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
;
420 u64 ontime_migration_time
;
423 unsigned long load_avg
;
426 struct ontime_entity
{
427 struct ontime_avg avg
;
432 struct sched_statistics
{
433 #ifdef CONFIG_SCHEDSTATS
443 s64 sum_sleep_runtime
;
450 u64 nr_migrations_cold
;
451 u64 nr_failed_migrations_affine
;
452 u64 nr_failed_migrations_running
;
453 u64 nr_failed_migrations_hot
;
454 u64 nr_forced_migrations
;
458 u64 nr_wakeups_migrate
;
459 u64 nr_wakeups_local
;
460 u64 nr_wakeups_remote
;
461 u64 nr_wakeups_affine
;
462 u64 nr_wakeups_affine_attempts
;
463 u64 nr_wakeups_passive
;
468 struct sched_entity
{
469 /* For load-balancing: */
470 struct load_weight load
;
471 struct rb_node run_node
;
472 struct list_head group_node
;
476 u64 sum_exec_runtime
;
478 u64 prev_sum_exec_runtime
;
482 struct sched_statistics statistics
;
484 #ifdef CONFIG_FAIR_GROUP_SCHED
486 struct sched_entity
*parent
;
487 /* rq on which this entity is (to be) queued: */
488 struct cfs_rq
*cfs_rq
;
489 /* rq "owned" by this entity/group: */
495 * Per entity load average tracking.
497 * Put into separate cache line so it does not
498 * collide with read-mostly values above.
500 struct sched_avg avg ____cacheline_aligned_in_smp
;
502 struct ontime_entity ontime
;
505 #ifdef CONFIG_SCHED_WALT
506 #define RAVG_HIST_SIZE_MAX 5
508 /* ravg represents frequency scaled cpu-demand of tasks */
511 * 'mark_start' marks the beginning of an event (task waking up, task
512 * starting to execute, task being preempted) within a window
514 * 'sum' represents how runnable a task has been within current
515 * window. It incorporates both running time and wait time and is
518 * 'sum_history' keeps track of history of 'sum' seen over previous
519 * RAVG_HIST_SIZE windows. Windows where task was entirely sleeping are
522 * 'demand' represents maximum sum seen over previous
523 * sysctl_sched_ravg_hist_size windows. 'demand' could drive frequency
526 * 'curr_window' represents task's contribution to cpu busy time
527 * statistics (rq->curr_runnable_sum) in current window
529 * 'prev_window' represents task's contribution to cpu busy time
530 * statistics (rq->prev_runnable_sum) in previous window
534 u32 sum_history
[RAVG_HIST_SIZE_MAX
];
535 u32 curr_window
, prev_window
;
540 struct sched_rt_entity
{
541 struct list_head run_list
;
542 unsigned long timeout
;
543 unsigned long watchdog_stamp
;
544 unsigned int time_slice
;
545 unsigned short on_rq
;
546 unsigned short on_list
;
548 struct sched_rt_entity
*back
;
549 #ifdef CONFIG_RT_GROUP_SCHED
550 struct sched_rt_entity
*parent
;
551 /* rq on which this entity is (to be) queued: */
553 /* rq "owned" by this entity/group: */
558 #ifdef CONFIG_SCHED_USE_FLUID_RT
562 * Per entity load average tracking.
564 * Put into separate cache line so it does not
565 * collide with read-mostly values above.
567 struct sched_avg avg
;// ____cacheline_aligned_in_smp;
569 } __randomize_layout
;
571 struct sched_dl_entity
{
572 struct rb_node rb_node
;
575 * Original scheduling parameters. Copied here from sched_attr
576 * during sched_setattr(), they will remain the same until
577 * the next sched_setattr().
579 u64 dl_runtime
; /* Maximum runtime for each instance */
580 u64 dl_deadline
; /* Relative deadline of each instance */
581 u64 dl_period
; /* Separation of two instances (period) */
582 u64 dl_bw
; /* dl_runtime / dl_period */
583 u64 dl_density
; /* dl_runtime / dl_deadline */
586 * Actual scheduling parameters. Initialized with the values above,
587 * they are continously updated during task execution. Note that
588 * the remaining runtime could be < 0 in case we are in overrun.
590 s64 runtime
; /* Remaining runtime for this instance */
591 u64 deadline
; /* Absolute deadline for this instance */
592 unsigned int flags
; /* Specifying the scheduler behaviour */
597 * @dl_throttled tells if we exhausted the runtime. If so, the
598 * task has to wait for a replenishment to be performed at the
599 * next firing of dl_timer.
601 * @dl_boosted tells if we are boosted due to DI. If so we are
602 * outside bandwidth enforcement mechanism (but only until we
603 * exit the critical section);
605 * @dl_yielded tells if task gave up the CPU before consuming
606 * all its available runtime during the last job.
608 * @dl_non_contending tells if the task is inactive while still
609 * contributing to the active utilization. In other words, it
610 * indicates if the inactive timer has been armed and its handler
611 * has not been executed yet. This flag is useful to avoid race
612 * conditions between the inactive timer handler and the wakeup
618 int dl_non_contending
;
621 * Bandwidth enforcement timer. Each -deadline task has its
622 * own bandwidth to be enforced, thus we need one timer per task.
624 struct hrtimer dl_timer
;
627 * Inactive timer, responsible for decreasing the active utilization
628 * at the "0-lag time". When a -deadline task blocks, it contributes
629 * to GRUB's active utilization until the "0-lag time", hence a
630 * timer is needed to decrease the active utilization at the correct
633 struct hrtimer inactive_timer
;
642 /* Otherwise the compiler can store garbage here: */
645 u32 s
; /* Set of bits. */
648 enum perf_event_task_context
{
649 perf_invalid_context
= -1,
652 perf_nr_task_contexts
,
656 struct wake_q_node
*next
;
660 #ifdef CONFIG_THREAD_INFO_IN_TASK
662 * For reasons of header soup (see current_thread_info()), this
663 * must be the first element of task_struct.
665 struct thread_info thread_info
;
667 /* -1 unrunnable, 0 runnable, >0 stopped: */
671 * This begins the randomizable portion of task_struct. Only
672 * scheduling-critical items should be added above here.
674 randomized_struct_fields_start
678 /* Per task flags (PF_*), defined further below: */
683 struct llist_node wake_entry
;
685 #ifdef CONFIG_THREAD_INFO_IN_TASK
689 unsigned int wakee_flips
;
690 unsigned long wakee_flip_decay_ts
;
691 struct task_struct
*last_wakee
;
700 unsigned int rt_priority
;
702 const struct sched_class
*sched_class
;
703 struct sched_entity se
;
704 struct sched_rt_entity rt
;
705 #ifdef CONFIG_SCHED_WALT
708 * 'init_load_pct' represents the initial task load assigned to children
714 #ifdef CONFIG_SCHED_USE_FLUID_RT
718 #ifdef CONFIG_SCHED_EMS
719 struct task_band
*band
;
720 struct list_head band_members
;
723 #ifdef CONFIG_CGROUP_SCHED
724 struct task_group
*sched_task_group
;
726 struct sched_dl_entity dl
;
728 #ifdef CONFIG_PREEMPT_NOTIFIERS
729 /* List of struct preempt_notifier: */
730 struct hlist_head preempt_notifiers
;
733 #ifdef CONFIG_BLK_DEV_IO_TRACE
734 unsigned int btrace_seq
;
739 cpumask_t cpus_allowed
;
741 #ifdef CONFIG_PREEMPT_RCU
742 int rcu_read_lock_nesting
;
743 union rcu_special rcu_read_unlock_special
;
744 struct list_head rcu_node_entry
;
745 struct rcu_node
*rcu_blocked_node
;
746 #endif /* #ifdef CONFIG_PREEMPT_RCU */
748 #ifdef CONFIG_TASKS_RCU
749 unsigned long rcu_tasks_nvcsw
;
750 u8 rcu_tasks_holdout
;
752 int rcu_tasks_idle_cpu
;
753 struct list_head rcu_tasks_holdout_list
;
754 #endif /* #ifdef CONFIG_TASKS_RCU */
756 struct sched_info sched_info
;
758 struct list_head tasks
;
760 struct plist_node pushable_tasks
;
761 struct rb_node pushable_dl_tasks
;
764 struct mm_struct
*mm
;
765 struct mm_struct
*active_mm
;
767 /* Per-thread vma caching: */
768 struct vmacache vmacache
;
770 #ifdef SPLIT_RSS_COUNTING
771 struct task_rss_stat rss_stat
;
776 /* The signal sent when the parent dies: */
778 /* JOBCTL_*, siglock protected: */
779 unsigned long jobctl
;
781 /* Used for emulating ABI behavior of previous Linux versions: */
782 unsigned int personality
;
784 /* Scheduler bits, serialized by scheduler locks: */
785 unsigned sched_reset_on_fork
:1;
786 unsigned sched_contributes_to_load
:1;
787 unsigned sched_migrated
:1;
788 unsigned sched_remote_wakeup
:1;
789 /* Force alignment to the next boundary: */
792 /* Unserialized, strictly 'current' */
794 /* Bit to tell LSMs we're in execve(): */
795 unsigned in_execve
:1;
796 unsigned in_iowait
:1;
797 #ifndef TIF_RESTORE_SIGMASK
798 unsigned restore_sigmask
:1;
801 unsigned memcg_may_oom
:1;
803 unsigned memcg_kmem_skip_account
:1;
806 #ifdef CONFIG_COMPAT_BRK
807 unsigned brk_randomized
:1;
809 #ifdef CONFIG_CGROUPS
810 /* disallow userland-initiated cgroup migration */
811 unsigned no_cgroup_migration
:1;
814 unsigned long atomic_flags
; /* Flags requiring atomic access. */
816 struct restart_block restart_block
;
821 #ifdef CONFIG_CC_STACKPROTECTOR
822 /* Canary value for the -fstack-protector GCC feature: */
823 unsigned long stack_canary
;
826 * Pointers to the (original) parent process, youngest child, younger sibling,
827 * older sibling, respectively. (p->father can be replaced with
828 * p->real_parent->pid)
831 /* Real parent process: */
832 struct task_struct __rcu
*real_parent
;
834 /* Recipient of SIGCHLD, wait4() reports: */
835 struct task_struct __rcu
*parent
;
838 * Children/sibling form the list of natural children:
840 struct list_head children
;
841 struct list_head sibling
;
842 struct task_struct
*group_leader
;
845 * 'ptraced' is the list of tasks this task is using ptrace() on.
847 * This includes both natural children and PTRACE_ATTACH targets.
848 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
850 struct list_head ptraced
;
851 struct list_head ptrace_entry
;
853 /* PID/PID hash table linkage. */
854 struct pid_link pids
[PIDTYPE_MAX
];
855 struct list_head thread_group
;
856 struct list_head thread_node
;
858 struct completion
*vfork_done
;
860 /* CLONE_CHILD_SETTID: */
861 int __user
*set_child_tid
;
863 /* CLONE_CHILD_CLEARTID: */
864 int __user
*clear_child_tid
;
868 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
873 #ifdef CONFIG_CPU_FREQ_TIMES
875 unsigned int max_state
;
877 struct prev_cputime prev_cputime
;
878 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
882 #ifdef CONFIG_NO_HZ_FULL
883 atomic_t tick_dep_mask
;
885 /* Context switch counts: */
887 unsigned long nivcsw
;
889 /* Monotonic time in nsecs: */
892 /* Boot based time in nsecs: */
895 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
896 unsigned long min_flt
;
897 unsigned long maj_flt
;
899 #ifdef CONFIG_POSIX_TIMERS
900 struct task_cputime cputime_expires
;
901 struct list_head cpu_timers
[3];
904 /* Process credentials: */
906 /* Tracer's credentials at attach: */
907 const struct cred __rcu
*ptracer_cred
;
909 /* Objective and real subjective task credentials (COW): */
910 const struct cred __rcu
*real_cred
;
912 /* Effective (overridable) subjective task credentials (COW): */
913 const struct cred __rcu
*cred
;
916 * executable name, excluding path.
918 * - normally initialized setup_new_exec()
919 * - access it with [gs]et_task_comm()
920 * - lock it with task_lock()
922 char comm
[TASK_COMM_LEN
];
924 struct nameidata
*nameidata
;
926 #ifdef CONFIG_SYSVIPC
927 struct sysv_sem sysvsem
;
928 struct sysv_shm sysvshm
;
930 #ifdef CONFIG_DETECT_HUNG_TASK
931 unsigned long last_switch_count
;
933 /* Filesystem information: */
934 struct fs_struct
*fs
;
936 /* Open file information: */
937 struct files_struct
*files
;
940 struct nsproxy
*nsproxy
;
942 /* Signal handlers: */
943 struct signal_struct
*signal
;
944 struct sighand_struct
*sighand
;
946 sigset_t real_blocked
;
947 /* Restored if set_restore_sigmask() was used: */
948 sigset_t saved_sigmask
;
949 struct sigpending pending
;
950 unsigned long sas_ss_sp
;
952 unsigned int sas_ss_flags
;
954 struct callback_head
*task_works
;
956 struct audit_context
*audit_context
;
957 #ifdef CONFIG_AUDITSYSCALL
959 unsigned int sessionid
;
961 struct seccomp seccomp
;
963 /* Thread group tracking: */
967 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
968 spinlock_t alloc_lock
;
970 /* Protection of the PI data structures: */
971 raw_spinlock_t pi_lock
;
973 struct wake_q_node wake_q
;
975 #ifdef CONFIG_RT_MUTEXES
976 /* PI waiters blocked on a rt_mutex held by this task: */
977 struct rb_root_cached pi_waiters
;
978 /* Updated under owner's pi_lock and rq lock */
979 struct task_struct
*pi_top_task
;
980 /* Deadlock detection and priority inheritance handling: */
981 struct rt_mutex_waiter
*pi_blocked_on
;
984 #ifdef CONFIG_DEBUG_MUTEXES
985 /* Mutex deadlock detection: */
986 struct mutex_waiter
*blocked_on
;
989 #ifdef CONFIG_TRACE_IRQFLAGS
990 unsigned int irq_events
;
991 unsigned long hardirq_enable_ip
;
992 unsigned long hardirq_disable_ip
;
993 unsigned int hardirq_enable_event
;
994 unsigned int hardirq_disable_event
;
995 int hardirqs_enabled
;
997 unsigned long softirq_disable_ip
;
998 unsigned long softirq_enable_ip
;
999 unsigned int softirq_disable_event
;
1000 unsigned int softirq_enable_event
;
1001 int softirqs_enabled
;
1002 int softirq_context
;
1005 #ifdef CONFIG_LOCKDEP
1006 # define MAX_LOCK_DEPTH 48UL
1009 unsigned int lockdep_recursion
;
1010 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1013 #ifdef CONFIG_LOCKDEP_CROSSRELEASE
1014 #define MAX_XHLOCKS_NR 64UL
1015 struct hist_lock
*xhlocks
; /* Crossrelease history locks */
1016 unsigned int xhlock_idx
;
1017 /* For restoring at history boundaries */
1018 unsigned int xhlock_idx_hist
[XHLOCK_CTX_NR
];
1019 unsigned int hist_id
;
1020 /* For overwrite check at each context exit */
1021 unsigned int hist_id_save
[XHLOCK_CTX_NR
];
1025 unsigned int in_ubsan
;
1028 /* Journalling filesystem info: */
1031 /* Stacked block device info: */
1032 struct bio_list
*bio_list
;
1035 /* Stack plugging: */
1036 struct blk_plug
*plug
;
1040 struct reclaim_state
*reclaim_state
;
1042 struct backing_dev_info
*backing_dev_info
;
1044 struct io_context
*io_context
;
1047 unsigned long ptrace_message
;
1048 siginfo_t
*last_siginfo
;
1050 struct task_io_accounting ioac
;
1051 #ifdef CONFIG_TASK_XACCT
1052 /* Accumulated RSS usage: */
1054 /* Accumulated virtual memory usage: */
1056 /* stime + utime since last update: */
1059 #ifdef CONFIG_CPUSETS
1060 /* Protected by ->alloc_lock: */
1061 nodemask_t mems_allowed
;
1062 /* Seqence number to catch updates: */
1063 seqcount_t mems_allowed_seq
;
1064 int cpuset_mem_spread_rotor
;
1065 int cpuset_slab_spread_rotor
;
1067 #ifdef CONFIG_CGROUPS
1068 /* Control Group info protected by css_set_lock: */
1069 struct css_set __rcu
*cgroups
;
1070 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
1071 struct list_head cg_list
;
1073 #ifdef CONFIG_INTEL_RDT
1078 struct robust_list_head __user
*robust_list
;
1079 #ifdef CONFIG_COMPAT
1080 struct compat_robust_list_head __user
*compat_robust_list
;
1082 struct list_head pi_state_list
;
1083 struct futex_pi_state
*pi_state_cache
;
1085 #ifdef CONFIG_PERF_EVENTS
1086 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1087 struct mutex perf_event_mutex
;
1088 struct list_head perf_event_list
;
1090 #ifdef CONFIG_DEBUG_PREEMPT
1091 unsigned long preempt_disable_ip
;
1094 /* Protected by alloc_lock: */
1095 struct mempolicy
*mempolicy
;
1097 short pref_node_fork
;
1099 #ifdef CONFIG_NUMA_BALANCING
1101 unsigned int numa_scan_period
;
1102 unsigned int numa_scan_period_max
;
1103 int numa_preferred_nid
;
1104 unsigned long numa_migrate_retry
;
1105 /* Migration stamp: */
1107 u64 last_task_numa_placement
;
1108 u64 last_sum_exec_runtime
;
1109 struct callback_head numa_work
;
1111 struct list_head numa_entry
;
1112 struct numa_group
*numa_group
;
1115 * numa_faults is an array split into four regions:
1116 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1117 * in this precise order.
1119 * faults_memory: Exponential decaying average of faults on a per-node
1120 * basis. Scheduling placement decisions are made based on these
1121 * counts. The values remain static for the duration of a PTE scan.
1122 * faults_cpu: Track the nodes the process was running on when a NUMA
1123 * hinting fault was incurred.
1124 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1125 * during the current scan window. When the scan completes, the counts
1126 * in faults_memory and faults_cpu decay and these values are copied.
1128 unsigned long *numa_faults
;
1129 unsigned long total_numa_faults
;
1132 * numa_faults_locality tracks if faults recorded during the last
1133 * scan window were remote/local or failed to migrate. The task scan
1134 * period is adapted based on the locality of the faults with different
1135 * weights depending on whether they were shared or private faults
1137 unsigned long numa_faults_locality
[3];
1139 unsigned long numa_pages_migrated
;
1140 #endif /* CONFIG_NUMA_BALANCING */
1142 struct tlbflush_unmap_batch tlb_ubc
;
1144 struct rcu_head rcu
;
1146 /* Cache last used pipe for splice(): */
1147 struct pipe_inode_info
*splice_pipe
;
1149 struct page_frag task_frag
;
1151 #ifdef CONFIG_TASK_DELAY_ACCT
1152 struct task_delay_info
*delays
;
1155 #ifdef CONFIG_FAULT_INJECTION
1157 unsigned int fail_nth
;
1160 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1161 * balance_dirty_pages() for a dirty throttling pause:
1164 int nr_dirtied_pause
;
1165 /* Start of a write-and-pause period: */
1166 unsigned long dirty_paused_when
;
1168 #ifdef CONFIG_LATENCYTOP
1169 int latency_record_count
;
1170 struct latency_record latency_record
[LT_SAVECOUNT
];
1173 * Time slack values; these are used to round up poll() and
1174 * select() etc timeout values. These are in nanoseconds.
1177 u64 default_timer_slack_ns
;
1180 unsigned int kasan_depth
;
1183 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1184 /* Index of current stored address in ret_stack: */
1187 /* Stack of return addresses for return function tracing: */
1188 struct ftrace_ret_stack
*ret_stack
;
1190 /* Timestamp for last schedule: */
1191 unsigned long long ftrace_timestamp
;
1194 * Number of functions that haven't been traced
1195 * because of depth overrun:
1197 atomic_t trace_overrun
;
1199 /* Pause tracing: */
1200 atomic_t tracing_graph_pause
;
1203 #ifdef CONFIG_TRACING
1204 /* State flags for use by tracers: */
1205 unsigned long trace
;
1207 /* Bitmask and counter of trace recursion: */
1208 unsigned long trace_recursion
;
1209 #endif /* CONFIG_TRACING */
1212 /* Coverage collection mode enabled for this task (0 if disabled): */
1213 enum kcov_mode kcov_mode
;
1215 /* Size of the kcov_area: */
1216 unsigned int kcov_size
;
1218 /* Buffer for coverage collection: */
1221 /* KCOV descriptor wired with this task or NULL: */
1226 struct mem_cgroup
*memcg_in_oom
;
1227 gfp_t memcg_oom_gfp_mask
;
1228 int memcg_oom_order
;
1230 /* Number of pages to reclaim on returning to userland: */
1231 unsigned int memcg_nr_pages_over_high
;
1234 #ifdef CONFIG_UPROBES
1235 struct uprobe_task
*utask
;
1237 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1238 unsigned int sequential_io
;
1239 unsigned int sequential_io_avg
;
1241 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1242 unsigned long task_state_change
;
1244 int pagefault_disabled
;
1246 struct task_struct
*oom_reaper_list
;
1248 #ifdef CONFIG_VMAP_STACK
1249 struct vm_struct
*stack_vm_area
;
1251 #ifdef CONFIG_THREAD_INFO_IN_TASK
1252 /* A live task holds one reference: */
1253 atomic_t stack_refcount
;
1255 #ifdef CONFIG_LIVEPATCH
1258 #ifdef CONFIG_SECURITY
1259 /* Used by LSM modules for access restriction: */
1264 * New fields for task_struct should be added above here, so that
1265 * they are included in the randomized portion of task_struct.
1267 randomized_struct_fields_end
1269 /* CPU-specific state of this task: */
1270 struct thread_struct thread
;
1273 * WARNING: on x86, 'thread_struct' contains a variable-sized
1274 * structure. It *MUST* be at the end of 'task_struct'.
1276 * Do not put anything below here!
1280 static inline struct pid
*task_pid(struct task_struct
*task
)
1282 return task
->pids
[PIDTYPE_PID
].pid
;
1285 static inline struct pid
*task_tgid(struct task_struct
*task
)
1287 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1291 * Without tasklist or RCU lock it is not safe to dereference
1292 * the result of task_pgrp/task_session even if task == current,
1293 * we can race with another thread doing sys_setsid/sys_setpgid.
1295 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1297 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1300 static inline struct pid
*task_session(struct task_struct
*task
)
1302 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1306 * the helpers to get the task's different pids as they are seen
1307 * from various namespaces
1309 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1310 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1312 * task_xid_nr_ns() : id seen from the ns specified;
1314 * see also pid_nr() etc in include/linux/pid.h
1316 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
, struct pid_namespace
*ns
);
1318 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1323 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1325 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1328 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1330 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1334 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1340 * pid_alive - check that a task structure is not stale
1341 * @p: Task structure to be checked.
1343 * Test if a process is not yet dead (at most zombie state)
1344 * If pid_alive fails, then pointers within the task structure
1345 * can be stale and must not be dereferenced.
1347 * Return: 1 if the process is alive. 0 otherwise.
1349 static inline int pid_alive(const struct task_struct
*p
)
1351 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1354 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1356 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1359 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1361 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1365 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1367 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1370 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1372 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1375 static inline pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
)
1377 return __task_pid_nr_ns(tsk
, __PIDTYPE_TGID
, ns
);
1380 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1382 return __task_pid_nr_ns(tsk
, __PIDTYPE_TGID
, NULL
);
1385 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1391 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1397 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1399 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
1402 /* Obsolete, do not use: */
1403 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1405 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1408 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1409 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1411 static inline unsigned int __get_task_state(struct task_struct
*tsk
)
1413 unsigned int tsk_state
= READ_ONCE(tsk
->state
);
1414 unsigned int state
= (tsk_state
| tsk
->exit_state
) & TASK_REPORT
;
1416 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX
);
1418 if (tsk_state
== TASK_IDLE
)
1419 state
= TASK_REPORT_IDLE
;
1424 static inline char __task_state_to_char(unsigned int state
)
1426 static const char state_char
[] = "RSDTtXZPI";
1428 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX
) != sizeof(state_char
) - 1);
1430 return state_char
[state
];
1433 static inline char task_state_to_char(struct task_struct
*tsk
)
1435 return __task_state_to_char(__get_task_state(tsk
));
1439 * is_global_init - check if a task structure is init. Since init
1440 * is free to have sub-threads we need to check tgid.
1441 * @tsk: Task structure to be checked.
1443 * Check if a task structure is the first user space task the kernel created.
1445 * Return: 1 if the task structure is init. 0 otherwise.
1447 static inline int is_global_init(struct task_struct
*tsk
)
1449 return task_tgid_nr(tsk
) == 1;
1452 extern struct pid
*cad_pid
;
1457 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1458 #define PF_EXITING 0x00000004 /* Getting shut down */
1459 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1460 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1461 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1462 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1463 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1464 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1465 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1466 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1467 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1468 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1469 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1470 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1471 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1472 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1473 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1474 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1475 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1476 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1477 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1478 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1479 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1480 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1481 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1482 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
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_LMK_WAITING 7 /* Lowmemorykiller is waiting */
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_FORCE_DISABLE
, spec_ssb_force_disable
)
1562 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE
, spec_ssb_force_disable
)
1564 TASK_PFA_TEST(SPEC_IB_DISABLE
, spec_ib_disable
)
1565 TASK_PFA_SET(SPEC_IB_DISABLE
, spec_ib_disable
)
1566 TASK_PFA_CLEAR(SPEC_IB_DISABLE
, spec_ib_disable
)
1568 TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE
, spec_ib_force_disable
)
1569 TASK_PFA_SET(SPEC_IB_FORCE_DISABLE
, spec_ib_force_disable
)
1571 TASK_PFA_TEST(LMK_WAITING
, lmk_waiting
)
1572 TASK_PFA_SET(LMK_WAITING
, lmk_waiting
)
1575 current_restore_flags(unsigned long orig_flags
, unsigned long flags
)
1577 current
->flags
&= ~flags
;
1578 current
->flags
|= orig_flags
& flags
;
1581 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
, const struct cpumask
*trial
);
1582 extern int task_can_attach(struct task_struct
*p
, const struct cpumask
*cs_cpus_allowed
);
1584 extern void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
);
1585 extern int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
);
1587 static inline void do_set_cpus_allowed(struct task_struct
*p
, const struct cpumask
*new_mask
)
1590 static inline int set_cpus_allowed_ptr(struct task_struct
*p
, const struct cpumask
*new_mask
)
1592 if (!cpumask_test_cpu(0, new_mask
))
1598 #ifndef cpu_relax_yield
1599 #define cpu_relax_yield() cpu_relax()
1602 extern int yield_to(struct task_struct
*p
, bool preempt
);
1603 extern void set_user_nice(struct task_struct
*p
, long nice
);
1604 extern int task_prio(const struct task_struct
*p
);
1607 * task_nice - return the nice value of a given task.
1608 * @p: the task in question.
1610 * Return: The nice value [ -20 ... 0 ... 19 ].
1612 static inline int task_nice(const struct task_struct
*p
)
1614 return PRIO_TO_NICE((p
)->static_prio
);
1617 extern int can_nice(const struct task_struct
*p
, const int nice
);
1618 extern int task_curr(const struct task_struct
*p
);
1619 extern int idle_cpu(int cpu
);
1620 extern int sched_setscheduler(struct task_struct
*, int, const struct sched_param
*);
1621 extern int sched_setscheduler_nocheck(struct task_struct
*, int, const struct sched_param
*);
1622 extern int sched_setattr(struct task_struct
*, const struct sched_attr
*);
1623 extern struct task_struct
*idle_task(int cpu
);
1626 * is_idle_task - is the specified task an idle task?
1627 * @p: the task in question.
1629 * Return: 1 if @p is an idle task. 0 otherwise.
1631 static inline bool is_idle_task(const struct task_struct
*p
)
1633 return !!(p
->flags
& PF_IDLE
);
1636 extern struct task_struct
*curr_task(int cpu
);
1637 extern void ia64_set_curr_task(int cpu
, struct task_struct
*p
);
1641 union thread_union
{
1642 #ifndef CONFIG_THREAD_INFO_IN_TASK
1643 struct thread_info thread_info
;
1645 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1648 #ifdef CONFIG_THREAD_INFO_IN_TASK
1649 static inline struct thread_info
*task_thread_info(struct task_struct
*task
)
1651 return &task
->thread_info
;
1653 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1654 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1658 * find a task by one of its numerical ids
1660 * find_task_by_pid_ns():
1661 * finds a task by its pid in the specified namespace
1662 * find_task_by_vpid():
1663 * finds a task by its virtual pid
1665 * see also find_vpid() etc in include/linux/pid.h
1668 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1669 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
, struct pid_namespace
*ns
);
1671 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1672 extern int wake_up_process(struct task_struct
*tsk
);
1673 extern void wake_up_new_task(struct task_struct
*tsk
);
1676 extern void kick_process(struct task_struct
*tsk
);
1678 static inline void kick_process(struct task_struct
*tsk
) { }
1681 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
1683 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
1685 __set_task_comm(tsk
, from
, false);
1688 extern char *__get_task_comm(char *to
, size_t len
, struct task_struct
*tsk
);
1689 #define get_task_comm(buf, tsk) ({ \
1690 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1691 __get_task_comm(buf, sizeof(buf), tsk); \
1695 void scheduler_ipi(void);
1696 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
1698 static inline void scheduler_ipi(void) { }
1699 static inline unsigned long wait_task_inactive(struct task_struct
*p
, long match_state
)
1706 * Set thread flags in other task's structures.
1707 * See asm/thread_info.h for TIF_xxxx flags available:
1709 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1711 set_ti_thread_flag(task_thread_info(tsk
), flag
);
1714 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1716 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1719 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1721 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
1724 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1726 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
1729 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
1731 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
1734 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
1736 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1739 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
1741 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
1744 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
1746 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
1750 * cond_resched() and cond_resched_lock(): latency reduction via
1751 * explicit rescheduling in places that are safe. The return
1752 * value indicates whether a reschedule was done in fact.
1753 * cond_resched_lock() will drop the spinlock before scheduling,
1754 * cond_resched_softirq() will enable bhs before scheduling.
1756 #ifndef CONFIG_PREEMPT
1757 extern int _cond_resched(void);
1759 static inline int _cond_resched(void) { return 0; }
1762 #define cond_resched() ({ \
1763 ___might_sleep(__FILE__, __LINE__, 0); \
1767 extern int __cond_resched_lock(spinlock_t
*lock
);
1769 #define cond_resched_lock(lock) ({ \
1770 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1771 __cond_resched_lock(lock); \
1774 extern int __cond_resched_softirq(void);
1776 #define cond_resched_softirq() ({ \
1777 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
1778 __cond_resched_softirq(); \
1781 static inline void cond_resched_rcu(void)
1783 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1791 * Does a critical section need to be broken due to another
1792 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1793 * but a general need for low latency)
1795 static inline int spin_needbreak(spinlock_t
*lock
)
1797 #ifdef CONFIG_PREEMPT
1798 return spin_is_contended(lock
);
1804 static __always_inline
bool need_resched(void)
1806 return unlikely(tif_need_resched());
1810 * Wrappers for p->thread_info->cpu access. No-op on UP.
1814 static inline unsigned int task_cpu(const struct task_struct
*p
)
1816 #ifdef CONFIG_THREAD_INFO_IN_TASK
1819 return task_thread_info(p
)->cpu
;
1823 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
1827 static inline unsigned int task_cpu(const struct task_struct
*p
)
1832 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
1836 #endif /* CONFIG_SMP */
1839 * In order to reduce various lock holder preemption latencies provide an
1840 * interface to see if a vCPU is currently running or not.
1842 * This allows us to terminate optimistic spin loops and block, analogous to
1843 * the native optimistic spin heuristic of testing if the lock owner task is
1846 #ifndef vcpu_is_preempted
1847 # define vcpu_is_preempted(cpu) false
1850 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
1851 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
1853 #ifndef TASK_SIZE_OF
1854 #define TASK_SIZE_OF(tsk) TASK_SIZE