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>
30 #include <linux/rseq.h>
32 /* task_struct member predeclarations (sorted alphabetically): */
34 struct backing_dev_info;
39 struct futex_pi_state;
44 struct perf_event_context;
46 struct pipe_inode_info;
49 struct robust_list_head;
53 struct sighand_struct;
55 struct task_delay_info;
59 * Task state bitmask. NOTE! These bits are also
60 * encoded in fs/proc/array.c: get_task_state().
62 * We have two separate sets of flags: task->state
63 * is about runnability, while task->exit_state are
64 * about the task exiting. Confusing, but this way
65 * modifying one set can't modify the other one by
69 /* Used in tsk->state: */
70 #define TASK_RUNNING 0x0000
71 #define TASK_INTERRUPTIBLE 0x0001
72 #define TASK_UNINTERRUPTIBLE 0x0002
73 #define __TASK_STOPPED 0x0004
74 #define __TASK_TRACED 0x0008
75 /* Used in tsk->exit_state: */
76 #define EXIT_DEAD 0x0010
77 #define EXIT_ZOMBIE 0x0020
78 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
79 /* Used in tsk->state again: */
80 #define TASK_PARKED 0x0040
81 #define TASK_DEAD 0x0080
82 #define TASK_WAKEKILL 0x0100
83 #define TASK_WAKING 0x0200
84 #define TASK_NOLOAD 0x0400
85 #define TASK_NEW 0x0800
86 #define TASK_STATE_MAX 0x1000
88 /* Convenience macros for the sake of set_current_state: */
89 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
90 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
91 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
93 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
95 /* Convenience macros for the sake of wake_up(): */
96 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
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_PARKED | 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() executes a full memory barrier before accessing the
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
209 extern void scheduler_tick(void);
211 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
213 extern long schedule_timeout(long timeout);
214 extern long schedule_timeout_interruptible(long timeout);
215 extern long schedule_timeout_killable(long timeout);
216 extern long schedule_timeout_uninterruptible(long timeout);
217 extern long schedule_timeout_idle(long timeout);
218 asmlinkage void schedule(void);
219 extern void schedule_preempt_disabled(void);
221 extern int __must_check io_schedule_prepare(void);
222 extern void io_schedule_finish(int token);
223 extern long io_schedule_timeout(long timeout);
224 extern void io_schedule(void);
227 * struct prev_cputime - snapshot of system and user cputime
228 * @utime: time spent in user mode
229 * @stime: time spent in system mode
230 * @lock: protects the above two fields
232 * Stores previous user/system time values such that we can guarantee
235 struct prev_cputime {
236 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
244 * struct task_cputime - collected CPU time counts
245 * @utime: time spent in user mode, in nanoseconds
246 * @stime: time spent in kernel mode, in nanoseconds
247 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
249 * This structure groups together three kinds of CPU time that are tracked for
250 * threads and thread groups. Most things considering CPU time want to group
251 * these counts together and treat all three of them in parallel.
253 struct task_cputime {
256 unsigned long long sum_exec_runtime;
259 /* Alternate field names when used on cache expirations: */
260 #define virt_exp utime
261 #define prof_exp stime
262 #define sched_exp sum_exec_runtime
265 /* Task is sleeping or running in a CPU with VTIME inactive: */
267 /* Task runs in userspace in a CPU with VTIME active: */
269 /* Task runs in kernelspace in a CPU with VTIME active: */
275 unsigned long long starttime;
276 enum vtime_state state;
283 #ifdef CONFIG_SCHED_INFO
284 /* Cumulative counters: */
286 /* # of times we have run on this CPU: */
287 unsigned long pcount;
289 /* Time spent waiting on a runqueue: */
290 unsigned long long run_delay;
294 /* When did we last run on a CPU? */
295 unsigned long long last_arrival;
297 /* When were we last queued to run? */
298 unsigned long long last_queued;
300 #endif /* CONFIG_SCHED_INFO */
304 * Integer metrics need fixed point arithmetic, e.g., sched/fair
305 * has a few: load, load_avg, util_avg, freq, and capacity.
307 * We define a basic fixed point arithmetic range, and then formalize
308 * all these metrics based on that basic range.
310 # define SCHED_FIXEDPOINT_SHIFT 10
311 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
314 unsigned long weight;
319 * struct util_est - Estimation utilization of FAIR tasks
320 * @enqueued: instantaneous estimated utilization of a task/cpu
321 * @ewma: the Exponential Weighted Moving Average (EWMA)
322 * utilization of a task
324 * Support data structure to track an Exponential Weighted Moving Average
325 * (EWMA) of a FAIR task's utilization. New samples are added to the moving
326 * average each time a task completes an activation. Sample's weight is chosen
327 * so that the EWMA will be relatively insensitive to transient changes to the
330 * The enqueued attribute has a slightly different meaning for tasks and cpus:
331 * - task: the task's util_avg at last task dequeue time
332 * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU
333 * Thus, the util_est.enqueued of a task represents the contribution on the
334 * estimated utilization of the CPU where that task is currently enqueued.
336 * Only for tasks we track a moving average of the past instantaneous
337 * estimated utilization. This allows to absorb sporadic drops in utilization
338 * of an otherwise almost periodic task.
341 unsigned int enqueued;
343 #define UTIL_EST_WEIGHT_SHIFT 2
344 } __attribute__((__aligned__(sizeof(u64))));
347 * The load_avg/util_avg accumulates an infinite geometric series
348 * (see __update_load_avg() in kernel/sched/fair.c).
350 * [load_avg definition]
352 * load_avg = runnable% * scale_load_down(load)
354 * where runnable% is the time ratio that a sched_entity is runnable.
355 * For cfs_rq, it is the aggregated load_avg of all runnable and
356 * blocked sched_entities.
358 * load_avg may also take frequency scaling into account:
360 * load_avg = runnable% * scale_load_down(load) * freq%
362 * where freq% is the CPU frequency normalized to the highest frequency.
364 * [util_avg definition]
366 * util_avg = running% * SCHED_CAPACITY_SCALE
368 * where running% is the time ratio that a sched_entity is running on
369 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
370 * and blocked sched_entities.
372 * util_avg may also factor frequency scaling and CPU capacity scaling:
374 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
376 * where freq% is the same as above, and capacity% is the CPU capacity
377 * normalized to the greatest capacity (due to uarch differences, etc).
379 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
380 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
381 * we therefore scale them to as large a range as necessary. This is for
382 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
386 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
387 * with the highest load (=88761), always runnable on a single cfs_rq,
388 * and should not overflow as the number already hits PID_MAX_LIMIT.
390 * For all other cases (including 32-bit kernels), struct load_weight's
391 * weight will overflow first before we do, because:
393 * Max(load_avg) <= Max(load.weight)
395 * Then it is the load_weight's responsibility to consider overflow
399 u64 last_update_time;
401 u64 runnable_load_sum;
404 unsigned long load_avg;
405 unsigned long runnable_load_avg;
406 unsigned long util_avg;
407 struct util_est util_est;
408 } ____cacheline_aligned;
410 struct sched_statistics {
411 #ifdef CONFIG_SCHEDSTATS
421 s64 sum_sleep_runtime;
428 u64 nr_migrations_cold;
429 u64 nr_failed_migrations_affine;
430 u64 nr_failed_migrations_running;
431 u64 nr_failed_migrations_hot;
432 u64 nr_forced_migrations;
436 u64 nr_wakeups_migrate;
437 u64 nr_wakeups_local;
438 u64 nr_wakeups_remote;
439 u64 nr_wakeups_affine;
440 u64 nr_wakeups_affine_attempts;
441 u64 nr_wakeups_passive;
446 struct sched_entity {
447 /* For load-balancing: */
448 struct load_weight load;
449 unsigned long runnable_weight;
450 struct rb_node run_node;
451 struct list_head group_node;
455 u64 sum_exec_runtime;
457 u64 prev_sum_exec_runtime;
461 struct sched_statistics statistics;
463 #ifdef CONFIG_FAIR_GROUP_SCHED
465 struct sched_entity *parent;
466 /* rq on which this entity is (to be) queued: */
467 struct cfs_rq *cfs_rq;
468 /* rq "owned" by this entity/group: */
474 * Per entity load average tracking.
476 * Put into separate cache line so it does not
477 * collide with read-mostly values above.
479 struct sched_avg avg;
483 struct sched_rt_entity {
484 struct list_head run_list;
485 unsigned long timeout;
486 unsigned long watchdog_stamp;
487 unsigned int time_slice;
488 unsigned short on_rq;
489 unsigned short on_list;
491 struct sched_rt_entity *back;
492 #ifdef CONFIG_RT_GROUP_SCHED
493 struct sched_rt_entity *parent;
494 /* rq on which this entity is (to be) queued: */
496 /* rq "owned" by this entity/group: */
499 } __randomize_layout;
501 struct sched_dl_entity {
502 struct rb_node rb_node;
505 * Original scheduling parameters. Copied here from sched_attr
506 * during sched_setattr(), they will remain the same until
507 * the next sched_setattr().
509 u64 dl_runtime; /* Maximum runtime for each instance */
510 u64 dl_deadline; /* Relative deadline of each instance */
511 u64 dl_period; /* Separation of two instances (period) */
512 u64 dl_bw; /* dl_runtime / dl_period */
513 u64 dl_density; /* dl_runtime / dl_deadline */
516 * Actual scheduling parameters. Initialized with the values above,
517 * they are continously updated during task execution. Note that
518 * the remaining runtime could be < 0 in case we are in overrun.
520 s64 runtime; /* Remaining runtime for this instance */
521 u64 deadline; /* Absolute deadline for this instance */
522 unsigned int flags; /* Specifying the scheduler behaviour */
527 * @dl_throttled tells if we exhausted the runtime. If so, the
528 * task has to wait for a replenishment to be performed at the
529 * next firing of dl_timer.
531 * @dl_boosted tells if we are boosted due to DI. If so we are
532 * outside bandwidth enforcement mechanism (but only until we
533 * exit the critical section);
535 * @dl_yielded tells if task gave up the CPU before consuming
536 * all its available runtime during the last job.
538 * @dl_non_contending tells if the task is inactive while still
539 * contributing to the active utilization. In other words, it
540 * indicates if the inactive timer has been armed and its handler
541 * has not been executed yet. This flag is useful to avoid race
542 * conditions between the inactive timer handler and the wakeup
545 * @dl_overrun tells if the task asked to be informed about runtime
548 unsigned int dl_throttled : 1;
549 unsigned int dl_boosted : 1;
550 unsigned int dl_yielded : 1;
551 unsigned int dl_non_contending : 1;
552 unsigned int dl_overrun : 1;
555 * Bandwidth enforcement timer. Each -deadline task has its
556 * own bandwidth to be enforced, thus we need one timer per task.
558 struct hrtimer dl_timer;
561 * Inactive timer, responsible for decreasing the active utilization
562 * at the "0-lag time". When a -deadline task blocks, it contributes
563 * to GRUB's active utilization until the "0-lag time", hence a
564 * timer is needed to decrease the active utilization at the correct
567 struct hrtimer inactive_timer;
576 /* Otherwise the compiler can store garbage here: */
579 u32 s; /* Set of bits. */
582 enum perf_event_task_context {
583 perf_invalid_context = -1,
586 perf_nr_task_contexts,
590 struct wake_q_node *next;
594 #ifdef CONFIG_THREAD_INFO_IN_TASK
596 * For reasons of header soup (see current_thread_info()), this
597 * must be the first element of task_struct.
599 struct thread_info thread_info;
601 /* -1 unrunnable, 0 runnable, >0 stopped: */
605 * This begins the randomizable portion of task_struct. Only
606 * scheduling-critical items should be added above here.
608 randomized_struct_fields_start
612 /* Per task flags (PF_*), defined further below: */
617 struct llist_node wake_entry;
619 #ifdef CONFIG_THREAD_INFO_IN_TASK
623 unsigned int wakee_flips;
624 unsigned long wakee_flip_decay_ts;
625 struct task_struct *last_wakee;
628 * recent_used_cpu is initially set as the last CPU used by a task
629 * that wakes affine another task. Waker/wakee relationships can
630 * push tasks around a CPU where each wakeup moves to the next one.
631 * Tracking a recently used CPU allows a quick search for a recently
632 * used CPU that may be idle.
642 unsigned int rt_priority;
644 const struct sched_class *sched_class;
645 struct sched_entity se;
646 struct sched_rt_entity rt;
647 #ifdef CONFIG_CGROUP_SCHED
648 struct task_group *sched_task_group;
650 struct sched_dl_entity dl;
652 #ifdef CONFIG_PREEMPT_NOTIFIERS
653 /* List of struct preempt_notifier: */
654 struct hlist_head preempt_notifiers;
657 #ifdef CONFIG_BLK_DEV_IO_TRACE
658 unsigned int btrace_seq;
663 cpumask_t cpus_allowed;
665 #ifdef CONFIG_PREEMPT_RCU
666 int rcu_read_lock_nesting;
667 union rcu_special rcu_read_unlock_special;
668 struct list_head rcu_node_entry;
669 struct rcu_node *rcu_blocked_node;
670 #endif /* #ifdef CONFIG_PREEMPT_RCU */
672 #ifdef CONFIG_TASKS_RCU
673 unsigned long rcu_tasks_nvcsw;
674 u8 rcu_tasks_holdout;
676 int rcu_tasks_idle_cpu;
677 struct list_head rcu_tasks_holdout_list;
678 #endif /* #ifdef CONFIG_TASKS_RCU */
680 struct sched_info sched_info;
682 struct list_head tasks;
684 struct plist_node pushable_tasks;
685 struct rb_node pushable_dl_tasks;
688 struct mm_struct *mm;
689 struct mm_struct *active_mm;
691 /* Per-thread vma caching: */
692 struct vmacache vmacache;
694 #ifdef SPLIT_RSS_COUNTING
695 struct task_rss_stat rss_stat;
700 /* The signal sent when the parent dies: */
702 /* JOBCTL_*, siglock protected: */
703 unsigned long jobctl;
705 /* Used for emulating ABI behavior of previous Linux versions: */
706 unsigned int personality;
708 /* Scheduler bits, serialized by scheduler locks: */
709 unsigned sched_reset_on_fork:1;
710 unsigned sched_contributes_to_load:1;
711 unsigned sched_migrated:1;
712 unsigned sched_remote_wakeup:1;
713 /* Force alignment to the next boundary: */
716 /* Unserialized, strictly 'current' */
718 /* Bit to tell LSMs we're in execve(): */
719 unsigned in_execve:1;
720 unsigned in_iowait:1;
721 #ifndef TIF_RESTORE_SIGMASK
722 unsigned restore_sigmask:1;
725 unsigned in_user_fault:1;
726 #ifdef CONFIG_MEMCG_KMEM
727 unsigned memcg_kmem_skip_account:1;
730 #ifdef CONFIG_COMPAT_BRK
731 unsigned brk_randomized:1;
733 #ifdef CONFIG_CGROUPS
734 /* disallow userland-initiated cgroup migration */
735 unsigned no_cgroup_migration:1;
737 #ifdef CONFIG_BLK_CGROUP
738 /* to be used once the psi infrastructure lands upstream. */
739 unsigned use_memdelay:1;
743 * May usercopy functions fault on kernel addresses?
744 * This is not just a single bit because this can potentially nest.
746 unsigned int kernel_uaccess_faults_ok;
748 unsigned long atomic_flags; /* Flags requiring atomic access. */
750 struct restart_block restart_block;
755 #ifdef CONFIG_STACKPROTECTOR
756 /* Canary value for the -fstack-protector GCC feature: */
757 unsigned long stack_canary;
760 * Pointers to the (original) parent process, youngest child, younger sibling,
761 * older sibling, respectively. (p->father can be replaced with
762 * p->real_parent->pid)
765 /* Real parent process: */
766 struct task_struct __rcu *real_parent;
768 /* Recipient of SIGCHLD, wait4() reports: */
769 struct task_struct __rcu *parent;
772 * Children/sibling form the list of natural children:
774 struct list_head children;
775 struct list_head sibling;
776 struct task_struct *group_leader;
779 * 'ptraced' is the list of tasks this task is using ptrace() on.
781 * This includes both natural children and PTRACE_ATTACH targets.
782 * 'ptrace_entry' is this task's link on the p->parent->ptraced list.
784 struct list_head ptraced;
785 struct list_head ptrace_entry;
787 /* PID/PID hash table linkage. */
788 struct pid *thread_pid;
789 struct hlist_node pid_links[PIDTYPE_MAX];
790 struct list_head thread_group;
791 struct list_head thread_node;
793 struct completion *vfork_done;
795 /* CLONE_CHILD_SETTID: */
796 int __user *set_child_tid;
798 /* CLONE_CHILD_CLEARTID: */
799 int __user *clear_child_tid;
803 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
808 struct prev_cputime prev_cputime;
809 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
813 #ifdef CONFIG_NO_HZ_FULL
814 atomic_t tick_dep_mask;
816 /* Context switch counts: */
818 unsigned long nivcsw;
820 /* Monotonic time in nsecs: */
823 /* Boot based time in nsecs: */
826 /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */
827 unsigned long min_flt;
828 unsigned long maj_flt;
830 #ifdef CONFIG_POSIX_TIMERS
831 struct task_cputime cputime_expires;
832 struct list_head cpu_timers[3];
835 /* Process credentials: */
837 /* Tracer's credentials at attach: */
838 const struct cred __rcu *ptracer_cred;
840 /* Objective and real subjective task credentials (COW): */
841 const struct cred __rcu *real_cred;
843 /* Effective (overridable) subjective task credentials (COW): */
844 const struct cred __rcu *cred;
847 * executable name, excluding path.
849 * - normally initialized setup_new_exec()
850 * - access it with [gs]et_task_comm()
851 * - lock it with task_lock()
853 char comm[TASK_COMM_LEN];
855 struct nameidata *nameidata;
857 #ifdef CONFIG_SYSVIPC
858 struct sysv_sem sysvsem;
859 struct sysv_shm sysvshm;
861 #ifdef CONFIG_DETECT_HUNG_TASK
862 unsigned long last_switch_count;
863 unsigned long last_switch_time;
865 /* Filesystem information: */
866 struct fs_struct *fs;
868 /* Open file information: */
869 struct files_struct *files;
872 struct nsproxy *nsproxy;
874 /* Signal handlers: */
875 struct signal_struct *signal;
876 struct sighand_struct *sighand;
878 sigset_t real_blocked;
879 /* Restored if set_restore_sigmask() was used: */
880 sigset_t saved_sigmask;
881 struct sigpending pending;
882 unsigned long sas_ss_sp;
884 unsigned int sas_ss_flags;
886 struct callback_head *task_works;
888 struct audit_context *audit_context;
889 #ifdef CONFIG_AUDITSYSCALL
891 unsigned int sessionid;
893 struct seccomp seccomp;
895 /* Thread group tracking: */
899 /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */
900 spinlock_t alloc_lock;
902 /* Protection of the PI data structures: */
903 raw_spinlock_t pi_lock;
905 struct wake_q_node wake_q;
907 #ifdef CONFIG_RT_MUTEXES
908 /* PI waiters blocked on a rt_mutex held by this task: */
909 struct rb_root_cached pi_waiters;
910 /* Updated under owner's pi_lock and rq lock */
911 struct task_struct *pi_top_task;
912 /* Deadlock detection and priority inheritance handling: */
913 struct rt_mutex_waiter *pi_blocked_on;
916 #ifdef CONFIG_DEBUG_MUTEXES
917 /* Mutex deadlock detection: */
918 struct mutex_waiter *blocked_on;
921 #ifdef CONFIG_TRACE_IRQFLAGS
922 unsigned int irq_events;
923 unsigned long hardirq_enable_ip;
924 unsigned long hardirq_disable_ip;
925 unsigned int hardirq_enable_event;
926 unsigned int hardirq_disable_event;
927 int hardirqs_enabled;
929 unsigned long softirq_disable_ip;
930 unsigned long softirq_enable_ip;
931 unsigned int softirq_disable_event;
932 unsigned int softirq_enable_event;
933 int softirqs_enabled;
937 #ifdef CONFIG_LOCKDEP
938 # define MAX_LOCK_DEPTH 48UL
941 unsigned int lockdep_recursion;
942 struct held_lock held_locks[MAX_LOCK_DEPTH];
946 unsigned int in_ubsan;
949 /* Journalling filesystem info: */
952 /* Stacked block device info: */
953 struct bio_list *bio_list;
956 /* Stack plugging: */
957 struct blk_plug *plug;
961 struct reclaim_state *reclaim_state;
963 struct backing_dev_info *backing_dev_info;
965 struct io_context *io_context;
968 unsigned long ptrace_message;
969 siginfo_t *last_siginfo;
971 struct task_io_accounting ioac;
972 #ifdef CONFIG_TASK_XACCT
973 /* Accumulated RSS usage: */
975 /* Accumulated virtual memory usage: */
977 /* stime + utime since last update: */
980 #ifdef CONFIG_CPUSETS
981 /* Protected by ->alloc_lock: */
982 nodemask_t mems_allowed;
983 /* Seqence number to catch updates: */
984 seqcount_t mems_allowed_seq;
985 int cpuset_mem_spread_rotor;
986 int cpuset_slab_spread_rotor;
988 #ifdef CONFIG_CGROUPS
989 /* Control Group info protected by css_set_lock: */
990 struct css_set __rcu *cgroups;
991 /* cg_list protected by css_set_lock and tsk->alloc_lock: */
992 struct list_head cg_list;
994 #ifdef CONFIG_INTEL_RDT
999 struct robust_list_head __user *robust_list;
1000 #ifdef CONFIG_COMPAT
1001 struct compat_robust_list_head __user *compat_robust_list;
1003 struct list_head pi_state_list;
1004 struct futex_pi_state *pi_state_cache;
1006 #ifdef CONFIG_PERF_EVENTS
1007 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1008 struct mutex perf_event_mutex;
1009 struct list_head perf_event_list;
1011 #ifdef CONFIG_DEBUG_PREEMPT
1012 unsigned long preempt_disable_ip;
1015 /* Protected by alloc_lock: */
1016 struct mempolicy *mempolicy;
1018 short pref_node_fork;
1020 #ifdef CONFIG_NUMA_BALANCING
1022 unsigned int numa_scan_period;
1023 unsigned int numa_scan_period_max;
1024 int numa_preferred_nid;
1025 unsigned long numa_migrate_retry;
1026 /* Migration stamp: */
1028 u64 last_task_numa_placement;
1029 u64 last_sum_exec_runtime;
1030 struct callback_head numa_work;
1032 struct numa_group *numa_group;
1035 * numa_faults is an array split into four regions:
1036 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1037 * in this precise order.
1039 * faults_memory: Exponential decaying average of faults on a per-node
1040 * basis. Scheduling placement decisions are made based on these
1041 * counts. The values remain static for the duration of a PTE scan.
1042 * faults_cpu: Track the nodes the process was running on when a NUMA
1043 * hinting fault was incurred.
1044 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1045 * during the current scan window. When the scan completes, the counts
1046 * in faults_memory and faults_cpu decay and these values are copied.
1048 unsigned long *numa_faults;
1049 unsigned long total_numa_faults;
1052 * numa_faults_locality tracks if faults recorded during the last
1053 * scan window were remote/local or failed to migrate. The task scan
1054 * period is adapted based on the locality of the faults with different
1055 * weights depending on whether they were shared or private faults
1057 unsigned long numa_faults_locality[3];
1059 unsigned long numa_pages_migrated;
1060 #endif /* CONFIG_NUMA_BALANCING */
1063 struct rseq __user *rseq;
1067 * RmW on rseq_event_mask must be performed atomically
1068 * with respect to preemption.
1070 unsigned long rseq_event_mask;
1073 struct tlbflush_unmap_batch tlb_ubc;
1075 struct rcu_head rcu;
1077 /* Cache last used pipe for splice(): */
1078 struct pipe_inode_info *splice_pipe;
1080 struct page_frag task_frag;
1082 #ifdef CONFIG_TASK_DELAY_ACCT
1083 struct task_delay_info *delays;
1086 #ifdef CONFIG_FAULT_INJECTION
1088 unsigned int fail_nth;
1091 * When (nr_dirtied >= nr_dirtied_pause), it's time to call
1092 * balance_dirty_pages() for a dirty throttling pause:
1095 int nr_dirtied_pause;
1096 /* Start of a write-and-pause period: */
1097 unsigned long dirty_paused_when;
1099 #ifdef CONFIG_LATENCYTOP
1100 int latency_record_count;
1101 struct latency_record latency_record[LT_SAVECOUNT];
1104 * Time slack values; these are used to round up poll() and
1105 * select() etc timeout values. These are in nanoseconds.
1108 u64 default_timer_slack_ns;
1111 unsigned int kasan_depth;
1114 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1115 /* Index of current stored address in ret_stack: */
1118 /* Stack of return addresses for return function tracing: */
1119 struct ftrace_ret_stack *ret_stack;
1121 /* Timestamp for last schedule: */
1122 unsigned long long ftrace_timestamp;
1125 * Number of functions that haven't been traced
1126 * because of depth overrun:
1128 atomic_t trace_overrun;
1130 /* Pause tracing: */
1131 atomic_t tracing_graph_pause;
1134 #ifdef CONFIG_TRACING
1135 /* State flags for use by tracers: */
1136 unsigned long trace;
1138 /* Bitmask and counter of trace recursion: */
1139 unsigned long trace_recursion;
1140 #endif /* CONFIG_TRACING */
1143 /* Coverage collection mode enabled for this task (0 if disabled): */
1144 unsigned int kcov_mode;
1146 /* Size of the kcov_area: */
1147 unsigned int kcov_size;
1149 /* Buffer for coverage collection: */
1152 /* KCOV descriptor wired with this task or NULL: */
1157 struct mem_cgroup *memcg_in_oom;
1158 gfp_t memcg_oom_gfp_mask;
1159 int memcg_oom_order;
1161 /* Number of pages to reclaim on returning to userland: */
1162 unsigned int memcg_nr_pages_over_high;
1164 /* Used by memcontrol for targeted memcg charge: */
1165 struct mem_cgroup *active_memcg;
1168 #ifdef CONFIG_BLK_CGROUP
1169 struct request_queue *throttle_queue;
1172 #ifdef CONFIG_UPROBES
1173 struct uprobe_task *utask;
1175 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1176 unsigned int sequential_io;
1177 unsigned int sequential_io_avg;
1179 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1180 unsigned long task_state_change;
1182 int pagefault_disabled;
1184 struct task_struct *oom_reaper_list;
1186 #ifdef CONFIG_VMAP_STACK
1187 struct vm_struct *stack_vm_area;
1189 #ifdef CONFIG_THREAD_INFO_IN_TASK
1190 /* A live task holds one reference: */
1191 atomic_t stack_refcount;
1193 #ifdef CONFIG_LIVEPATCH
1196 #ifdef CONFIG_SECURITY
1197 /* Used by LSM modules for access restriction: */
1202 * New fields for task_struct should be added above here, so that
1203 * they are included in the randomized portion of task_struct.
1205 randomized_struct_fields_end
1207 /* CPU-specific state of this task: */
1208 struct thread_struct thread;
1211 * WARNING: on x86, 'thread_struct' contains a variable-sized
1212 * structure. It *MUST* be at the end of 'task_struct'.
1214 * Do not put anything below here!
1218 static inline struct pid *task_pid(struct task_struct *task)
1220 return task->thread_pid;
1224 * the helpers to get the task's different pids as they are seen
1225 * from various namespaces
1227 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1228 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1230 * task_xid_nr_ns() : id seen from the ns specified;
1232 * see also pid_nr() etc in include/linux/pid.h
1234 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns);
1236 static inline pid_t task_pid_nr(struct task_struct *tsk)
1241 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1243 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1246 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1248 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1252 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1258 * pid_alive - check that a task structure is not stale
1259 * @p: Task structure to be checked.
1261 * Test if a process is not yet dead (at most zombie state)
1262 * If pid_alive fails, then pointers within the task structure
1263 * can be stale and must not be dereferenced.
1265 * Return: 1 if the process is alive. 0 otherwise.
1267 static inline int pid_alive(const struct task_struct *p)
1269 return p->thread_pid != NULL;
1272 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1274 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1277 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1279 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1283 static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1285 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1288 static inline pid_t task_session_vnr(struct task_struct *tsk)
1290 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1293 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
1295 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns);
1298 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1300 return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL);
1303 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1309 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1315 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1317 return task_ppid_nr_ns(tsk, &init_pid_ns);
1320 /* Obsolete, do not use: */
1321 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1323 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1326 #define TASK_REPORT_IDLE (TASK_REPORT + 1)
1327 #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1)
1329 static inline unsigned int task_state_index(struct task_struct *tsk)
1331 unsigned int tsk_state = READ_ONCE(tsk->state);
1332 unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT;
1334 BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX);
1336 if (tsk_state == TASK_IDLE)
1337 state = TASK_REPORT_IDLE;
1342 static inline char task_index_to_char(unsigned int state)
1344 static const char state_char[] = "RSDTtXZPI";
1346 BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1);
1348 return state_char[state];
1351 static inline char task_state_to_char(struct task_struct *tsk)
1353 return task_index_to_char(task_state_index(tsk));
1357 * is_global_init - check if a task structure is init. Since init
1358 * is free to have sub-threads we need to check tgid.
1359 * @tsk: Task structure to be checked.
1361 * Check if a task structure is the first user space task the kernel created.
1363 * Return: 1 if the task structure is init. 0 otherwise.
1365 static inline int is_global_init(struct task_struct *tsk)
1367 return task_tgid_nr(tsk) == 1;
1370 extern struct pid *cad_pid;
1375 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1376 #define PF_EXITING 0x00000004 /* Getting shut down */
1377 #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */
1378 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1379 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1380 #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */
1381 #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */
1382 #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */
1383 #define PF_DUMPCORE 0x00000200 /* Dumped core */
1384 #define PF_SIGNALED 0x00000400 /* Killed by a signal */
1385 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1386 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */
1387 #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */
1388 #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */
1389 #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */
1390 #define PF_FROZEN 0x00010000 /* Frozen for system suspend */
1391 #define PF_KSWAPD 0x00020000 /* I am kswapd */
1392 #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */
1393 #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */
1394 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1395 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1396 #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */
1397 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1398 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1399 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1400 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1401 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1402 #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */
1405 * Only the _current_ task can read/write to tsk->flags, but other
1406 * tasks can access tsk->flags in readonly mode for example
1407 * with tsk_used_math (like during threaded core dumping).
1408 * There is however an exception to this rule during ptrace
1409 * or during fork: the ptracer task is allowed to write to the
1410 * child->flags of its traced child (same goes for fork, the parent
1411 * can write to the child->flags), because we're guaranteed the
1412 * child is not running and in turn not changing child->flags
1413 * at the same time the parent does it.
1415 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1416 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1417 #define clear_used_math() clear_stopped_child_used_math(current)
1418 #define set_used_math() set_stopped_child_used_math(current)
1420 #define conditional_stopped_child_used_math(condition, child) \
1421 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1423 #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current)
1425 #define copy_to_stopped_child_used_math(child) \
1426 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1428 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1429 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1430 #define used_math() tsk_used_math(current)
1432 static inline bool is_percpu_thread(void)
1435 return (current->flags & PF_NO_SETAFFINITY) &&
1436 (current->nr_cpus_allowed == 1);
1442 /* Per-process atomic flags. */
1443 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1444 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1445 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1446 #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */
1447 #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/
1449 #define TASK_PFA_TEST(name, func) \
1450 static inline bool task_##func(struct task_struct *p) \
1451 { return test_bit(PFA_##name, &p->atomic_flags); }
1453 #define TASK_PFA_SET(name, func) \
1454 static inline void task_set_##func(struct task_struct *p) \
1455 { set_bit(PFA_##name, &p->atomic_flags); }
1457 #define TASK_PFA_CLEAR(name, func) \
1458 static inline void task_clear_##func(struct task_struct *p) \
1459 { clear_bit(PFA_##name, &p->atomic_flags); }
1461 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1462 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1464 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1465 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1466 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1468 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1469 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1470 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1472 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
1473 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
1474 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
1476 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1477 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
1480 current_restore_flags(unsigned long orig_flags, unsigned long flags)
1482 current->flags &= ~flags;
1483 current->flags |= orig_flags & flags;
1486 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial);
1487 extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed);
1489 extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask);
1490 extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask);
1492 static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask)
1495 static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask)
1497 if (!cpumask_test_cpu(0, new_mask))
1503 #ifndef cpu_relax_yield
1504 #define cpu_relax_yield() cpu_relax()
1507 extern int yield_to(struct task_struct *p, bool preempt);
1508 extern void set_user_nice(struct task_struct *p, long nice);
1509 extern int task_prio(const struct task_struct *p);
1512 * task_nice - return the nice value of a given task.
1513 * @p: the task in question.
1515 * Return: The nice value [ -20 ... 0 ... 19 ].
1517 static inline int task_nice(const struct task_struct *p)
1519 return PRIO_TO_NICE((p)->static_prio);
1522 extern int can_nice(const struct task_struct *p, const int nice);
1523 extern int task_curr(const struct task_struct *p);
1524 extern int idle_cpu(int cpu);
1525 extern int available_idle_cpu(int cpu);
1526 extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *);
1527 extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *);
1528 extern int sched_setattr(struct task_struct *, const struct sched_attr *);
1529 extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *);
1530 extern struct task_struct *idle_task(int cpu);
1533 * is_idle_task - is the specified task an idle task?
1534 * @p: the task in question.
1536 * Return: 1 if @p is an idle task. 0 otherwise.
1538 static inline bool is_idle_task(const struct task_struct *p)
1540 return !!(p->flags & PF_IDLE);
1543 extern struct task_struct *curr_task(int cpu);
1544 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1548 union thread_union {
1549 #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK
1550 struct task_struct task;
1552 #ifndef CONFIG_THREAD_INFO_IN_TASK
1553 struct thread_info thread_info;
1555 unsigned long stack[THREAD_SIZE/sizeof(long)];
1558 #ifndef CONFIG_THREAD_INFO_IN_TASK
1559 extern struct thread_info init_thread_info;
1562 extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)];
1564 #ifdef CONFIG_THREAD_INFO_IN_TASK
1565 static inline struct thread_info *task_thread_info(struct task_struct *task)
1567 return &task->thread_info;
1569 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1570 # define task_thread_info(task) ((struct thread_info *)(task)->stack)
1574 * find a task by one of its numerical ids
1576 * find_task_by_pid_ns():
1577 * finds a task by its pid in the specified namespace
1578 * find_task_by_vpid():
1579 * finds a task by its virtual pid
1581 * see also find_vpid() etc in include/linux/pid.h
1584 extern struct task_struct *find_task_by_vpid(pid_t nr);
1585 extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns);
1588 * find a task by its virtual pid and get the task struct
1590 extern struct task_struct *find_get_task_by_vpid(pid_t nr);
1592 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1593 extern int wake_up_process(struct task_struct *tsk);
1594 extern void wake_up_new_task(struct task_struct *tsk);
1597 extern void kick_process(struct task_struct *tsk);
1599 static inline void kick_process(struct task_struct *tsk) { }
1602 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1604 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1606 __set_task_comm(tsk, from, false);
1609 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
1610 #define get_task_comm(buf, tsk) ({ \
1611 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
1612 __get_task_comm(buf, sizeof(buf), tsk); \
1616 void scheduler_ipi(void);
1617 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1619 static inline void scheduler_ipi(void) { }
1620 static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state)
1627 * Set thread flags in other task's structures.
1628 * See asm/thread_info.h for TIF_xxxx flags available:
1630 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
1632 set_ti_thread_flag(task_thread_info(tsk), flag);
1635 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1637 clear_ti_thread_flag(task_thread_info(tsk), flag);
1640 static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag,
1643 update_ti_thread_flag(task_thread_info(tsk), flag, value);
1646 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
1648 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
1651 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
1653 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
1656 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
1658 return test_ti_thread_flag(task_thread_info(tsk), flag);
1661 static inline void set_tsk_need_resched(struct task_struct *tsk)
1663 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1666 static inline void clear_tsk_need_resched(struct task_struct *tsk)
1668 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
1671 static inline int test_tsk_need_resched(struct task_struct *tsk)
1673 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
1677 * cond_resched() and cond_resched_lock(): latency reduction via
1678 * explicit rescheduling in places that are safe. The return
1679 * value indicates whether a reschedule was done in fact.
1680 * cond_resched_lock() will drop the spinlock before scheduling,
1682 #ifndef CONFIG_PREEMPT
1683 extern int _cond_resched(void);
1685 static inline int _cond_resched(void) { return 0; }
1688 #define cond_resched() ({ \
1689 ___might_sleep(__FILE__, __LINE__, 0); \
1693 extern int __cond_resched_lock(spinlock_t *lock);
1695 #define cond_resched_lock(lock) ({ \
1696 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
1697 __cond_resched_lock(lock); \
1700 static inline void cond_resched_rcu(void)
1702 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
1710 * Does a critical section need to be broken due to another
1711 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
1712 * but a general need for low latency)
1714 static inline int spin_needbreak(spinlock_t *lock)
1716 #ifdef CONFIG_PREEMPT
1717 return spin_is_contended(lock);
1723 static __always_inline bool need_resched(void)
1725 return unlikely(tif_need_resched());
1729 * Wrappers for p->thread_info->cpu access. No-op on UP.
1733 static inline unsigned int task_cpu(const struct task_struct *p)
1735 #ifdef CONFIG_THREAD_INFO_IN_TASK
1738 return task_thread_info(p)->cpu;
1742 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
1746 static inline unsigned int task_cpu(const struct task_struct *p)
1751 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
1755 #endif /* CONFIG_SMP */
1758 * In order to reduce various lock holder preemption latencies provide an
1759 * interface to see if a vCPU is currently running or not.
1761 * This allows us to terminate optimistic spin loops and block, analogous to
1762 * the native optimistic spin heuristic of testing if the lock owner task is
1765 #ifndef vcpu_is_preempted
1766 # define vcpu_is_preempted(cpu) false
1769 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
1770 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
1772 #ifndef TASK_SIZE_OF
1773 #define TASK_SIZE_OF(tsk) TASK_SIZE
1779 * Map the event mask on the user-space ABI enum rseq_cs_flags
1780 * for direct mask checks.
1782 enum rseq_event_mask_bits {
1783 RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT,
1784 RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT,
1785 RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT,
1788 enum rseq_event_mask {
1789 RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT),
1790 RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT),
1791 RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT),
1794 static inline void rseq_set_notify_resume(struct task_struct *t)
1797 set_tsk_thread_flag(t, TIF_NOTIFY_RESUME);
1800 void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs);
1802 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1803 struct pt_regs *regs)
1806 __rseq_handle_notify_resume(ksig, regs);
1809 static inline void rseq_signal_deliver(struct ksignal *ksig,
1810 struct pt_regs *regs)
1813 __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask);
1815 rseq_handle_notify_resume(ksig, regs);
1818 /* rseq_preempt() requires preemption to be disabled. */
1819 static inline void rseq_preempt(struct task_struct *t)
1821 __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask);
1822 rseq_set_notify_resume(t);
1825 /* rseq_migrate() requires preemption to be disabled. */
1826 static inline void rseq_migrate(struct task_struct *t)
1828 __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask);
1829 rseq_set_notify_resume(t);
1833 * If parent process has a registered restartable sequences area, the
1834 * child inherits. Only applies when forking a process, not a thread.
1836 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1838 if (clone_flags & CLONE_THREAD) {
1842 t->rseq_event_mask = 0;
1844 t->rseq = current->rseq;
1845 t->rseq_len = current->rseq_len;
1846 t->rseq_sig = current->rseq_sig;
1847 t->rseq_event_mask = current->rseq_event_mask;
1851 static inline void rseq_execve(struct task_struct *t)
1856 t->rseq_event_mask = 0;
1861 static inline void rseq_set_notify_resume(struct task_struct *t)
1864 static inline void rseq_handle_notify_resume(struct ksignal *ksig,
1865 struct pt_regs *regs)
1868 static inline void rseq_signal_deliver(struct ksignal *ksig,
1869 struct pt_regs *regs)
1872 static inline void rseq_preempt(struct task_struct *t)
1875 static inline void rseq_migrate(struct task_struct *t)
1878 static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags)
1881 static inline void rseq_execve(struct task_struct *t)
1887 #ifdef CONFIG_DEBUG_RSEQ
1889 void rseq_syscall(struct pt_regs *regs);
1893 static inline void rseq_syscall(struct pt_regs *regs)