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/resource.h>
25 #include <linux/latencytop.h>
26 #include <linux/sched/prio.h>
27 #include <linux/signal_types.h>
28 #include <linux/mm_types_task.h>
29 #include <linux/task_io_accounting.h>
31 /* task_struct member predeclarations (sorted alphabetically): */
33 struct backing_dev_info
;
38 struct futex_pi_state
;
43 struct perf_event_context
;
45 struct pipe_inode_info
;
48 struct robust_list_head
;
52 struct sighand_struct
;
54 struct task_delay_info
;
58 * Task state bitmask. NOTE! These bits are also
59 * encoded in fs/proc/array.c: get_task_state().
61 * We have two separate sets of flags: task->state
62 * is about runnability, while task->exit_state are
63 * about the task exiting. Confusing, but this way
64 * modifying one set can't modify the other one by
68 /* Used in tsk->state: */
69 #define TASK_RUNNING 0x0000
70 #define TASK_INTERRUPTIBLE 0x0001
71 #define TASK_UNINTERRUPTIBLE 0x0002
72 #define __TASK_STOPPED 0x0004
73 #define __TASK_TRACED 0x0008
74 /* Used in tsk->exit_state: */
75 #define EXIT_DEAD 0x0010
76 #define EXIT_ZOMBIE 0x0020
77 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
78 /* Used in tsk->state again: */
79 #define TASK_PARKED 0x0040
80 #define TASK_DEAD 0x0080
81 #define TASK_WAKEKILL 0x0100
82 #define TASK_WAKING 0x0200
83 #define TASK_NOLOAD 0x0400
84 #define TASK_NEW 0x0800
85 #define TASK_STATE_MAX 0x1000
87 /* Convenience macros for the sake of set_current_state: */
88 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
89 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
90 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
92 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
94 /* Convenience macros for the sake of wake_up(): */
95 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
96 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
98 /* get_task_state(): */
99 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
100 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
101 __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \
104 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
106 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
108 #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
110 #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
111 (task->flags & PF_FROZEN) == 0 && \
112 (task->state & TASK_NOLOAD) == 0)
114 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
117 * Special states are those that do not use the normal wait-loop pattern. See
118 * the comment with set_special_state().
120 #define is_special_task_state(state) \
121 ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_DEAD))
123 #define __set_current_state(state_value) \
125 WARN_ON_ONCE(is_special_task_state(state_value));\
126 current->task_state_change = _THIS_IP_; \
127 current->state = (state_value); \
130 #define set_current_state(state_value) \
132 WARN_ON_ONCE(is_special_task_state(state_value));\
133 current->task_state_change = _THIS_IP_; \
134 smp_store_mb(current->state, (state_value)); \
137 #define set_special_state(state_value) \
139 unsigned long flags; /* may shadow */ \
140 WARN_ON_ONCE(!is_special_task_state(state_value)); \
141 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
142 current->task_state_change = _THIS_IP_; \
143 current->state = (state_value); \
144 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
148 * set_current_state() includes a barrier so that the write of current->state
149 * is correctly serialised wrt the caller's subsequent test of whether to
153 * set_current_state(TASK_UNINTERRUPTIBLE);
159 * __set_current_state(TASK_RUNNING);
161 * If the caller does not need such serialisation (because, for instance, the
162 * condition test and condition change and wakeup are under the same lock) then
163 * use __set_current_state().
165 * The above is typically ordered against the wakeup, which does:
167 * need_sleep = false;
168 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
170 * Where wake_up_state() (and all other wakeup primitives) imply enough
171 * barriers to order the store of the variable against wakeup.
173 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
174 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
175 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
177 * However, with slightly different timing the wakeup TASK_RUNNING store can
178 * also collide with the TASK_UNINTERRUPTIBLE store. Loosing that store is not
179 * a problem either because that will result in one extra go around the loop
180 * and our @cond test will save the day.
182 * Also see the comments of try_to_wake_up().
184 #define __set_current_state(state_value) \
185 current->state = (state_value)
187 #define set_current_state(state_value) \
188 smp_store_mb(current->state, (state_value))
191 * set_special_state() should be used for those states when the blocking task
192 * can not use the regular condition based wait-loop. In that case we must
193 * serialize against wakeups such that any possible in-flight TASK_RUNNING stores
194 * will not collide with our state change.
196 #define set_special_state(state_value) \
198 unsigned long flags; /* may shadow */ \
199 raw_spin_lock_irqsave(¤t->pi_lock, flags); \
200 current->state = (state_value); \
201 raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \
206 /* Task command name length: */
207 #define TASK_COMM_LEN 16
218 extern cpumask_var_t cpu_isolated_map
;
220 extern void scheduler_tick(void);
222 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
224 extern long schedule_timeout(long timeout
);
225 extern long schedule_timeout_interruptible(long timeout
);
226 extern long schedule_timeout_killable(long timeout
);
227 extern long schedule_timeout_uninterruptible(long timeout
);
228 extern long schedule_timeout_idle(long timeout
);
229 asmlinkage
void schedule(void);
230 extern void schedule_preempt_disabled(void);
232 extern int __must_check
io_schedule_prepare(void);
233 extern void io_schedule_finish(int token
);
234 extern long io_schedule_timeout(long timeout
);
235 extern void io_schedule(void);
238 * struct prev_cputime - snapshot of system and user cputime
239 * @utime: time spent in user mode
240 * @stime: time spent in system mode
241 * @lock: protects the above two fields
243 * Stores previous user/system time values such that we can guarantee
246 struct prev_cputime
{
247 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
255 * struct task_cputime - collected CPU time counts
256 * @utime: time spent in user mode, in nanoseconds
257 * @stime: time spent in kernel mode, in nanoseconds
258 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
260 * This structure groups together three kinds of CPU time that are tracked for
261 * threads and thread groups. Most things considering CPU time want to group
262 * these counts together and treat all three of them in parallel.
264 struct task_cputime
{
267 unsigned long long sum_exec_runtime
;
270 /* Alternate field names when used on cache expirations: */
271 #define virt_exp utime
272 #define prof_exp stime
273 #define sched_exp sum_exec_runtime
276 /* Task is sleeping or running in a CPU with VTIME inactive: */
278 /* Task runs in userspace in a CPU with VTIME active: */
280 /* Task runs in kernelspace in a CPU with VTIME active: */
286 unsigned long long starttime
;
287 enum vtime_state state
;
294 #ifdef CONFIG_SCHED_INFO
295 /* Cumulative counters: */
297 /* # of times we have run on this CPU: */
298 unsigned long pcount
;
300 /* Time spent waiting on a runqueue: */
301 unsigned long long run_delay
;
305 /* When did we last run on a CPU? */
306 unsigned long long last_arrival
;
308 /* When were we last queued to run? */
309 unsigned long long last_queued
;
311 #endif /* CONFIG_SCHED_INFO */
315 * Integer metrics need fixed point arithmetic, e.g., sched/fair
316 * has a few: load, load_avg, util_avg, freq, and capacity.
318 * We define a basic fixed point arithmetic range, and then formalize
319 * all these metrics based on that basic range.
321 # define SCHED_FIXEDPOINT_SHIFT 10
322 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
325 unsigned long weight
;
330 * struct util_est - Estimation utilization of FAIR tasks
331 * @enqueued: instantaneous estimated utilization of a task/cpu
332 * @ewma: the Exponential Weighted Moving Average (EWMA)
333 * utilization of a task
335 * Support data structure to track an Exponential Weighted Moving Average
336 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
337 * average each time a task completes an activation. Sample's weight is chosen
338 * so that the EWMA will be relatively insensitive to transient changes to the
341 * The enqueued attribute has a slightly different meaning for tasks and cpus:
342 * - task: the task's util_avg at last task dequeue time
343 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
344 * Thus, the util_est.enqueued of a task represents the contribution on the
345 * estimated utilization of the CPU where that task is currently enqueued.
347 * Only for tasks we track a moving average of the past instantaneous
348 * estimated utilization. This allows to absorb sporadic drops in utilization
349 * of an otherwise almost periodic task.
352 unsigned int enqueued
;
354 #define UTIL_EST_WEIGHT_SHIFT 2
358 * The load_avg/util_avg accumulates an infinite geometric series
359 * (see __update_load_avg() in kernel/sched/fair.c).
361 * [load_avg definition]
363 * load_avg = runnable% * scale_load_down(load)
365 * where runnable% is the time ratio that a sched_entity is runnable.
366 * For cfs_rq, it is the aggregated load_avg of all runnable and
367 * blocked sched_entities.
369 * load_avg may also take frequency scaling into account:
371 * load_avg = runnable% * scale_load_down(load) * freq%
373 * where freq% is the CPU frequency normalized to the highest frequency.
375 * [util_avg definition]
377 * util_avg = running% * SCHED_CAPACITY_SCALE
379 * where running% is the time ratio that a sched_entity is running on
380 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
381 * and blocked sched_entities.
383 * util_avg may also factor frequency scaling and CPU capacity scaling:
385 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
387 * where freq% is the same as above, and capacity% is the CPU capacity
388 * normalized to the greatest capacity (due to uarch differences, etc).
390 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
391 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
392 * we therefore scale them to as large a range as necessary. This is for
393 * example reflected by 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
;
414 unsigned long load_avg
;
415 unsigned long util_avg
;
416 struct util_est util_est
;
419 struct sched_statistics
{
420 #ifdef CONFIG_SCHEDSTATS
430 s64 sum_sleep_runtime
;
437 u64 nr_migrations_cold
;
438 u64 nr_failed_migrations_affine
;
439 u64 nr_failed_migrations_running
;
440 u64 nr_failed_migrations_hot
;
441 u64 nr_forced_migrations
;
445 u64 nr_wakeups_migrate
;
446 u64 nr_wakeups_local
;
447 u64 nr_wakeups_remote
;
448 u64 nr_wakeups_affine
;
449 u64 nr_wakeups_affine_attempts
;
450 u64 nr_wakeups_passive
;
455 struct sched_entity
{
456 /* For load-balancing: */
457 struct load_weight load
;
458 struct rb_node run_node
;
459 struct list_head group_node
;
463 u64 sum_exec_runtime
;
465 u64 prev_sum_exec_runtime
;
469 struct sched_statistics statistics
;
471 #ifdef CONFIG_FAIR_GROUP_SCHED
473 struct sched_entity
*parent
;
474 /* rq on which this entity is (to be) queued: */
475 struct cfs_rq
*cfs_rq
;
476 /* rq "owned" by this entity/group: */
482 * Per entity load average tracking.
484 * Put into separate cache line so it does not
485 * collide with read-mostly values above.
487 struct sched_avg avg ____cacheline_aligned_in_smp
;
491 #ifdef CONFIG_SCHED_WALT
492 #define RAVG_HIST_SIZE_MAX 5
494 /* ravg represents frequency scaled cpu-demand of tasks */
497 * 'mark_start' marks the beginning of an event (task waking up, task
498 * starting to execute, task being preempted) within a window
500 * 'sum' represents how runnable a task has been within current
501 * window. It incorporates both running time and wait time and is
504 * 'sum_history' keeps track of history of 'sum' seen over previous
505 * RAVG_HIST_SIZE windows. Windows where task was entirely sleeping are
508 * 'demand' represents maximum sum seen over previous
509 * sysctl_sched_ravg_hist_size windows. 'demand' could drive frequency
512 * 'curr_window' represents task's contribution to cpu busy time
513 * statistics (rq->curr_runnable_sum) in current window
515 * 'prev_window' represents task's contribution to cpu busy time
516 * statistics (rq->prev_runnable_sum) in previous window
520 u32 sum_history
[RAVG_HIST_SIZE_MAX
];
521 u32 curr_window
, prev_window
;
526 struct sched_rt_entity
{
527 struct list_head run_list
;
528 unsigned long timeout
;
529 unsigned long watchdog_stamp
;
530 unsigned int time_slice
;
531 unsigned short on_rq
;
532 unsigned short on_list
;
534 struct sched_rt_entity
*back
;
535 #ifdef CONFIG_RT_GROUP_SCHED
536 struct sched_rt_entity
*parent
;
537 /* rq on which this entity is (to be) queued: */
539 /* rq "owned" by this entity/group: */
542 } __randomize_layout
;
544 struct sched_dl_entity
{
545 struct rb_node rb_node
;
548 * Original scheduling parameters. Copied here from sched_attr
549 * during sched_setattr(), they will remain the same until
550 * the next sched_setattr().
552 u64 dl_runtime
; /* Maximum runtime for each instance */
553 u64 dl_deadline
; /* Relative deadline of each instance */
554 u64 dl_period
; /* Separation of two instances (period) */
555 u64 dl_bw
; /* dl_runtime / dl_period */
556 u64 dl_density
; /* dl_runtime / dl_deadline */
559 * Actual scheduling parameters. Initialized with the values above,
560 * they are continously updated during task execution. Note that
561 * the remaining runtime could be < 0 in case we are in overrun.
563 s64 runtime
; /* Remaining runtime for this instance */
564 u64 deadline
; /* Absolute deadline for this instance */
565 unsigned int flags
; /* Specifying the scheduler behaviour */
570 * @dl_throttled tells if we exhausted the runtime. If so, the
571 * task has to wait for a replenishment to be performed at the
572 * next firing of dl_timer.
574 * @dl_boosted tells if we are boosted due to DI. If so we are
575 * outside bandwidth enforcement mechanism (but only until we
576 * exit the critical section);
578 * @dl_yielded tells if task gave up the CPU before consuming
579 * all its available runtime during the last job.
581 * @dl_non_contending tells if the task is inactive while still
582 * contributing to the active utilization. In other words, it
583 * indicates if the inactive timer has been armed and its handler
584 * has not been executed yet. This flag is useful to avoid race
585 * conditions between the inactive timer handler and the wakeup
591 int dl_non_contending
;
594 * Bandwidth enforcement timer. Each -deadline task has its
595 * own bandwidth to be enforced, thus we need one timer per task.
597 struct hrtimer dl_timer
;
600 * Inactive timer, responsible for decreasing the active utilization
601 * at the "0-lag time". When a -deadline task blocks, it contributes
602 * to GRUB's active utilization until the "0-lag time", hence a
603 * timer is needed to decrease the active utilization at the correct
606 struct hrtimer inactive_timer
;
615 /* Otherwise the compiler can store garbage here: */
618 u32 s
; /* Set of bits. */
621 enum perf_event_task_context
{
622 perf_invalid_context
= -1,
625 perf_nr_task_contexts
,
629 struct wake_q_node
*next
;
633 #ifdef CONFIG_THREAD_INFO_IN_TASK
635 * For reasons of header soup (see current_thread_info()), this
636 * must be the first element of task_struct.
638 struct thread_info thread_info
;
640 /* -1 unrunnable, 0 runnable, >0 stopped: */
644 * This begins the randomizable portion of task_struct. Only
645 * scheduling-critical items should be added above here.
647 randomized_struct_fields_start
651 /* Per task flags (PF_*), defined further below: */
656 struct llist_node wake_entry
;
658 #ifdef CONFIG_THREAD_INFO_IN_TASK
662 unsigned int wakee_flips
;
663 unsigned long wakee_flip_decay_ts
;
664 struct task_struct
*last_wakee
;
673 unsigned int rt_priority
;
675 const struct sched_class
*sched_class
;
676 struct sched_entity se
;
677 struct sched_rt_entity rt
;
678 #ifdef CONFIG_SCHED_WALT
681 * 'init_load_pct' represents the initial task load assigned to children
688 #ifdef CONFIG_CGROUP_SCHED
689 struct task_group
*sched_task_group
;
691 struct sched_dl_entity dl
;
693 #ifdef CONFIG_PREEMPT_NOTIFIERS
694 /* List of struct preempt_notifier: */
695 struct hlist_head preempt_notifiers
;
698 #ifdef CONFIG_BLK_DEV_IO_TRACE
699 unsigned int btrace_seq
;
704 cpumask_t cpus_allowed
;
706 #ifdef CONFIG_PREEMPT_RCU
707 int rcu_read_lock_nesting
;
708 union rcu_special rcu_read_unlock_special
;
709 struct list_head rcu_node_entry
;
710 struct rcu_node
*rcu_blocked_node
;
711 #endif /* #ifdef CONFIG_PREEMPT_RCU */
713 #ifdef CONFIG_TASKS_RCU
714 unsigned long rcu_tasks_nvcsw
;
715 u8 rcu_tasks_holdout
;
717 int rcu_tasks_idle_cpu
;
718 struct list_head rcu_tasks_holdout_list
;
719 #endif /* #ifdef CONFIG_TASKS_RCU */
721 struct sched_info sched_info
;
723 struct list_head tasks
;
725 struct plist_node pushable_tasks
;
726 struct rb_node pushable_dl_tasks
;
729 struct mm_struct
*mm
;
730 struct mm_struct
*active_mm
;
732 /* Per-thread vma caching: */
733 struct vmacache vmacache
;
735 #ifdef SPLIT_RSS_COUNTING
736 struct task_rss_stat rss_stat
;
741 /* The signal sent when the parent dies: */
743 /* JOBCTL_*, siglock protected: */
744 unsigned long jobctl
;
746 /* Used for emulating ABI behavior of previous Linux versions: */
747 unsigned int personality
;
749 /* Scheduler bits, serialized by scheduler locks: */
750 unsigned sched_reset_on_fork
:1;
751 unsigned sched_contributes_to_load
:1;
752 unsigned sched_migrated
:1;
753 unsigned sched_remote_wakeup
:1;
754 /* Force alignment to the next boundary: */
757 /* Unserialized, strictly 'current' */
759 /* Bit to tell LSMs we're in execve(): */
760 unsigned in_execve
:1;
761 unsigned in_iowait
:1;
762 #ifndef TIF_RESTORE_SIGMASK
763 unsigned restore_sigmask
:1;
766 unsigned memcg_may_oom
:1;
768 unsigned memcg_kmem_skip_account
:1;
771 #ifdef CONFIG_COMPAT_BRK
772 unsigned brk_randomized
:1;
774 #ifdef CONFIG_CGROUPS
775 /* disallow userland-initiated cgroup migration */
776 unsigned no_cgroup_migration
:1;
779 unsigned long atomic_flags
; /* Flags requiring atomic access. */
781 struct restart_block restart_block
;
786 #ifdef CONFIG_CC_STACKPROTECTOR
787 /* Canary value for the -fstack-protector GCC feature: */
788 unsigned long stack_canary
;
791 * Pointers to the (original) parent process, youngest child, younger sibling,
792 * older sibling, respectively. (p->father can be replaced with
793 * p->real_parent->pid)
796 /* Real parent process: */
797 struct task_struct __rcu
*real_parent
;
799 /* Recipient of SIGCHLD, wait4() reports: */
800 struct task_struct __rcu
*parent
;
803 * Children/sibling form the list of natural children:
805 struct list_head children
;
806 struct list_head sibling
;
807 struct task_struct
*group_leader
;
810 * 'ptraced' is the list of tasks this task is using ptrace() on.
812 * This includes both natural children and PTRACE_ATTACH targets.
813 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
815 struct list_head ptraced
;
816 struct list_head ptrace_entry
;
818 /* PID/PID hash table linkage. */
819 struct pid_link pids
[PIDTYPE_MAX
];
820 struct list_head thread_group
;
821 struct list_head thread_node
;
823 struct completion
*vfork_done
;
825 /* CLONE_CHILD_SETTID: */
826 int __user
*set_child_tid
;
828 /* CLONE_CHILD_CLEARTID: */
829 int __user
*clear_child_tid
;
833 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
838 #ifdef CONFIG_CPU_FREQ_TIMES
840 unsigned int max_state
;
842 struct prev_cputime prev_cputime
;
843 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
847 #ifdef CONFIG_NO_HZ_FULL
848 atomic_t tick_dep_mask
;
850 /* Context switch counts: */
852 unsigned long nivcsw
;
854 /* Monotonic time in nsecs: */
857 /* Boot based time in nsecs: */
860 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
861 unsigned long min_flt
;
862 unsigned long maj_flt
;
864 #ifdef CONFIG_POSIX_TIMERS
865 struct task_cputime cputime_expires
;
866 struct list_head cpu_timers
[3];
869 /* Process credentials: */
871 /* Tracer's credentials at attach: */
872 const struct cred __rcu
*ptracer_cred
;
874 /* Objective and real subjective task credentials (COW): */
875 const struct cred __rcu
*real_cred
;
877 /* Effective (overridable) subjective task credentials (COW): */
878 const struct cred __rcu
*cred
;
881 * executable name, excluding path.
883 * - normally initialized setup_new_exec()
884 * - access it with [gs]et_task_comm()
885 * - lock it with task_lock()
887 char comm
[TASK_COMM_LEN
];
889 struct nameidata
*nameidata
;
891 #ifdef CONFIG_SYSVIPC
892 struct sysv_sem sysvsem
;
893 struct sysv_shm sysvshm
;
895 #ifdef CONFIG_DETECT_HUNG_TASK
896 unsigned long last_switch_count
;
898 /* Filesystem information: */
899 struct fs_struct
*fs
;
901 /* Open file information: */
902 struct files_struct
*files
;
905 struct nsproxy
*nsproxy
;
907 /* Signal handlers: */
908 struct signal_struct
*signal
;
909 struct sighand_struct
*sighand
;
911 sigset_t real_blocked
;
912 /* Restored if set_restore_sigmask() was used: */
913 sigset_t saved_sigmask
;
914 struct sigpending pending
;
915 unsigned long sas_ss_sp
;
917 unsigned int sas_ss_flags
;
919 struct callback_head
*task_works
;
921 struct audit_context
*audit_context
;
922 #ifdef CONFIG_AUDITSYSCALL
924 unsigned int sessionid
;
926 struct seccomp seccomp
;
928 /* Thread group tracking: */
932 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
933 spinlock_t alloc_lock
;
935 /* Protection of the PI data structures: */
936 raw_spinlock_t pi_lock
;
938 struct wake_q_node wake_q
;
940 #ifdef CONFIG_RT_MUTEXES
941 /* PI waiters blocked on a rt_mutex held by this task: */
942 struct rb_root_cached pi_waiters
;
943 /* Updated under owner's pi_lock and rq lock */
944 struct task_struct
*pi_top_task
;
945 /* Deadlock detection and priority inheritance handling: */
946 struct rt_mutex_waiter
*pi_blocked_on
;
949 #ifdef CONFIG_DEBUG_MUTEXES
950 /* Mutex deadlock detection: */
951 struct mutex_waiter
*blocked_on
;
954 #ifdef CONFIG_TRACE_IRQFLAGS
955 unsigned int irq_events
;
956 unsigned long hardirq_enable_ip
;
957 unsigned long hardirq_disable_ip
;
958 unsigned int hardirq_enable_event
;
959 unsigned int hardirq_disable_event
;
960 int hardirqs_enabled
;
962 unsigned long softirq_disable_ip
;
963 unsigned long softirq_enable_ip
;
964 unsigned int softirq_disable_event
;
965 unsigned int softirq_enable_event
;
966 int softirqs_enabled
;
970 #ifdef CONFIG_LOCKDEP
971 # define MAX_LOCK_DEPTH 48UL
974 unsigned int lockdep_recursion
;
975 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
978 #ifdef CONFIG_LOCKDEP_CROSSRELEASE
979 #define MAX_XHLOCKS_NR 64UL
980 struct hist_lock
*xhlocks
; /* Crossrelease history locks */
981 unsigned int xhlock_idx
;
982 /* For restoring at history boundaries */
983 unsigned int xhlock_idx_hist
[XHLOCK_CTX_NR
];
984 unsigned int hist_id
;
985 /* For overwrite check at each context exit */
986 unsigned int hist_id_save
[XHLOCK_CTX_NR
];
990 unsigned int in_ubsan
;
993 /* Journalling filesystem info: */
996 /* Stacked block device info: */
997 struct bio_list
*bio_list
;
1000 /* Stack plugging: */
1001 struct blk_plug
*plug
;
1005 struct reclaim_state
*reclaim_state
;
1007 struct backing_dev_info
*backing_dev_info
;
1009 struct io_context
*io_context
;
1012 unsigned long ptrace_message
;
1013 siginfo_t
*last_siginfo
;
1015 struct task_io_accounting ioac
;
1016 #ifdef CONFIG_TASK_XACCT
1017 /* Accumulated RSS usage: */
1019 /* Accumulated virtual memory usage: */
1021 /* stime + utime since last update: */
1024 #ifdef CONFIG_CPUSETS
1025 /* Protected by ->alloc_lock: */
1026 nodemask_t mems_allowed
;
1027 /* Seqence number to catch updates: */
1028 seqcount_t mems_allowed_seq
;
1029 int cpuset_mem_spread_rotor
;
1030 int cpuset_slab_spread_rotor
;
1032 #ifdef CONFIG_CGROUPS
1033 /* Control Group info protected by css_set_lock: */
1034 struct css_set __rcu
*cgroups
;
1035 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1036 struct list_head cg_list
;
1038 #ifdef CONFIG_INTEL_RDT
1043 struct robust_list_head __user
*robust_list
;
1044 #ifdef CONFIG_COMPAT
1045 struct compat_robust_list_head __user
*compat_robust_list
;
1047 struct list_head pi_state_list
;
1048 struct futex_pi_state
*pi_state_cache
;
1050 #ifdef CONFIG_PERF_EVENTS
1051 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1052 struct mutex perf_event_mutex
;
1053 struct list_head perf_event_list
;
1055 #ifdef CONFIG_DEBUG_PREEMPT
1056 unsigned long preempt_disable_ip
;
1059 /* Protected by alloc_lock: */
1060 struct mempolicy
*mempolicy
;
1062 short pref_node_fork
;
1064 #ifdef CONFIG_NUMA_BALANCING
1066 unsigned int numa_scan_period
;
1067 unsigned int numa_scan_period_max
;
1068 int numa_preferred_nid
;
1069 unsigned long numa_migrate_retry
;
1070 /* Migration stamp: */
1072 u64 last_task_numa_placement
;
1073 u64 last_sum_exec_runtime
;
1074 struct callback_head numa_work
;
1076 struct list_head numa_entry
;
1077 struct numa_group
*numa_group
;
1080 * numa_faults is an array split into four regions:
1081 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1082 * in this precise order.
1084 * faults_memory: Exponential decaying average of faults on a per-node
1085 * basis. Scheduling placement decisions are made based on these
1086 * counts. The values remain static for the duration of a PTE scan.
1087 * faults_cpu: Track the nodes the process was running on when a NUMA
1088 * hinting fault was incurred.
1089 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1090 * during the current scan window. When the scan completes, the counts
1091 * in faults_memory and faults_cpu decay and these values are copied.
1093 unsigned long *numa_faults
;
1094 unsigned long total_numa_faults
;
1097 * numa_faults_locality tracks if faults recorded during the last
1098 * scan window were remote/local or failed to migrate. The task scan
1099 * period is adapted based on the locality of the faults with different
1100 * weights depending on whether they were shared or private faults
1102 unsigned long numa_faults_locality
[3];
1104 unsigned long numa_pages_migrated
;
1105 #endif /* CONFIG_NUMA_BALANCING */
1107 struct tlbflush_unmap_batch tlb_ubc
;
1109 struct rcu_head rcu
;
1111 /* Cache last used pipe for splice(): */
1112 struct pipe_inode_info
*splice_pipe
;
1114 struct page_frag task_frag
;
1116 #ifdef CONFIG_TASK_DELAY_ACCT
1117 struct task_delay_info
*delays
;
1120 #ifdef CONFIG_FAULT_INJECTION
1122 unsigned int fail_nth
;
1125 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1126 * balance_dirty_pages() for a dirty throttling pause:
1129 int nr_dirtied_pause
;
1130 /* Start of a write-and-pause period: */
1131 unsigned long dirty_paused_when
;
1133 #ifdef CONFIG_LATENCYTOP
1134 int latency_record_count
;
1135 struct latency_record latency_record
[LT_SAVECOUNT
];
1138 * Time slack values; these are used to round up poll() and
1139 * select() etc timeout values. These are in nanoseconds.
1142 u64 default_timer_slack_ns
;
1145 unsigned int kasan_depth
;
1148 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1149 /* Index of current stored address in ret_stack: */
1152 /* Stack of return addresses for return function tracing: */
1153 struct ftrace_ret_stack
*ret_stack
;
1155 /* Timestamp for last schedule: */
1156 unsigned long long ftrace_timestamp
;
1159 * Number of functions that haven't been traced
1160 * because of depth overrun:
1162 atomic_t trace_overrun
;
1164 /* Pause tracing: */
1165 atomic_t tracing_graph_pause
;
1168 #ifdef CONFIG_TRACING
1169 /* State flags for use by tracers: */
1170 unsigned long trace
;
1172 /* Bitmask and counter of trace recursion: */
1173 unsigned long trace_recursion
;
1174 #endif /* CONFIG_TRACING */
1177 /* Coverage collection mode enabled for this task (0 if disabled): */
1178 enum kcov_mode kcov_mode
;
1180 /* Size of the kcov_area: */
1181 unsigned int kcov_size
;
1183 /* Buffer for coverage collection: */
1186 /* KCOV descriptor wired with this task or NULL: */
1191 struct mem_cgroup
*memcg_in_oom
;
1192 gfp_t memcg_oom_gfp_mask
;
1193 int memcg_oom_order
;
1195 /* Number of pages to reclaim on returning to userland: */
1196 unsigned int memcg_nr_pages_over_high
;
1199 #ifdef CONFIG_UPROBES
1200 struct uprobe_task
*utask
;
1202 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1203 unsigned int sequential_io
;
1204 unsigned int sequential_io_avg
;
1206 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1207 unsigned long task_state_change
;
1209 int pagefault_disabled
;
1211 struct task_struct
*oom_reaper_list
;
1213 #ifdef CONFIG_VMAP_STACK
1214 struct vm_struct
*stack_vm_area
;
1216 #ifdef CONFIG_THREAD_INFO_IN_TASK
1217 /* A live task holds one reference: */
1218 atomic_t stack_refcount
;
1220 #ifdef CONFIG_LIVEPATCH
1223 #ifdef CONFIG_SECURITY
1224 /* Used by LSM modules for access restriction: */
1229 * New fields for task_struct should be added above here, so that
1230 * they are included in the randomized portion of task_struct.
1232 randomized_struct_fields_end
1234 /* CPU-specific state of this task: */
1235 struct thread_struct thread
;
1238 * WARNING: on x86, 'thread_struct' contains a variable-sized
1239 * structure. It *MUST* be at the end of 'task_struct'.
1241 * Do not put anything below here!
1245 static inline struct pid
*task_pid(struct task_struct
*task
)
1247 return task
->pids
[PIDTYPE_PID
].pid
;
1250 static inline struct pid
*task_tgid(struct task_struct
*task
)
1252 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1256 * Without tasklist or RCU lock it is not safe to dereference
1257 * the result of task_pgrp/task_session even if task == current,
1258 * we can race with another thread doing sys_setsid/sys_setpgid.
1260 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1262 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1265 static inline struct pid
*task_session(struct task_struct
*task
)
1267 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1271 * the helpers to get the task's different pids as they are seen
1272 * from various namespaces
1274 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1275 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1277 * task_xid_nr_ns() : id seen from the ns specified;
1279 * see also pid_nr() etc in include/linux/pid.h
1281 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
, struct pid_namespace
*ns
);
1283 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1288 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1290 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1293 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1295 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1299 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1305 * pid_alive - check that a task structure is not stale
1306 * @p: Task structure to be checked.
1308 * Test if a process is not yet dead (at most zombie state)
1309 * If pid_alive fails, then pointers within the task structure
1310 * can be stale and must not be dereferenced.
1312 * Return: 1 if the process is alive. 0 otherwise.
1314 static inline int pid_alive(const struct task_struct
*p
)
1316 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1319 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1321 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1324 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1326 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1330 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1332 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1335 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1337 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1340 static inline pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1342 return __task_pid_nr_ns(tsk
, __PIDTYPE_TGID
, ns
);
1345 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1347 return __task_pid_nr_ns(tsk
, __PIDTYPE_TGID
, NULL
);
1350 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1356 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1362 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1364 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
1367 /* Obsolete, do not use: */
1368 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1370 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1373 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1374 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1376 static inline unsigned int __get_task_state(struct task_struct
*tsk
)
1378 unsigned int tsk_state
= READ_ONCE(tsk
->state
);
1379 unsigned int state
= (tsk_state
| tsk
->exit_state
) & TASK_REPORT
;
1381 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX
);
1383 if (tsk_state
== TASK_IDLE
)
1384 state
= TASK_REPORT_IDLE
;
1389 static inline char __task_state_to_char(unsigned int state
)
1391 static const char state_char
[] = "RSDTtXZPI";
1393 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX
) != sizeof(state_char
) - 1);
1395 return state_char
[state
];
1398 static inline char task_state_to_char(struct task_struct
*tsk
)
1400 return __task_state_to_char(__get_task_state(tsk
));
1404 * is_global_init - check if a task structure is init. Since init
1405 * is free to have sub-threads we need to check tgid.
1406 * @tsk: Task structure to be checked.
1408 * Check if a task structure is the first user space task the kernel created.
1410 * Return: 1 if the task structure is init. 0 otherwise.
1412 static inline int is_global_init(struct task_struct
*tsk
)
1414 return task_tgid_nr(tsk
) == 1;
1417 extern struct pid
*cad_pid
;
1422 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1423 #define PF_EXITING 0x00000004 /* Getting shut down */
1424 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1425 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1426 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1427 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1428 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1429 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1430 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1431 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1432 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1433 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1434 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1435 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1436 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1437 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1438 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1439 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1440 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1441 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1442 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1443 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1444 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1445 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1446 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1447 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1448 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1449 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1452 * Only the _current_ task can read/write to tsk->flags, but other
1453 * tasks can access tsk->flags in readonly mode for example
1454 * with tsk_used_math (like during threaded core dumping).
1455 * There is however an exception to this rule during ptrace
1456 * or during fork: the ptracer task is allowed to write to the
1457 * child->flags of its traced child (same goes for fork, the parent
1458 * can write to the child->flags), because we're guaranteed the
1459 * child is not running and in turn not changing child->flags
1460 * at the same time the parent does it.
1462 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1463 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1464 #define clear_used_math() clear_stopped_child_used_math(current)
1465 #define set_used_math() set_stopped_child_used_math(current)
1467 #define conditional_stopped_child_used_math(condition, child) \
1468 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1470 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1472 #define copy_to_stopped_child_used_math(child) \
1473 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1475 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1476 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1477 #define used_math() tsk_used_math(current)
1479 static inline bool is_percpu_thread(void)
1482 return (current
->flags
& PF_NO_SETAFFINITY
) &&
1483 (current
->nr_cpus_allowed
== 1);
1489 /* Per-process atomic flags. */
1490 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1491 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1492 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1493 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1494 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1496 #define TASK_PFA_TEST(name, func) \
1497 static inline bool task_##func(struct task_struct *p) \
1498 { return test_bit(PFA_##name, &p->atomic_flags); }
1500 #define TASK_PFA_SET(name, func) \
1501 static inline void task_set_##func(struct task_struct *p) \
1502 { set_bit(PFA_##name, &p->atomic_flags); }
1504 #define TASK_PFA_CLEAR(name, func) \
1505 static inline void task_clear_##func(struct task_struct *p) \
1506 { clear_bit(PFA_##name, &p->atomic_flags); }
1508 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
1509 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
1511 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
1512 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
1513 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
1515 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
1516 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
1517 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
1519 TASK_PFA_TEST(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1520 TASK_PFA_SET(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1521 TASK_PFA_CLEAR(SPEC_SSB_DISABLE
, spec_ssb_disable
)
1523 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE
, spec_ssb_force_disable
)
1524 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE
, spec_ssb_force_disable
)
1527 current_restore_flags(unsigned long orig_flags
, unsigned long flags
)
1529 current
->flags
&= ~flags
;
1530 current
->flags
|= orig_flags
& flags
;
1533 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
, const struct cpumask
*trial
);
1534 extern int task_can_attach(struct task_struct
*p
, const struct cpumask
*cs_cpus_allowed
);
1536 extern void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
);
1537 extern int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
);
1539 static inline void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
)
1542 static inline int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
)
1544 if (!cpumask_test_cpu(0, new_mask
))
1550 #ifndef cpu_relax_yield
1551 #define cpu_relax_yield() cpu_relax()
1554 extern int yield_to(struct task_struct
*p
, bool preempt
);
1555 extern void set_user_nice(struct task_struct
*p
, long nice
);
1556 extern int task_prio(const struct task_struct
*p
);
1559 * task_nice - return the nice value of a given task.
1560 * @p: the task in question.
1562 * Return: The nice value [ -20 ... 0 ... 19 ].
1564 static inline int task_nice(const struct task_struct
*p
)
1566 return PRIO_TO_NICE((p
)->static_prio
);
1569 extern int can_nice(const struct task_struct
*p
, const int nice
);
1570 extern int task_curr(const struct task_struct
*p
);
1571 extern int idle_cpu(int cpu
);
1572 extern int sched_setscheduler(struct task_struct
*, int, const struct sched_param
*);
1573 extern int sched_setscheduler_nocheck(struct task_struct
*, int, const struct sched_param
*);
1574 extern int sched_setattr(struct task_struct
*, const struct sched_attr
*);
1575 extern struct task_struct
*idle_task(int cpu
);
1578 * is_idle_task - is the specified task an idle task?
1579 * @p: the task in question.
1581 * Return: 1 if @p is an idle task. 0 otherwise.
1583 static inline bool is_idle_task(const struct task_struct
*p
)
1585 return !!(p
->flags
& PF_IDLE
);
1588 extern struct task_struct
*curr_task(int cpu
);
1589 extern void ia64_set_curr_task(int cpu
, struct task_struct
*p
);
1593 union thread_union
{
1594 #ifndef CONFIG_THREAD_INFO_IN_TASK
1595 struct thread_info thread_info
;
1597 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1600 #ifdef CONFIG_THREAD_INFO_IN_TASK
1601 static inline struct thread_info
*task_thread_info(struct task_struct
*task
)
1603 return &task
->thread_info
;
1605 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1606 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1610 * find a task by one of its numerical ids
1612 * find_task_by_pid_ns():
1613 * finds a task by its pid in the specified namespace
1614 * find_task_by_vpid():
1615 * finds a task by its virtual pid
1617 * see also find_vpid() etc in include/linux/pid.h
1620 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1621 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
, struct pid_namespace
*ns
);
1623 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1624 extern int wake_up_process(struct task_struct
*tsk
);
1625 extern void wake_up_new_task(struct task_struct
*tsk
);
1628 extern void kick_process(struct task_struct
*tsk
);
1630 static inline void kick_process(struct task_struct
*tsk
) { }
1633 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
1635 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
1637 __set_task_comm(tsk
, from
, false);
1640 extern char *__get_task_comm(char *to
, size_t len
, struct task_struct
*tsk
);
1641 #define get_task_comm(buf, tsk) ({ \
1642 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1643 __get_task_comm(buf, sizeof(buf), tsk); \
1647 void scheduler_ipi(void);
1648 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
1650 static inline void scheduler_ipi(void) { }
1651 static inline unsigned long wait_task_inactive(struct task_struct
*p
, long match_state
)
1658 * Set thread flags in other task's structures.
1659 * See asm/thread_info.h for TIF_xxxx flags available:
1661 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1663 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1666 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1668 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1671 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1673 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1676 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1678 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1681 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1683 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1686 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1688 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1691 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1693 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1696 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
1698 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
1702 * cond_resched() and cond_resched_lock(): latency reduction via
1703 * explicit rescheduling in places that are safe. The return
1704 * value indicates whether a reschedule was done in fact.
1705 * cond_resched_lock() will drop the spinlock before scheduling,
1706 * cond_resched_softirq() will enable bhs before scheduling.
1708 #ifndef CONFIG_PREEMPT
1709 extern int _cond_resched(void);
1711 static inline int _cond_resched(void) { return 0; }
1714 #define cond_resched() ({ \
1715 ___might_sleep(__FILE__, __LINE__, 0); \
1719 extern int __cond_resched_lock(spinlock_t
*lock
);
1721 #define cond_resched_lock(lock) ({ \
1722 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1723 __cond_resched_lock(lock); \
1726 extern int __cond_resched_softirq(void);
1728 #define cond_resched_softirq() ({ \
1729 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
1730 __cond_resched_softirq(); \
1733 static inline void cond_resched_rcu(void)
1735 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1743 * Does a critical section need to be broken due to another
1744 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1745 * but a general need for low latency)
1747 static inline int spin_needbreak(spinlock_t
*lock
)
1749 #ifdef CONFIG_PREEMPT
1750 return spin_is_contended(lock
);
1756 static __always_inline
bool need_resched(void)
1758 return unlikely(tif_need_resched());
1762 * Wrappers for p->thread_info->cpu access. No-op on UP.
1766 static inline unsigned int task_cpu(const struct task_struct
*p
)
1768 #ifdef CONFIG_THREAD_INFO_IN_TASK
1771 return task_thread_info(p
)->cpu
;
1775 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1779 static inline unsigned int task_cpu(const struct task_struct
*p
)
1784 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1788 #endif /* CONFIG_SMP */
1791 * In order to reduce various lock holder preemption latencies provide an
1792 * interface to see if a vCPU is currently running or not.
1794 * This allows us to terminate optimistic spin loops and block, analogous to
1795 * the native optimistic spin heuristic of testing if the lock owner task is
1798 #ifndef vcpu_is_preempted
1799 # define vcpu_is_preempted(cpu) false
1802 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
1803 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
1805 #ifndef TASK_SIZE_OF
1806 #define TASK_SIZE_OF(tsk) TASK_SIZE