4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
8 #include <asm/param.h> /* for HZ */
10 #include <linux/capability.h>
11 #include <linux/threads.h>
12 #include <linux/kernel.h>
13 #include <linux/types.h>
14 #include <linux/timex.h>
15 #include <linux/jiffies.h>
16 #include <linux/mutex.h>
17 #include <linux/plist.h>
18 #include <linux/rbtree.h>
19 #include <linux/thread_info.h>
20 #include <linux/cpumask.h>
21 #include <linux/errno.h>
22 #include <linux/nodemask.h>
23 #include <linux/mm_types.h>
24 #include <linux/preempt.h>
27 #include <asm/ptrace.h>
29 #include <linux/smp.h>
30 #include <linux/sem.h>
31 #include <linux/shm.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/signal_types.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/kcov.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
56 #include <linux/topology.h>
57 #include <linux/magic.h>
58 #include <linux/cgroup-defs.h>
60 #include <asm/processor.h>
65 struct futex_pi_state;
66 struct robust_list_head;
69 struct perf_event_context;
75 struct sighand_struct;
77 extern unsigned long total_forks;
78 extern int nr_threads;
79 DECLARE_PER_CPU(unsigned long, process_counts);
80 extern int nr_processes(void);
81 extern unsigned long nr_running(void);
82 extern bool single_task_running(void);
83 extern unsigned long nr_iowait(void);
84 extern unsigned long nr_iowait_cpu(int cpu);
85 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
87 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
88 extern void cpu_load_update_nohz_start(void);
89 extern void cpu_load_update_nohz_stop(void);
91 static inline void cpu_load_update_nohz_start(void) { }
92 static inline void cpu_load_update_nohz_stop(void) { }
95 extern void dump_cpu_task(int cpu);
100 #ifdef CONFIG_SCHED_DEBUG
101 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
102 extern void proc_sched_set_task(struct task_struct *p);
106 * Task state bitmask. NOTE! These bits are also
107 * encoded in fs/proc/array.c: get_task_state().
109 * We have two separate sets of flags: task->state
110 * is about runnability, while task->exit_state are
111 * about the task exiting. Confusing, but this way
112 * modifying one set can't modify the other one by
115 #define TASK_RUNNING 0
116 #define TASK_INTERRUPTIBLE 1
117 #define TASK_UNINTERRUPTIBLE 2
118 #define __TASK_STOPPED 4
119 #define __TASK_TRACED 8
120 /* in tsk->exit_state */
122 #define EXIT_ZOMBIE 32
123 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
124 /* in tsk->state again */
126 #define TASK_WAKEKILL 128
127 #define TASK_WAKING 256
128 #define TASK_PARKED 512
129 #define TASK_NOLOAD 1024
130 #define TASK_NEW 2048
131 #define TASK_STATE_MAX 4096
133 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
135 /* Convenience macros for the sake of set_current_state */
136 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
137 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
138 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
140 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
142 /* Convenience macros for the sake of wake_up */
143 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
144 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
146 /* get_task_state() */
147 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
148 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
149 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
151 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
152 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
153 #define task_is_stopped_or_traced(task) \
154 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
155 #define task_contributes_to_load(task) \
156 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
157 (task->flags & PF_FROZEN) == 0 && \
158 (task->state & TASK_NOLOAD) == 0)
160 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
162 #define __set_current_state(state_value) \
164 current->task_state_change = _THIS_IP_; \
165 current->state = (state_value); \
167 #define set_current_state(state_value) \
169 current->task_state_change = _THIS_IP_; \
170 smp_store_mb(current->state, (state_value)); \
175 * set_current_state() includes a barrier so that the write of current->state
176 * is correctly serialised wrt the caller's subsequent test of whether to
180 * set_current_state(TASK_UNINTERRUPTIBLE);
186 * __set_current_state(TASK_RUNNING);
188 * If the caller does not need such serialisation (because, for instance, the
189 * condition test and condition change and wakeup are under the same lock) then
190 * use __set_current_state().
192 * The above is typically ordered against the wakeup, which does:
194 * need_sleep = false;
195 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
197 * Where wake_up_state() (and all other wakeup primitives) imply enough
198 * barriers to order the store of the variable against wakeup.
200 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
201 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
202 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
204 * This is obviously fine, since they both store the exact same value.
206 * Also see the comments of try_to_wake_up().
208 #define __set_current_state(state_value) \
209 do { current->state = (state_value); } while (0)
210 #define set_current_state(state_value) \
211 smp_store_mb(current->state, (state_value))
215 /* Task command name length */
216 #define TASK_COMM_LEN 16
218 #include <linux/spinlock.h>
221 * This serializes "schedule()" and also protects
222 * the run-queue from deletions/modifications (but
223 * _adding_ to the beginning of the run-queue has
226 extern rwlock_t tasklist_lock;
227 extern spinlock_t mmlist_lock;
231 #ifdef CONFIG_PROVE_RCU
232 extern int lockdep_tasklist_lock_is_held(void);
233 #endif /* #ifdef CONFIG_PROVE_RCU */
235 extern void sched_init(void);
236 extern void sched_init_smp(void);
237 extern asmlinkage void schedule_tail(struct task_struct *prev);
238 extern void init_idle(struct task_struct *idle, int cpu);
239 extern void init_idle_bootup_task(struct task_struct *idle);
241 extern cpumask_var_t cpu_isolated_map;
243 extern int runqueue_is_locked(int cpu);
245 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
246 extern void nohz_balance_enter_idle(int cpu);
247 extern void set_cpu_sd_state_idle(void);
248 extern int get_nohz_timer_target(void);
250 static inline void nohz_balance_enter_idle(int cpu) { }
251 static inline void set_cpu_sd_state_idle(void) { }
255 * Only dump TASK_* tasks. (0 for all tasks)
257 extern void show_state_filter(unsigned long state_filter);
259 static inline void show_state(void)
261 show_state_filter(0);
264 extern void show_regs(struct pt_regs *);
267 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
268 * task), SP is the stack pointer of the first frame that should be shown in the back
269 * trace (or NULL if the entire call-chain of the task should be shown).
271 extern void show_stack(struct task_struct *task, unsigned long *sp);
273 extern void cpu_init (void);
274 extern void trap_init(void);
275 extern void update_process_times(int user);
276 extern void scheduler_tick(void);
277 extern int sched_cpu_starting(unsigned int cpu);
278 extern int sched_cpu_activate(unsigned int cpu);
279 extern int sched_cpu_deactivate(unsigned int cpu);
281 #ifdef CONFIG_HOTPLUG_CPU
282 extern int sched_cpu_dying(unsigned int cpu);
284 # define sched_cpu_dying NULL
287 extern void sched_show_task(struct task_struct *p);
289 /* Attach to any functions which should be ignored in wchan output. */
290 #define __sched __attribute__((__section__(".sched.text")))
292 /* Linker adds these: start and end of __sched functions */
293 extern char __sched_text_start[], __sched_text_end[];
295 /* Is this address in the __sched functions? */
296 extern int in_sched_functions(unsigned long addr);
298 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
299 extern signed long schedule_timeout(signed long timeout);
300 extern signed long schedule_timeout_interruptible(signed long timeout);
301 extern signed long schedule_timeout_killable(signed long timeout);
302 extern signed long schedule_timeout_uninterruptible(signed long timeout);
303 extern signed long schedule_timeout_idle(signed long timeout);
304 asmlinkage void schedule(void);
305 extern void schedule_preempt_disabled(void);
307 extern int __must_check io_schedule_prepare(void);
308 extern void io_schedule_finish(int token);
309 extern long io_schedule_timeout(long timeout);
310 extern void io_schedule(void);
312 void __noreturn do_task_dead(void);
317 * struct prev_cputime - snaphsot of system and user cputime
318 * @utime: time spent in user mode
319 * @stime: time spent in system mode
320 * @lock: protects the above two fields
322 * Stores previous user/system time values such that we can guarantee
325 struct prev_cputime {
326 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
333 static inline void prev_cputime_init(struct prev_cputime *prev)
335 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
336 prev->utime = prev->stime = 0;
337 raw_spin_lock_init(&prev->lock);
342 * struct task_cputime - collected CPU time counts
343 * @utime: time spent in user mode, in nanoseconds
344 * @stime: time spent in kernel mode, in nanoseconds
345 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
347 * This structure groups together three kinds of CPU time that are tracked for
348 * threads and thread groups. Most things considering CPU time want to group
349 * these counts together and treat all three of them in parallel.
351 struct task_cputime {
354 unsigned long long sum_exec_runtime;
357 /* Alternate field names when used to cache expirations. */
358 #define virt_exp utime
359 #define prof_exp stime
360 #define sched_exp sum_exec_runtime
363 * This is the atomic variant of task_cputime, which can be used for
364 * storing and updating task_cputime statistics without locking.
366 struct task_cputime_atomic {
369 atomic64_t sum_exec_runtime;
372 #define INIT_CPUTIME_ATOMIC \
373 (struct task_cputime_atomic) { \
374 .utime = ATOMIC64_INIT(0), \
375 .stime = ATOMIC64_INIT(0), \
376 .sum_exec_runtime = ATOMIC64_INIT(0), \
379 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
382 * Disable preemption until the scheduler is running -- use an unconditional
383 * value so that it also works on !PREEMPT_COUNT kernels.
385 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
387 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
390 * Initial preempt_count value; reflects the preempt_count schedule invariant
391 * which states that during context switches:
393 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
395 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
396 * Note: See finish_task_switch().
398 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
401 * struct thread_group_cputimer - thread group interval timer counts
402 * @cputime_atomic: atomic thread group interval timers.
403 * @running: true when there are timers running and
404 * @cputime_atomic receives updates.
405 * @checking_timer: true when a thread in the group is in the
406 * process of checking for thread group timers.
408 * This structure contains the version of task_cputime, above, that is
409 * used for thread group CPU timer calculations.
411 struct thread_group_cputimer {
412 struct task_cputime_atomic cputime_atomic;
417 #include <linux/rwsem.h>
420 struct backing_dev_info;
421 struct reclaim_state;
423 #ifdef CONFIG_SCHED_INFO
425 /* cumulative counters */
426 unsigned long pcount; /* # of times run on this cpu */
427 unsigned long long run_delay; /* time spent waiting on a runqueue */
430 unsigned long long last_arrival,/* when we last ran on a cpu */
431 last_queued; /* when we were last queued to run */
433 #endif /* CONFIG_SCHED_INFO */
435 struct task_delay_info;
437 static inline int sched_info_on(void)
439 #ifdef CONFIG_SCHEDSTATS
441 #elif defined(CONFIG_TASK_DELAY_ACCT)
442 extern int delayacct_on;
449 #ifdef CONFIG_SCHEDSTATS
450 void force_schedstat_enabled(void);
454 * Integer metrics need fixed point arithmetic, e.g., sched/fair
455 * has a few: load, load_avg, util_avg, freq, and capacity.
457 * We define a basic fixed point arithmetic range, and then formalize
458 * all these metrics based on that basic range.
460 # define SCHED_FIXEDPOINT_SHIFT 10
461 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
463 struct io_context; /* See blkdev.h */
466 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
467 extern void prefetch_stack(struct task_struct *t);
469 static inline void prefetch_stack(struct task_struct *t) { }
472 struct audit_context; /* See audit.c */
474 struct pipe_inode_info;
475 struct uts_namespace;
478 unsigned long weight;
483 * The load_avg/util_avg accumulates an infinite geometric series
484 * (see __update_load_avg() in kernel/sched/fair.c).
486 * [load_avg definition]
488 * load_avg = runnable% * scale_load_down(load)
490 * where runnable% is the time ratio that a sched_entity is runnable.
491 * For cfs_rq, it is the aggregated load_avg of all runnable and
492 * blocked sched_entities.
494 * load_avg may also take frequency scaling into account:
496 * load_avg = runnable% * scale_load_down(load) * freq%
498 * where freq% is the CPU frequency normalized to the highest frequency.
500 * [util_avg definition]
502 * util_avg = running% * SCHED_CAPACITY_SCALE
504 * where running% is the time ratio that a sched_entity is running on
505 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
506 * and blocked sched_entities.
508 * util_avg may also factor frequency scaling and CPU capacity scaling:
510 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
512 * where freq% is the same as above, and capacity% is the CPU capacity
513 * normalized to the greatest capacity (due to uarch differences, etc).
515 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
516 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
517 * we therefore scale them to as large a range as necessary. This is for
518 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
522 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
523 * with the highest load (=88761), always runnable on a single cfs_rq,
524 * and should not overflow as the number already hits PID_MAX_LIMIT.
526 * For all other cases (including 32-bit kernels), struct load_weight's
527 * weight will overflow first before we do, because:
529 * Max(load_avg) <= Max(load.weight)
531 * Then it is the load_weight's responsibility to consider overflow
535 u64 last_update_time, load_sum;
536 u32 util_sum, period_contrib;
537 unsigned long load_avg, util_avg;
540 #ifdef CONFIG_SCHEDSTATS
541 struct sched_statistics {
551 s64 sum_sleep_runtime;
558 u64 nr_migrations_cold;
559 u64 nr_failed_migrations_affine;
560 u64 nr_failed_migrations_running;
561 u64 nr_failed_migrations_hot;
562 u64 nr_forced_migrations;
566 u64 nr_wakeups_migrate;
567 u64 nr_wakeups_local;
568 u64 nr_wakeups_remote;
569 u64 nr_wakeups_affine;
570 u64 nr_wakeups_affine_attempts;
571 u64 nr_wakeups_passive;
576 struct sched_entity {
577 struct load_weight load; /* for load-balancing */
578 struct rb_node run_node;
579 struct list_head group_node;
583 u64 sum_exec_runtime;
585 u64 prev_sum_exec_runtime;
589 #ifdef CONFIG_SCHEDSTATS
590 struct sched_statistics statistics;
593 #ifdef CONFIG_FAIR_GROUP_SCHED
595 struct sched_entity *parent;
596 /* rq on which this entity is (to be) queued: */
597 struct cfs_rq *cfs_rq;
598 /* rq "owned" by this entity/group: */
604 * Per entity load average tracking.
606 * Put into separate cache line so it does not
607 * collide with read-mostly values above.
609 struct sched_avg avg ____cacheline_aligned_in_smp;
613 struct sched_rt_entity {
614 struct list_head run_list;
615 unsigned long timeout;
616 unsigned long watchdog_stamp;
617 unsigned int time_slice;
618 unsigned short on_rq;
619 unsigned short on_list;
621 struct sched_rt_entity *back;
622 #ifdef CONFIG_RT_GROUP_SCHED
623 struct sched_rt_entity *parent;
624 /* rq on which this entity is (to be) queued: */
626 /* rq "owned" by this entity/group: */
631 struct sched_dl_entity {
632 struct rb_node rb_node;
635 * Original scheduling parameters. Copied here from sched_attr
636 * during sched_setattr(), they will remain the same until
637 * the next sched_setattr().
639 u64 dl_runtime; /* maximum runtime for each instance */
640 u64 dl_deadline; /* relative deadline of each instance */
641 u64 dl_period; /* separation of two instances (period) */
642 u64 dl_bw; /* dl_runtime / dl_deadline */
645 * Actual scheduling parameters. Initialized with the values above,
646 * they are continously updated during task execution. Note that
647 * the remaining runtime could be < 0 in case we are in overrun.
649 s64 runtime; /* remaining runtime for this instance */
650 u64 deadline; /* absolute deadline for this instance */
651 unsigned int flags; /* specifying the scheduler behaviour */
656 * @dl_throttled tells if we exhausted the runtime. If so, the
657 * task has to wait for a replenishment to be performed at the
658 * next firing of dl_timer.
660 * @dl_boosted tells if we are boosted due to DI. If so we are
661 * outside bandwidth enforcement mechanism (but only until we
662 * exit the critical section);
664 * @dl_yielded tells if task gave up the cpu before consuming
665 * all its available runtime during the last job.
667 int dl_throttled, dl_boosted, dl_yielded;
670 * Bandwidth enforcement timer. Each -deadline task has its
671 * own bandwidth to be enforced, thus we need one timer per task.
673 struct hrtimer dl_timer;
681 u8 pad; /* Otherwise the compiler can store garbage here. */
683 u32 s; /* Set of bits. */
687 enum perf_event_task_context {
688 perf_invalid_context = -1,
691 perf_nr_task_contexts,
695 struct wake_q_node *next;
698 /* Track pages that require TLB flushes */
699 struct tlbflush_unmap_batch {
701 * Each bit set is a CPU that potentially has a TLB entry for one of
702 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
704 struct cpumask cpumask;
706 /* True if any bit in cpumask is set */
710 * If true then the PTE was dirty when unmapped. The entry must be
711 * flushed before IO is initiated or a stale TLB entry potentially
712 * allows an update without redirtying the page.
718 #ifdef CONFIG_THREAD_INFO_IN_TASK
720 * For reasons of header soup (see current_thread_info()), this
721 * must be the first element of task_struct.
723 struct thread_info thread_info;
725 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
728 unsigned int flags; /* per process flags, defined below */
732 struct llist_node wake_entry;
734 #ifdef CONFIG_THREAD_INFO_IN_TASK
735 unsigned int cpu; /* current CPU */
737 unsigned int wakee_flips;
738 unsigned long wakee_flip_decay_ts;
739 struct task_struct *last_wakee;
745 int prio, static_prio, normal_prio;
746 unsigned int rt_priority;
747 const struct sched_class *sched_class;
748 struct sched_entity se;
749 struct sched_rt_entity rt;
750 #ifdef CONFIG_CGROUP_SCHED
751 struct task_group *sched_task_group;
753 struct sched_dl_entity dl;
755 #ifdef CONFIG_PREEMPT_NOTIFIERS
756 /* list of struct preempt_notifier: */
757 struct hlist_head preempt_notifiers;
760 #ifdef CONFIG_BLK_DEV_IO_TRACE
761 unsigned int btrace_seq;
766 cpumask_t cpus_allowed;
768 #ifdef CONFIG_PREEMPT_RCU
769 int rcu_read_lock_nesting;
770 union rcu_special rcu_read_unlock_special;
771 struct list_head rcu_node_entry;
772 struct rcu_node *rcu_blocked_node;
773 #endif /* #ifdef CONFIG_PREEMPT_RCU */
774 #ifdef CONFIG_TASKS_RCU
775 unsigned long rcu_tasks_nvcsw;
776 bool rcu_tasks_holdout;
777 struct list_head rcu_tasks_holdout_list;
778 int rcu_tasks_idle_cpu;
779 #endif /* #ifdef CONFIG_TASKS_RCU */
781 #ifdef CONFIG_SCHED_INFO
782 struct sched_info sched_info;
785 struct list_head tasks;
787 struct plist_node pushable_tasks;
788 struct rb_node pushable_dl_tasks;
791 struct mm_struct *mm, *active_mm;
793 /* Per-thread vma caching: */
794 struct vmacache vmacache;
796 #if defined(SPLIT_RSS_COUNTING)
797 struct task_rss_stat rss_stat;
801 int exit_code, exit_signal;
802 int pdeath_signal; /* The signal sent when the parent dies */
803 unsigned long jobctl; /* JOBCTL_*, siglock protected */
805 /* Used for emulating ABI behavior of previous Linux versions */
806 unsigned int personality;
808 /* scheduler bits, serialized by scheduler locks */
809 unsigned sched_reset_on_fork:1;
810 unsigned sched_contributes_to_load:1;
811 unsigned sched_migrated:1;
812 unsigned sched_remote_wakeup:1;
813 unsigned :0; /* force alignment to the next boundary */
815 /* unserialized, strictly 'current' */
816 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
817 unsigned in_iowait:1;
818 #if !defined(TIF_RESTORE_SIGMASK)
819 unsigned restore_sigmask:1;
822 unsigned memcg_may_oom:1;
824 unsigned memcg_kmem_skip_account:1;
827 #ifdef CONFIG_COMPAT_BRK
828 unsigned brk_randomized:1;
831 unsigned long atomic_flags; /* Flags needing atomic access. */
833 struct restart_block restart_block;
838 #ifdef CONFIG_CC_STACKPROTECTOR
839 /* Canary value for the -fstack-protector gcc feature */
840 unsigned long stack_canary;
843 * pointers to (original) parent process, youngest child, younger sibling,
844 * older sibling, respectively. (p->father can be replaced with
845 * p->real_parent->pid)
847 struct task_struct __rcu *real_parent; /* real parent process */
848 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
850 * children/sibling forms the list of my natural children
852 struct list_head children; /* list of my children */
853 struct list_head sibling; /* linkage in my parent's children list */
854 struct task_struct *group_leader; /* threadgroup leader */
857 * ptraced is the list of tasks this task is using ptrace on.
858 * This includes both natural children and PTRACE_ATTACH targets.
859 * p->ptrace_entry is p's link on the p->parent->ptraced list.
861 struct list_head ptraced;
862 struct list_head ptrace_entry;
864 /* PID/PID hash table linkage. */
865 struct pid_link pids[PIDTYPE_MAX];
866 struct list_head thread_group;
867 struct list_head thread_node;
869 struct completion *vfork_done; /* for vfork() */
870 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
871 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
874 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
875 u64 utimescaled, stimescaled;
878 struct prev_cputime prev_cputime;
879 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
880 seqcount_t vtime_seqcount;
881 unsigned long long vtime_snap;
883 /* Task is sleeping or running in a CPU with VTIME inactive */
885 /* Task runs in userspace in a CPU with VTIME active */
887 /* Task runs in kernelspace in a CPU with VTIME active */
892 #ifdef CONFIG_NO_HZ_FULL
893 atomic_t tick_dep_mask;
895 unsigned long nvcsw, nivcsw; /* context switch counts */
896 u64 start_time; /* monotonic time in nsec */
897 u64 real_start_time; /* boot based time in nsec */
898 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
899 unsigned long min_flt, maj_flt;
901 #ifdef CONFIG_POSIX_TIMERS
902 struct task_cputime cputime_expires;
903 struct list_head cpu_timers[3];
906 /* process credentials */
907 const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
908 const struct cred __rcu *real_cred; /* objective and real subjective task
909 * credentials (COW) */
910 const struct cred __rcu *cred; /* effective (overridable) subjective task
911 * credentials (COW) */
912 char comm[TASK_COMM_LEN]; /* executable name excluding path
913 - access with [gs]et_task_comm (which lock
915 - initialized normally by setup_new_exec */
916 /* file system info */
917 struct nameidata *nameidata;
918 #ifdef CONFIG_SYSVIPC
920 struct sysv_sem sysvsem;
921 struct sysv_shm sysvshm;
923 #ifdef CONFIG_DETECT_HUNG_TASK
924 /* hung task detection */
925 unsigned long last_switch_count;
927 /* filesystem information */
928 struct fs_struct *fs;
929 /* open file information */
930 struct files_struct *files;
932 struct nsproxy *nsproxy;
933 /* signal handlers */
934 struct signal_struct *signal;
935 struct sighand_struct *sighand;
937 sigset_t blocked, real_blocked;
938 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
939 struct sigpending pending;
941 unsigned long sas_ss_sp;
943 unsigned sas_ss_flags;
945 struct callback_head *task_works;
947 struct audit_context *audit_context;
948 #ifdef CONFIG_AUDITSYSCALL
950 unsigned int sessionid;
952 struct seccomp seccomp;
954 /* Thread group tracking */
957 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
959 spinlock_t alloc_lock;
961 /* Protection of the PI data structures: */
962 raw_spinlock_t pi_lock;
964 struct wake_q_node wake_q;
966 #ifdef CONFIG_RT_MUTEXES
967 /* PI waiters blocked on a rt_mutex held by this task */
968 struct rb_root pi_waiters;
969 struct rb_node *pi_waiters_leftmost;
970 /* Deadlock detection and priority inheritance handling */
971 struct rt_mutex_waiter *pi_blocked_on;
974 #ifdef CONFIG_DEBUG_MUTEXES
975 /* mutex deadlock detection */
976 struct mutex_waiter *blocked_on;
978 #ifdef CONFIG_TRACE_IRQFLAGS
979 unsigned int irq_events;
980 unsigned long hardirq_enable_ip;
981 unsigned long hardirq_disable_ip;
982 unsigned int hardirq_enable_event;
983 unsigned int hardirq_disable_event;
984 int hardirqs_enabled;
986 unsigned long softirq_disable_ip;
987 unsigned long softirq_enable_ip;
988 unsigned int softirq_disable_event;
989 unsigned int softirq_enable_event;
990 int softirqs_enabled;
993 #ifdef CONFIG_LOCKDEP
994 # define MAX_LOCK_DEPTH 48UL
997 unsigned int lockdep_recursion;
998 struct held_lock held_locks[MAX_LOCK_DEPTH];
999 gfp_t lockdep_reclaim_gfp;
1002 unsigned int in_ubsan;
1005 /* journalling filesystem info */
1008 /* stacked block device info */
1009 struct bio_list *bio_list;
1012 /* stack plugging */
1013 struct blk_plug *plug;
1017 struct reclaim_state *reclaim_state;
1019 struct backing_dev_info *backing_dev_info;
1021 struct io_context *io_context;
1023 unsigned long ptrace_message;
1024 siginfo_t *last_siginfo; /* For ptrace use. */
1025 struct task_io_accounting ioac;
1026 #if defined(CONFIG_TASK_XACCT)
1027 u64 acct_rss_mem1; /* accumulated rss usage */
1028 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1029 u64 acct_timexpd; /* stime + utime since last update */
1031 #ifdef CONFIG_CPUSETS
1032 nodemask_t mems_allowed; /* Protected by alloc_lock */
1033 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1034 int cpuset_mem_spread_rotor;
1035 int cpuset_slab_spread_rotor;
1037 #ifdef CONFIG_CGROUPS
1038 /* Control Group info protected by css_set_lock */
1039 struct css_set __rcu *cgroups;
1040 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1041 struct list_head cg_list;
1043 #ifdef CONFIG_INTEL_RDT_A
1047 struct robust_list_head __user *robust_list;
1048 #ifdef CONFIG_COMPAT
1049 struct compat_robust_list_head __user *compat_robust_list;
1051 struct list_head pi_state_list;
1052 struct futex_pi_state *pi_state_cache;
1054 #ifdef CONFIG_PERF_EVENTS
1055 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1056 struct mutex perf_event_mutex;
1057 struct list_head perf_event_list;
1059 #ifdef CONFIG_DEBUG_PREEMPT
1060 unsigned long preempt_disable_ip;
1063 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1065 short pref_node_fork;
1067 #ifdef CONFIG_NUMA_BALANCING
1069 unsigned int numa_scan_period;
1070 unsigned int numa_scan_period_max;
1071 int numa_preferred_nid;
1072 unsigned long numa_migrate_retry;
1073 u64 node_stamp; /* migration stamp */
1074 u64 last_task_numa_placement;
1075 u64 last_sum_exec_runtime;
1076 struct callback_head numa_work;
1078 struct list_head numa_entry;
1079 struct numa_group *numa_group;
1082 * numa_faults is an array split into four regions:
1083 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1084 * in this precise order.
1086 * faults_memory: Exponential decaying average of faults on a per-node
1087 * basis. Scheduling placement decisions are made based on these
1088 * counts. The values remain static for the duration of a PTE scan.
1089 * faults_cpu: Track the nodes the process was running on when a NUMA
1090 * hinting fault was incurred.
1091 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1092 * during the current scan window. When the scan completes, the counts
1093 * in faults_memory and faults_cpu decay and these values are copied.
1095 unsigned long *numa_faults;
1096 unsigned long total_numa_faults;
1099 * numa_faults_locality tracks if faults recorded during the last
1100 * scan window were remote/local or failed to migrate. The task scan
1101 * period is adapted based on the locality of the faults with different
1102 * weights depending on whether they were shared or private faults
1104 unsigned long numa_faults_locality[3];
1106 unsigned long numa_pages_migrated;
1107 #endif /* CONFIG_NUMA_BALANCING */
1109 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1110 struct tlbflush_unmap_batch tlb_ubc;
1113 struct rcu_head rcu;
1116 * cache last used pipe for splice
1118 struct pipe_inode_info *splice_pipe;
1120 struct page_frag task_frag;
1122 #ifdef CONFIG_TASK_DELAY_ACCT
1123 struct task_delay_info *delays;
1126 #ifdef CONFIG_FAULT_INJECTION
1130 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1131 * balance_dirty_pages() for some dirty throttling pause
1134 int nr_dirtied_pause;
1135 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1137 #ifdef CONFIG_LATENCYTOP
1138 int latency_record_count;
1139 struct latency_record latency_record[LT_SAVECOUNT];
1142 * time slack values; these are used to round up poll() and
1143 * select() etc timeout values. These are in nanoseconds.
1146 u64 default_timer_slack_ns;
1149 unsigned int kasan_depth;
1151 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1152 /* Index of current stored address in ret_stack */
1154 /* Stack of return addresses for return function tracing */
1155 struct ftrace_ret_stack *ret_stack;
1156 /* time stamp for last schedule */
1157 unsigned long long ftrace_timestamp;
1159 * Number of functions that haven't been traced
1160 * because of depth overrun.
1162 atomic_t trace_overrun;
1163 /* Pause for the tracing */
1164 atomic_t tracing_graph_pause;
1166 #ifdef CONFIG_TRACING
1167 /* state flags for use by tracers */
1168 unsigned long trace;
1169 /* bitmask and counter of trace recursion */
1170 unsigned long trace_recursion;
1171 #endif /* CONFIG_TRACING */
1173 /* Coverage collection mode enabled for this task (0 if disabled). */
1174 enum kcov_mode kcov_mode;
1175 /* Size of the kcov_area. */
1177 /* Buffer for coverage collection. */
1179 /* kcov desciptor wired with this task or NULL. */
1183 struct mem_cgroup *memcg_in_oom;
1184 gfp_t memcg_oom_gfp_mask;
1185 int memcg_oom_order;
1187 /* number of pages to reclaim on returning to userland */
1188 unsigned int memcg_nr_pages_over_high;
1190 #ifdef CONFIG_UPROBES
1191 struct uprobe_task *utask;
1193 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1194 unsigned int sequential_io;
1195 unsigned int sequential_io_avg;
1197 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1198 unsigned long task_state_change;
1200 int pagefault_disabled;
1202 struct task_struct *oom_reaper_list;
1204 #ifdef CONFIG_VMAP_STACK
1205 struct vm_struct *stack_vm_area;
1207 #ifdef CONFIG_THREAD_INFO_IN_TASK
1208 /* A live task holds one reference. */
1209 atomic_t stack_refcount;
1211 /* CPU-specific state of this task */
1212 struct thread_struct thread;
1214 * WARNING: on x86, 'thread_struct' contains a variable-sized
1215 * structure. It *MUST* be at the end of 'task_struct'.
1217 * Do not put anything below here!
1221 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1222 extern int arch_task_struct_size __read_mostly;
1224 # define arch_task_struct_size (sizeof(struct task_struct))
1227 #ifdef CONFIG_VMAP_STACK
1228 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
1230 return t->stack_vm_area;
1233 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
1239 #define TNF_MIGRATED 0x01
1240 #define TNF_NO_GROUP 0x02
1241 #define TNF_SHARED 0x04
1242 #define TNF_FAULT_LOCAL 0x08
1243 #define TNF_MIGRATE_FAIL 0x10
1245 static inline bool in_vfork(struct task_struct *tsk)
1250 * need RCU to access ->real_parent if CLONE_VM was used along with
1253 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
1256 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
1257 * ->real_parent is not necessarily the task doing vfork(), so in
1258 * theory we can't rely on task_lock() if we want to dereference it.
1260 * And in this case we can't trust the real_parent->mm == tsk->mm
1261 * check, it can be false negative. But we do not care, if init or
1262 * another oom-unkillable task does this it should blame itself.
1265 ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
1271 #ifdef CONFIG_NUMA_BALANCING
1272 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1273 extern pid_t task_numa_group_id(struct task_struct *p);
1274 extern void set_numabalancing_state(bool enabled);
1275 extern void task_numa_free(struct task_struct *p);
1276 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1277 int src_nid, int dst_cpu);
1279 static inline void task_numa_fault(int last_node, int node, int pages,
1283 static inline pid_t task_numa_group_id(struct task_struct *p)
1287 static inline void set_numabalancing_state(bool enabled)
1290 static inline void task_numa_free(struct task_struct *p)
1293 static inline bool should_numa_migrate_memory(struct task_struct *p,
1294 struct page *page, int src_nid, int dst_cpu)
1300 static inline struct pid *task_pid(struct task_struct *task)
1302 return task->pids[PIDTYPE_PID].pid;
1305 static inline struct pid *task_tgid(struct task_struct *task)
1307 return task->group_leader->pids[PIDTYPE_PID].pid;
1311 * Without tasklist or rcu lock it is not safe to dereference
1312 * the result of task_pgrp/task_session even if task == current,
1313 * we can race with another thread doing sys_setsid/sys_setpgid.
1315 static inline struct pid *task_pgrp(struct task_struct *task)
1317 return task->group_leader->pids[PIDTYPE_PGID].pid;
1320 static inline struct pid *task_session(struct task_struct *task)
1322 return task->group_leader->pids[PIDTYPE_SID].pid;
1325 struct pid_namespace;
1328 * the helpers to get the task's different pids as they are seen
1329 * from various namespaces
1331 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1332 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1334 * task_xid_nr_ns() : id seen from the ns specified;
1336 * set_task_vxid() : assigns a virtual id to a task;
1338 * see also pid_nr() etc in include/linux/pid.h
1340 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1341 struct pid_namespace *ns);
1343 static inline pid_t task_pid_nr(struct task_struct *tsk)
1348 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1349 struct pid_namespace *ns)
1351 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1354 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1356 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1360 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1365 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1367 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1369 return pid_vnr(task_tgid(tsk));
1373 static inline int pid_alive(const struct task_struct *p);
1374 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1380 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1386 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1388 return task_ppid_nr_ns(tsk, &init_pid_ns);
1391 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1392 struct pid_namespace *ns)
1394 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1397 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1399 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1403 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1404 struct pid_namespace *ns)
1406 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1409 static inline pid_t task_session_vnr(struct task_struct *tsk)
1411 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1414 /* obsolete, do not use */
1415 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1417 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1421 * pid_alive - check that a task structure is not stale
1422 * @p: Task structure to be checked.
1424 * Test if a process is not yet dead (at most zombie state)
1425 * If pid_alive fails, then pointers within the task structure
1426 * can be stale and must not be dereferenced.
1428 * Return: 1 if the process is alive. 0 otherwise.
1430 static inline int pid_alive(const struct task_struct *p)
1432 return p->pids[PIDTYPE_PID].pid != NULL;
1436 * is_global_init - check if a task structure is init. Since init
1437 * is free to have sub-threads we need to check tgid.
1438 * @tsk: Task structure to be checked.
1440 * Check if a task structure is the first user space task the kernel created.
1442 * Return: 1 if the task structure is init. 0 otherwise.
1444 static inline int is_global_init(struct task_struct *tsk)
1446 return task_tgid_nr(tsk) == 1;
1449 extern struct pid *cad_pid;
1451 extern void free_task(struct task_struct *tsk);
1452 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1454 extern void __put_task_struct(struct task_struct *t);
1456 static inline void put_task_struct(struct task_struct *t)
1458 if (atomic_dec_and_test(&t->usage))
1459 __put_task_struct(t);
1462 struct task_struct *task_rcu_dereference(struct task_struct **ptask);
1463 struct task_struct *try_get_task_struct(struct task_struct **ptask);
1465 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1466 extern void task_cputime(struct task_struct *t,
1467 u64 *utime, u64 *stime);
1468 extern u64 task_gtime(struct task_struct *t);
1470 static inline void task_cputime(struct task_struct *t,
1471 u64 *utime, u64 *stime)
1477 static inline u64 task_gtime(struct task_struct *t)
1483 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
1484 static inline void task_cputime_scaled(struct task_struct *t,
1488 *utimescaled = t->utimescaled;
1489 *stimescaled = t->stimescaled;
1492 static inline void task_cputime_scaled(struct task_struct *t,
1496 task_cputime(t, utimescaled, stimescaled);
1500 extern void task_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
1501 extern void thread_group_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
1506 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1507 #define PF_EXITING 0x00000004 /* getting shut down */
1508 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1509 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1510 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1511 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1512 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1513 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1514 #define PF_DUMPCORE 0x00000200 /* dumped core */
1515 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1516 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1517 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1518 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1519 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1520 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1521 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1522 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1523 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1524 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1525 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1526 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1527 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1528 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1529 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1530 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1531 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1532 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1533 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1536 * Only the _current_ task can read/write to tsk->flags, but other
1537 * tasks can access tsk->flags in readonly mode for example
1538 * with tsk_used_math (like during threaded core dumping).
1539 * There is however an exception to this rule during ptrace
1540 * or during fork: the ptracer task is allowed to write to the
1541 * child->flags of its traced child (same goes for fork, the parent
1542 * can write to the child->flags), because we're guaranteed the
1543 * child is not running and in turn not changing child->flags
1544 * at the same time the parent does it.
1546 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1547 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1548 #define clear_used_math() clear_stopped_child_used_math(current)
1549 #define set_used_math() set_stopped_child_used_math(current)
1550 #define conditional_stopped_child_used_math(condition, child) \
1551 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1552 #define conditional_used_math(condition) \
1553 conditional_stopped_child_used_math(condition, current)
1554 #define copy_to_stopped_child_used_math(child) \
1555 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1556 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1557 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1558 #define used_math() tsk_used_math(current)
1560 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1561 * __GFP_FS is also cleared as it implies __GFP_IO.
1563 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1565 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1566 flags &= ~(__GFP_IO | __GFP_FS);
1570 static inline unsigned int memalloc_noio_save(void)
1572 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1573 current->flags |= PF_MEMALLOC_NOIO;
1577 static inline void memalloc_noio_restore(unsigned int flags)
1579 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1582 /* Per-process atomic flags. */
1583 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1584 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1585 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1586 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
1589 #define TASK_PFA_TEST(name, func) \
1590 static inline bool task_##func(struct task_struct *p) \
1591 { return test_bit(PFA_##name, &p->atomic_flags); }
1592 #define TASK_PFA_SET(name, func) \
1593 static inline void task_set_##func(struct task_struct *p) \
1594 { set_bit(PFA_##name, &p->atomic_flags); }
1595 #define TASK_PFA_CLEAR(name, func) \
1596 static inline void task_clear_##func(struct task_struct *p) \
1597 { clear_bit(PFA_##name, &p->atomic_flags); }
1599 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1600 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1602 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1603 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1604 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1606 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1607 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1608 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1610 TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
1611 TASK_PFA_SET(LMK_WAITING, lmk_waiting)
1614 * task->jobctl flags
1616 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1618 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1619 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1620 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1621 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1622 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1623 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1624 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1626 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
1627 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
1628 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
1629 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
1630 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
1631 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
1632 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
1634 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1635 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1637 extern bool task_set_jobctl_pending(struct task_struct *task,
1638 unsigned long mask);
1639 extern void task_clear_jobctl_trapping(struct task_struct *task);
1640 extern void task_clear_jobctl_pending(struct task_struct *task,
1641 unsigned long mask);
1643 static inline void rcu_copy_process(struct task_struct *p)
1645 #ifdef CONFIG_PREEMPT_RCU
1646 p->rcu_read_lock_nesting = 0;
1647 p->rcu_read_unlock_special.s = 0;
1648 p->rcu_blocked_node = NULL;
1649 INIT_LIST_HEAD(&p->rcu_node_entry);
1650 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1651 #ifdef CONFIG_TASKS_RCU
1652 p->rcu_tasks_holdout = false;
1653 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
1654 p->rcu_tasks_idle_cpu = -1;
1655 #endif /* #ifdef CONFIG_TASKS_RCU */
1658 static inline void tsk_restore_flags(struct task_struct *task,
1659 unsigned long orig_flags, unsigned long flags)
1661 task->flags &= ~flags;
1662 task->flags |= orig_flags & flags;
1665 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
1666 const struct cpumask *trial);
1667 extern int task_can_attach(struct task_struct *p,
1668 const struct cpumask *cs_cpus_allowed);
1670 extern void do_set_cpus_allowed(struct task_struct *p,
1671 const struct cpumask *new_mask);
1673 extern int set_cpus_allowed_ptr(struct task_struct *p,
1674 const struct cpumask *new_mask);
1676 static inline void do_set_cpus_allowed(struct task_struct *p,
1677 const struct cpumask *new_mask)
1680 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1681 const struct cpumask *new_mask)
1683 if (!cpumask_test_cpu(0, new_mask))
1689 #ifdef CONFIG_NO_HZ_COMMON
1690 void calc_load_enter_idle(void);
1691 void calc_load_exit_idle(void);
1693 static inline void calc_load_enter_idle(void) { }
1694 static inline void calc_load_exit_idle(void) { }
1695 #endif /* CONFIG_NO_HZ_COMMON */
1697 #ifndef cpu_relax_yield
1698 #define cpu_relax_yield() cpu_relax()
1701 extern unsigned long long
1702 task_sched_runtime(struct task_struct *task);
1704 /* sched_exec is called by processes performing an exec */
1706 extern void sched_exec(void);
1708 #define sched_exec() {}
1711 #ifdef CONFIG_HOTPLUG_CPU
1712 extern void idle_task_exit(void);
1714 static inline void idle_task_exit(void) {}
1717 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1718 extern void wake_up_nohz_cpu(int cpu);
1720 static inline void wake_up_nohz_cpu(int cpu) { }
1723 #ifdef CONFIG_NO_HZ_FULL
1724 extern u64 scheduler_tick_max_deferment(void);
1727 extern int yield_to(struct task_struct *p, bool preempt);
1728 extern void set_user_nice(struct task_struct *p, long nice);
1729 extern int task_prio(const struct task_struct *p);
1731 * task_nice - return the nice value of a given task.
1732 * @p: the task in question.
1734 * Return: The nice value [ -20 ... 0 ... 19 ].
1736 static inline int task_nice(const struct task_struct *p)
1738 return PRIO_TO_NICE((p)->static_prio);
1740 extern int can_nice(const struct task_struct *p, const int nice);
1741 extern int task_curr(const struct task_struct *p);
1742 extern int idle_cpu(int cpu);
1743 extern int sched_setscheduler(struct task_struct *, int,
1744 const struct sched_param *);
1745 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1746 const struct sched_param *);
1747 extern int sched_setattr(struct task_struct *,
1748 const struct sched_attr *);
1749 extern struct task_struct *idle_task(int cpu);
1751 * is_idle_task - is the specified task an idle task?
1752 * @p: the task in question.
1754 * Return: 1 if @p is an idle task. 0 otherwise.
1756 static inline bool is_idle_task(const struct task_struct *p)
1758 return !!(p->flags & PF_IDLE);
1760 extern struct task_struct *curr_task(int cpu);
1761 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1765 union thread_union {
1766 #ifndef CONFIG_THREAD_INFO_IN_TASK
1767 struct thread_info thread_info;
1769 unsigned long stack[THREAD_SIZE/sizeof(long)];
1772 #ifndef __HAVE_ARCH_KSTACK_END
1773 static inline int kstack_end(void *addr)
1775 /* Reliable end of stack detection:
1776 * Some APM bios versions misalign the stack
1778 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1782 extern union thread_union init_thread_union;
1783 extern struct task_struct init_task;
1785 extern struct mm_struct init_mm;
1787 extern struct pid_namespace init_pid_ns;
1790 * find a task by one of its numerical ids
1792 * find_task_by_pid_ns():
1793 * finds a task by its pid in the specified namespace
1794 * find_task_by_vpid():
1795 * finds a task by its virtual pid
1797 * see also find_vpid() etc in include/linux/pid.h
1800 extern struct task_struct *find_task_by_vpid(pid_t nr);
1801 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1802 struct pid_namespace *ns);
1804 #include <asm/current.h>
1806 extern void xtime_update(unsigned long ticks);
1808 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1809 extern int wake_up_process(struct task_struct *tsk);
1810 extern void wake_up_new_task(struct task_struct *tsk);
1812 extern void kick_process(struct task_struct *tsk);
1814 static inline void kick_process(struct task_struct *tsk) { }
1816 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
1817 extern void sched_dead(struct task_struct *p);
1819 extern void proc_caches_init(void);
1821 extern void release_task(struct task_struct * p);
1823 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
1824 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
1825 struct task_struct *, unsigned long);
1827 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1828 struct task_struct *);
1830 /* Architectures that haven't opted into copy_thread_tls get the tls argument
1831 * via pt_regs, so ignore the tls argument passed via C. */
1832 static inline int copy_thread_tls(
1833 unsigned long clone_flags, unsigned long sp, unsigned long arg,
1834 struct task_struct *p, unsigned long tls)
1836 return copy_thread(clone_flags, sp, arg, p);
1839 extern void flush_thread(void);
1841 #ifdef CONFIG_HAVE_EXIT_THREAD
1842 extern void exit_thread(struct task_struct *tsk);
1844 static inline void exit_thread(struct task_struct *tsk)
1849 extern void exit_files(struct task_struct *);
1851 extern void exit_itimers(struct signal_struct *);
1853 extern void do_group_exit(int);
1855 extern int do_execve(struct filename *,
1856 const char __user * const __user *,
1857 const char __user * const __user *);
1858 extern int do_execveat(int, struct filename *,
1859 const char __user * const __user *,
1860 const char __user * const __user *,
1862 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
1863 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
1864 struct task_struct *fork_idle(int);
1865 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
1867 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1868 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1870 __set_task_comm(tsk, from, false);
1872 extern char *get_task_comm(char *to, struct task_struct *tsk);
1875 void scheduler_ipi(void);
1876 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1878 static inline void scheduler_ipi(void) { }
1879 static inline unsigned long wait_task_inactive(struct task_struct *p,
1887 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1888 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1889 * pins the final release of task.io_context. Also protects ->cpuset and
1890 * ->cgroup.subsys[]. And ->vfork_done.
1892 * Nests both inside and outside of read_lock(&tasklist_lock).
1893 * It must not be nested with write_lock_irq(&tasklist_lock),
1894 * neither inside nor outside.
1896 static inline void task_lock(struct task_struct *p)
1898 spin_lock(&p->alloc_lock);
1901 static inline void task_unlock(struct task_struct *p)
1903 spin_unlock(&p->alloc_lock);
1906 #ifdef CONFIG_THREAD_INFO_IN_TASK
1908 static inline struct thread_info *task_thread_info(struct task_struct *task)
1910 return &task->thread_info;
1914 * When accessing the stack of a non-current task that might exit, use
1915 * try_get_task_stack() instead. task_stack_page will return a pointer
1916 * that could get freed out from under you.
1918 static inline void *task_stack_page(const struct task_struct *task)
1923 #define setup_thread_stack(new,old) do { } while(0)
1925 static inline unsigned long *end_of_stack(const struct task_struct *task)
1930 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1932 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1933 #define task_stack_page(task) ((void *)(task)->stack)
1935 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1937 *task_thread_info(p) = *task_thread_info(org);
1938 task_thread_info(p)->task = p;
1942 * Return the address of the last usable long on the stack.
1944 * When the stack grows down, this is just above the thread
1945 * info struct. Going any lower will corrupt the threadinfo.
1947 * When the stack grows up, this is the highest address.
1948 * Beyond that position, we corrupt data on the next page.
1950 static inline unsigned long *end_of_stack(struct task_struct *p)
1952 #ifdef CONFIG_STACK_GROWSUP
1953 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
1955 return (unsigned long *)(task_thread_info(p) + 1);
1961 #ifdef CONFIG_THREAD_INFO_IN_TASK
1962 static inline void *try_get_task_stack(struct task_struct *tsk)
1964 return atomic_inc_not_zero(&tsk->stack_refcount) ?
1965 task_stack_page(tsk) : NULL;
1968 extern void put_task_stack(struct task_struct *tsk);
1970 static inline void *try_get_task_stack(struct task_struct *tsk)
1972 return task_stack_page(tsk);
1975 static inline void put_task_stack(struct task_struct *tsk) {}
1978 #define task_stack_end_corrupted(task) \
1979 (*(end_of_stack(task)) != STACK_END_MAGIC)
1981 static inline int object_is_on_stack(void *obj)
1983 void *stack = task_stack_page(current);
1985 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
1988 extern void thread_stack_cache_init(void);
1990 #ifdef CONFIG_DEBUG_STACK_USAGE
1991 static inline unsigned long stack_not_used(struct task_struct *p)
1993 unsigned long *n = end_of_stack(p);
1995 do { /* Skip over canary */
1996 # ifdef CONFIG_STACK_GROWSUP
2003 # ifdef CONFIG_STACK_GROWSUP
2004 return (unsigned long)end_of_stack(p) - (unsigned long)n;
2006 return (unsigned long)n - (unsigned long)end_of_stack(p);
2010 extern void set_task_stack_end_magic(struct task_struct *tsk);
2012 /* set thread flags in other task's structures
2013 * - see asm/thread_info.h for TIF_xxxx flags available
2015 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2017 set_ti_thread_flag(task_thread_info(tsk), flag);
2020 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2022 clear_ti_thread_flag(task_thread_info(tsk), flag);
2025 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2027 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2030 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2032 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2035 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2037 return test_ti_thread_flag(task_thread_info(tsk), flag);
2040 static inline void set_tsk_need_resched(struct task_struct *tsk)
2042 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2045 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2047 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2050 static inline int test_tsk_need_resched(struct task_struct *tsk)
2052 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2055 static inline int restart_syscall(void)
2057 set_tsk_thread_flag(current, TIF_SIGPENDING);
2058 return -ERESTARTNOINTR;
2061 static inline int signal_pending(struct task_struct *p)
2063 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2066 static inline int __fatal_signal_pending(struct task_struct *p)
2068 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2071 static inline int fatal_signal_pending(struct task_struct *p)
2073 return signal_pending(p) && __fatal_signal_pending(p);
2076 static inline int signal_pending_state(long state, struct task_struct *p)
2078 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2080 if (!signal_pending(p))
2083 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2087 * cond_resched() and cond_resched_lock(): latency reduction via
2088 * explicit rescheduling in places that are safe. The return
2089 * value indicates whether a reschedule was done in fact.
2090 * cond_resched_lock() will drop the spinlock before scheduling,
2091 * cond_resched_softirq() will enable bhs before scheduling.
2093 #ifndef CONFIG_PREEMPT
2094 extern int _cond_resched(void);
2096 static inline int _cond_resched(void) { return 0; }
2099 #define cond_resched() ({ \
2100 ___might_sleep(__FILE__, __LINE__, 0); \
2104 extern int __cond_resched_lock(spinlock_t *lock);
2106 #define cond_resched_lock(lock) ({ \
2107 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2108 __cond_resched_lock(lock); \
2111 extern int __cond_resched_softirq(void);
2113 #define cond_resched_softirq() ({ \
2114 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2115 __cond_resched_softirq(); \
2118 static inline void cond_resched_rcu(void)
2120 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2128 * Does a critical section need to be broken due to another
2129 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2130 * but a general need for low latency)
2132 static inline int spin_needbreak(spinlock_t *lock)
2134 #ifdef CONFIG_PREEMPT
2135 return spin_is_contended(lock);
2141 static __always_inline bool need_resched(void)
2143 return unlikely(tif_need_resched());
2147 * Thread group CPU time accounting.
2149 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2150 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2153 * Reevaluate whether the task has signals pending delivery.
2154 * Wake the task if so.
2155 * This is required every time the blocked sigset_t changes.
2156 * callers must hold sighand->siglock.
2158 extern void recalc_sigpending_and_wake(struct task_struct *t);
2159 extern void recalc_sigpending(void);
2161 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2163 static inline void signal_wake_up(struct task_struct *t, bool resume)
2165 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2167 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2169 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2173 * Wrappers for p->thread_info->cpu access. No-op on UP.
2177 static inline unsigned int task_cpu(const struct task_struct *p)
2179 #ifdef CONFIG_THREAD_INFO_IN_TASK
2182 return task_thread_info(p)->cpu;
2186 static inline int task_node(const struct task_struct *p)
2188 return cpu_to_node(task_cpu(p));
2191 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2195 static inline unsigned int task_cpu(const struct task_struct *p)
2200 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2204 #endif /* CONFIG_SMP */
2207 * In order to reduce various lock holder preemption latencies provide an
2208 * interface to see if a vCPU is currently running or not.
2210 * This allows us to terminate optimistic spin loops and block, analogous to
2211 * the native optimistic spin heuristic of testing if the lock owner task is
2214 #ifndef vcpu_is_preempted
2215 # define vcpu_is_preempted(cpu) false
2218 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2219 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2221 #ifdef CONFIG_CGROUP_SCHED
2222 extern struct task_group root_task_group;
2223 #endif /* CONFIG_CGROUP_SCHED */
2225 extern int task_can_switch_user(struct user_struct *up,
2226 struct task_struct *tsk);
2228 #ifdef CONFIG_TASK_XACCT
2229 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2231 tsk->ioac.rchar += amt;
2234 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2236 tsk->ioac.wchar += amt;
2239 static inline void inc_syscr(struct task_struct *tsk)
2244 static inline void inc_syscw(struct task_struct *tsk)
2249 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2253 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2257 static inline void inc_syscr(struct task_struct *tsk)
2261 static inline void inc_syscw(struct task_struct *tsk)
2266 #ifndef TASK_SIZE_OF
2267 #define TASK_SIZE_OF(tsk) TASK_SIZE