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 extern void arch_pick_mmap_layout(struct mm_struct *mm);
319 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
320 unsigned long, unsigned long);
322 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
323 unsigned long len, unsigned long pgoff,
324 unsigned long flags);
326 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
329 struct pacct_struct {
332 unsigned long ac_mem;
333 u64 ac_utime, ac_stime;
334 unsigned long ac_minflt, ac_majflt;
343 * struct prev_cputime - snaphsot of system and user cputime
344 * @utime: time spent in user mode
345 * @stime: time spent in system mode
346 * @lock: protects the above two fields
348 * Stores previous user/system time values such that we can guarantee
351 struct prev_cputime {
352 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
359 static inline void prev_cputime_init(struct prev_cputime *prev)
361 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
362 prev->utime = prev->stime = 0;
363 raw_spin_lock_init(&prev->lock);
368 * struct task_cputime - collected CPU time counts
369 * @utime: time spent in user mode, in nanoseconds
370 * @stime: time spent in kernel mode, in nanoseconds
371 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
373 * This structure groups together three kinds of CPU time that are tracked for
374 * threads and thread groups. Most things considering CPU time want to group
375 * these counts together and treat all three of them in parallel.
377 struct task_cputime {
380 unsigned long long sum_exec_runtime;
383 /* Alternate field names when used to cache expirations. */
384 #define virt_exp utime
385 #define prof_exp stime
386 #define sched_exp sum_exec_runtime
389 * This is the atomic variant of task_cputime, which can be used for
390 * storing and updating task_cputime statistics without locking.
392 struct task_cputime_atomic {
395 atomic64_t sum_exec_runtime;
398 #define INIT_CPUTIME_ATOMIC \
399 (struct task_cputime_atomic) { \
400 .utime = ATOMIC64_INIT(0), \
401 .stime = ATOMIC64_INIT(0), \
402 .sum_exec_runtime = ATOMIC64_INIT(0), \
405 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
408 * Disable preemption until the scheduler is running -- use an unconditional
409 * value so that it also works on !PREEMPT_COUNT kernels.
411 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
413 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
416 * Initial preempt_count value; reflects the preempt_count schedule invariant
417 * which states that during context switches:
419 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
421 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
422 * Note: See finish_task_switch().
424 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
427 * struct thread_group_cputimer - thread group interval timer counts
428 * @cputime_atomic: atomic thread group interval timers.
429 * @running: true when there are timers running and
430 * @cputime_atomic receives updates.
431 * @checking_timer: true when a thread in the group is in the
432 * process of checking for thread group timers.
434 * This structure contains the version of task_cputime, above, that is
435 * used for thread group CPU timer calculations.
437 struct thread_group_cputimer {
438 struct task_cputime_atomic cputime_atomic;
443 #include <linux/rwsem.h>
446 struct backing_dev_info;
447 struct reclaim_state;
449 #ifdef CONFIG_SCHED_INFO
451 /* cumulative counters */
452 unsigned long pcount; /* # of times run on this cpu */
453 unsigned long long run_delay; /* time spent waiting on a runqueue */
456 unsigned long long last_arrival,/* when we last ran on a cpu */
457 last_queued; /* when we were last queued to run */
459 #endif /* CONFIG_SCHED_INFO */
461 struct task_delay_info;
463 static inline int sched_info_on(void)
465 #ifdef CONFIG_SCHEDSTATS
467 #elif defined(CONFIG_TASK_DELAY_ACCT)
468 extern int delayacct_on;
475 #ifdef CONFIG_SCHEDSTATS
476 void force_schedstat_enabled(void);
480 * Integer metrics need fixed point arithmetic, e.g., sched/fair
481 * has a few: load, load_avg, util_avg, freq, and capacity.
483 * We define a basic fixed point arithmetic range, and then formalize
484 * all these metrics based on that basic range.
486 # define SCHED_FIXEDPOINT_SHIFT 10
487 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
489 struct io_context; /* See blkdev.h */
492 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
493 extern void prefetch_stack(struct task_struct *t);
495 static inline void prefetch_stack(struct task_struct *t) { }
498 struct audit_context; /* See audit.c */
500 struct pipe_inode_info;
501 struct uts_namespace;
504 unsigned long weight;
509 * The load_avg/util_avg accumulates an infinite geometric series
510 * (see __update_load_avg() in kernel/sched/fair.c).
512 * [load_avg definition]
514 * load_avg = runnable% * scale_load_down(load)
516 * where runnable% is the time ratio that a sched_entity is runnable.
517 * For cfs_rq, it is the aggregated load_avg of all runnable and
518 * blocked sched_entities.
520 * load_avg may also take frequency scaling into account:
522 * load_avg = runnable% * scale_load_down(load) * freq%
524 * where freq% is the CPU frequency normalized to the highest frequency.
526 * [util_avg definition]
528 * util_avg = running% * SCHED_CAPACITY_SCALE
530 * where running% is the time ratio that a sched_entity is running on
531 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
532 * and blocked sched_entities.
534 * util_avg may also factor frequency scaling and CPU capacity scaling:
536 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
538 * where freq% is the same as above, and capacity% is the CPU capacity
539 * normalized to the greatest capacity (due to uarch differences, etc).
541 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
542 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
543 * we therefore scale them to as large a range as necessary. This is for
544 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
548 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
549 * with the highest load (=88761), always runnable on a single cfs_rq,
550 * and should not overflow as the number already hits PID_MAX_LIMIT.
552 * For all other cases (including 32-bit kernels), struct load_weight's
553 * weight will overflow first before we do, because:
555 * Max(load_avg) <= Max(load.weight)
557 * Then it is the load_weight's responsibility to consider overflow
561 u64 last_update_time, load_sum;
562 u32 util_sum, period_contrib;
563 unsigned long load_avg, util_avg;
566 #ifdef CONFIG_SCHEDSTATS
567 struct sched_statistics {
577 s64 sum_sleep_runtime;
584 u64 nr_migrations_cold;
585 u64 nr_failed_migrations_affine;
586 u64 nr_failed_migrations_running;
587 u64 nr_failed_migrations_hot;
588 u64 nr_forced_migrations;
592 u64 nr_wakeups_migrate;
593 u64 nr_wakeups_local;
594 u64 nr_wakeups_remote;
595 u64 nr_wakeups_affine;
596 u64 nr_wakeups_affine_attempts;
597 u64 nr_wakeups_passive;
602 struct sched_entity {
603 struct load_weight load; /* for load-balancing */
604 struct rb_node run_node;
605 struct list_head group_node;
609 u64 sum_exec_runtime;
611 u64 prev_sum_exec_runtime;
615 #ifdef CONFIG_SCHEDSTATS
616 struct sched_statistics statistics;
619 #ifdef CONFIG_FAIR_GROUP_SCHED
621 struct sched_entity *parent;
622 /* rq on which this entity is (to be) queued: */
623 struct cfs_rq *cfs_rq;
624 /* rq "owned" by this entity/group: */
630 * Per entity load average tracking.
632 * Put into separate cache line so it does not
633 * collide with read-mostly values above.
635 struct sched_avg avg ____cacheline_aligned_in_smp;
639 struct sched_rt_entity {
640 struct list_head run_list;
641 unsigned long timeout;
642 unsigned long watchdog_stamp;
643 unsigned int time_slice;
644 unsigned short on_rq;
645 unsigned short on_list;
647 struct sched_rt_entity *back;
648 #ifdef CONFIG_RT_GROUP_SCHED
649 struct sched_rt_entity *parent;
650 /* rq on which this entity is (to be) queued: */
652 /* rq "owned" by this entity/group: */
657 struct sched_dl_entity {
658 struct rb_node rb_node;
661 * Original scheduling parameters. Copied here from sched_attr
662 * during sched_setattr(), they will remain the same until
663 * the next sched_setattr().
665 u64 dl_runtime; /* maximum runtime for each instance */
666 u64 dl_deadline; /* relative deadline of each instance */
667 u64 dl_period; /* separation of two instances (period) */
668 u64 dl_bw; /* dl_runtime / dl_deadline */
671 * Actual scheduling parameters. Initialized with the values above,
672 * they are continously updated during task execution. Note that
673 * the remaining runtime could be < 0 in case we are in overrun.
675 s64 runtime; /* remaining runtime for this instance */
676 u64 deadline; /* absolute deadline for this instance */
677 unsigned int flags; /* specifying the scheduler behaviour */
682 * @dl_throttled tells if we exhausted the runtime. If so, the
683 * task has to wait for a replenishment to be performed at the
684 * next firing of dl_timer.
686 * @dl_boosted tells if we are boosted due to DI. If so we are
687 * outside bandwidth enforcement mechanism (but only until we
688 * exit the critical section);
690 * @dl_yielded tells if task gave up the cpu before consuming
691 * all its available runtime during the last job.
693 int dl_throttled, dl_boosted, dl_yielded;
696 * Bandwidth enforcement timer. Each -deadline task has its
697 * own bandwidth to be enforced, thus we need one timer per task.
699 struct hrtimer dl_timer;
707 u8 pad; /* Otherwise the compiler can store garbage here. */
709 u32 s; /* Set of bits. */
713 enum perf_event_task_context {
714 perf_invalid_context = -1,
717 perf_nr_task_contexts,
721 struct wake_q_node *next;
724 /* Track pages that require TLB flushes */
725 struct tlbflush_unmap_batch {
727 * Each bit set is a CPU that potentially has a TLB entry for one of
728 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
730 struct cpumask cpumask;
732 /* True if any bit in cpumask is set */
736 * If true then the PTE was dirty when unmapped. The entry must be
737 * flushed before IO is initiated or a stale TLB entry potentially
738 * allows an update without redirtying the page.
744 #ifdef CONFIG_THREAD_INFO_IN_TASK
746 * For reasons of header soup (see current_thread_info()), this
747 * must be the first element of task_struct.
749 struct thread_info thread_info;
751 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
754 unsigned int flags; /* per process flags, defined below */
758 struct llist_node wake_entry;
760 #ifdef CONFIG_THREAD_INFO_IN_TASK
761 unsigned int cpu; /* current CPU */
763 unsigned int wakee_flips;
764 unsigned long wakee_flip_decay_ts;
765 struct task_struct *last_wakee;
771 int prio, static_prio, normal_prio;
772 unsigned int rt_priority;
773 const struct sched_class *sched_class;
774 struct sched_entity se;
775 struct sched_rt_entity rt;
776 #ifdef CONFIG_CGROUP_SCHED
777 struct task_group *sched_task_group;
779 struct sched_dl_entity dl;
781 #ifdef CONFIG_PREEMPT_NOTIFIERS
782 /* list of struct preempt_notifier: */
783 struct hlist_head preempt_notifiers;
786 #ifdef CONFIG_BLK_DEV_IO_TRACE
787 unsigned int btrace_seq;
792 cpumask_t cpus_allowed;
794 #ifdef CONFIG_PREEMPT_RCU
795 int rcu_read_lock_nesting;
796 union rcu_special rcu_read_unlock_special;
797 struct list_head rcu_node_entry;
798 struct rcu_node *rcu_blocked_node;
799 #endif /* #ifdef CONFIG_PREEMPT_RCU */
800 #ifdef CONFIG_TASKS_RCU
801 unsigned long rcu_tasks_nvcsw;
802 bool rcu_tasks_holdout;
803 struct list_head rcu_tasks_holdout_list;
804 int rcu_tasks_idle_cpu;
805 #endif /* #ifdef CONFIG_TASKS_RCU */
807 #ifdef CONFIG_SCHED_INFO
808 struct sched_info sched_info;
811 struct list_head tasks;
813 struct plist_node pushable_tasks;
814 struct rb_node pushable_dl_tasks;
817 struct mm_struct *mm, *active_mm;
819 /* Per-thread vma caching: */
820 struct vmacache vmacache;
822 #if defined(SPLIT_RSS_COUNTING)
823 struct task_rss_stat rss_stat;
827 int exit_code, exit_signal;
828 int pdeath_signal; /* The signal sent when the parent dies */
829 unsigned long jobctl; /* JOBCTL_*, siglock protected */
831 /* Used for emulating ABI behavior of previous Linux versions */
832 unsigned int personality;
834 /* scheduler bits, serialized by scheduler locks */
835 unsigned sched_reset_on_fork:1;
836 unsigned sched_contributes_to_load:1;
837 unsigned sched_migrated:1;
838 unsigned sched_remote_wakeup:1;
839 unsigned :0; /* force alignment to the next boundary */
841 /* unserialized, strictly 'current' */
842 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
843 unsigned in_iowait:1;
844 #if !defined(TIF_RESTORE_SIGMASK)
845 unsigned restore_sigmask:1;
848 unsigned memcg_may_oom:1;
850 unsigned memcg_kmem_skip_account:1;
853 #ifdef CONFIG_COMPAT_BRK
854 unsigned brk_randomized:1;
857 unsigned long atomic_flags; /* Flags needing atomic access. */
859 struct restart_block restart_block;
864 #ifdef CONFIG_CC_STACKPROTECTOR
865 /* Canary value for the -fstack-protector gcc feature */
866 unsigned long stack_canary;
869 * pointers to (original) parent process, youngest child, younger sibling,
870 * older sibling, respectively. (p->father can be replaced with
871 * p->real_parent->pid)
873 struct task_struct __rcu *real_parent; /* real parent process */
874 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
876 * children/sibling forms the list of my natural children
878 struct list_head children; /* list of my children */
879 struct list_head sibling; /* linkage in my parent's children list */
880 struct task_struct *group_leader; /* threadgroup leader */
883 * ptraced is the list of tasks this task is using ptrace on.
884 * This includes both natural children and PTRACE_ATTACH targets.
885 * p->ptrace_entry is p's link on the p->parent->ptraced list.
887 struct list_head ptraced;
888 struct list_head ptrace_entry;
890 /* PID/PID hash table linkage. */
891 struct pid_link pids[PIDTYPE_MAX];
892 struct list_head thread_group;
893 struct list_head thread_node;
895 struct completion *vfork_done; /* for vfork() */
896 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
897 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
900 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
901 u64 utimescaled, stimescaled;
904 struct prev_cputime prev_cputime;
905 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
906 seqcount_t vtime_seqcount;
907 unsigned long long vtime_snap;
909 /* Task is sleeping or running in a CPU with VTIME inactive */
911 /* Task runs in userspace in a CPU with VTIME active */
913 /* Task runs in kernelspace in a CPU with VTIME active */
918 #ifdef CONFIG_NO_HZ_FULL
919 atomic_t tick_dep_mask;
921 unsigned long nvcsw, nivcsw; /* context switch counts */
922 u64 start_time; /* monotonic time in nsec */
923 u64 real_start_time; /* boot based time in nsec */
924 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
925 unsigned long min_flt, maj_flt;
927 #ifdef CONFIG_POSIX_TIMERS
928 struct task_cputime cputime_expires;
929 struct list_head cpu_timers[3];
932 /* process credentials */
933 const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
934 const struct cred __rcu *real_cred; /* objective and real subjective task
935 * credentials (COW) */
936 const struct cred __rcu *cred; /* effective (overridable) subjective task
937 * credentials (COW) */
938 char comm[TASK_COMM_LEN]; /* executable name excluding path
939 - access with [gs]et_task_comm (which lock
941 - initialized normally by setup_new_exec */
942 /* file system info */
943 struct nameidata *nameidata;
944 #ifdef CONFIG_SYSVIPC
946 struct sysv_sem sysvsem;
947 struct sysv_shm sysvshm;
949 #ifdef CONFIG_DETECT_HUNG_TASK
950 /* hung task detection */
951 unsigned long last_switch_count;
953 /* filesystem information */
954 struct fs_struct *fs;
955 /* open file information */
956 struct files_struct *files;
958 struct nsproxy *nsproxy;
959 /* signal handlers */
960 struct signal_struct *signal;
961 struct sighand_struct *sighand;
963 sigset_t blocked, real_blocked;
964 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
965 struct sigpending pending;
967 unsigned long sas_ss_sp;
969 unsigned sas_ss_flags;
971 struct callback_head *task_works;
973 struct audit_context *audit_context;
974 #ifdef CONFIG_AUDITSYSCALL
976 unsigned int sessionid;
978 struct seccomp seccomp;
980 /* Thread group tracking */
983 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
985 spinlock_t alloc_lock;
987 /* Protection of the PI data structures: */
988 raw_spinlock_t pi_lock;
990 struct wake_q_node wake_q;
992 #ifdef CONFIG_RT_MUTEXES
993 /* PI waiters blocked on a rt_mutex held by this task */
994 struct rb_root pi_waiters;
995 struct rb_node *pi_waiters_leftmost;
996 /* Deadlock detection and priority inheritance handling */
997 struct rt_mutex_waiter *pi_blocked_on;
1000 #ifdef CONFIG_DEBUG_MUTEXES
1001 /* mutex deadlock detection */
1002 struct mutex_waiter *blocked_on;
1004 #ifdef CONFIG_TRACE_IRQFLAGS
1005 unsigned int irq_events;
1006 unsigned long hardirq_enable_ip;
1007 unsigned long hardirq_disable_ip;
1008 unsigned int hardirq_enable_event;
1009 unsigned int hardirq_disable_event;
1010 int hardirqs_enabled;
1011 int hardirq_context;
1012 unsigned long softirq_disable_ip;
1013 unsigned long softirq_enable_ip;
1014 unsigned int softirq_disable_event;
1015 unsigned int softirq_enable_event;
1016 int softirqs_enabled;
1017 int softirq_context;
1019 #ifdef CONFIG_LOCKDEP
1020 # define MAX_LOCK_DEPTH 48UL
1023 unsigned int lockdep_recursion;
1024 struct held_lock held_locks[MAX_LOCK_DEPTH];
1025 gfp_t lockdep_reclaim_gfp;
1028 unsigned int in_ubsan;
1031 /* journalling filesystem info */
1034 /* stacked block device info */
1035 struct bio_list *bio_list;
1038 /* stack plugging */
1039 struct blk_plug *plug;
1043 struct reclaim_state *reclaim_state;
1045 struct backing_dev_info *backing_dev_info;
1047 struct io_context *io_context;
1049 unsigned long ptrace_message;
1050 siginfo_t *last_siginfo; /* For ptrace use. */
1051 struct task_io_accounting ioac;
1052 #if defined(CONFIG_TASK_XACCT)
1053 u64 acct_rss_mem1; /* accumulated rss usage */
1054 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1055 u64 acct_timexpd; /* stime + utime since last update */
1057 #ifdef CONFIG_CPUSETS
1058 nodemask_t mems_allowed; /* Protected by alloc_lock */
1059 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1060 int cpuset_mem_spread_rotor;
1061 int cpuset_slab_spread_rotor;
1063 #ifdef CONFIG_CGROUPS
1064 /* Control Group info protected by css_set_lock */
1065 struct css_set __rcu *cgroups;
1066 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1067 struct list_head cg_list;
1069 #ifdef CONFIG_INTEL_RDT_A
1073 struct robust_list_head __user *robust_list;
1074 #ifdef CONFIG_COMPAT
1075 struct compat_robust_list_head __user *compat_robust_list;
1077 struct list_head pi_state_list;
1078 struct futex_pi_state *pi_state_cache;
1080 #ifdef CONFIG_PERF_EVENTS
1081 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1082 struct mutex perf_event_mutex;
1083 struct list_head perf_event_list;
1085 #ifdef CONFIG_DEBUG_PREEMPT
1086 unsigned long preempt_disable_ip;
1089 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1091 short pref_node_fork;
1093 #ifdef CONFIG_NUMA_BALANCING
1095 unsigned int numa_scan_period;
1096 unsigned int numa_scan_period_max;
1097 int numa_preferred_nid;
1098 unsigned long numa_migrate_retry;
1099 u64 node_stamp; /* migration stamp */
1100 u64 last_task_numa_placement;
1101 u64 last_sum_exec_runtime;
1102 struct callback_head numa_work;
1104 struct list_head numa_entry;
1105 struct numa_group *numa_group;
1108 * numa_faults is an array split into four regions:
1109 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1110 * in this precise order.
1112 * faults_memory: Exponential decaying average of faults on a per-node
1113 * basis. Scheduling placement decisions are made based on these
1114 * counts. The values remain static for the duration of a PTE scan.
1115 * faults_cpu: Track the nodes the process was running on when a NUMA
1116 * hinting fault was incurred.
1117 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1118 * during the current scan window. When the scan completes, the counts
1119 * in faults_memory and faults_cpu decay and these values are copied.
1121 unsigned long *numa_faults;
1122 unsigned long total_numa_faults;
1125 * numa_faults_locality tracks if faults recorded during the last
1126 * scan window were remote/local or failed to migrate. The task scan
1127 * period is adapted based on the locality of the faults with different
1128 * weights depending on whether they were shared or private faults
1130 unsigned long numa_faults_locality[3];
1132 unsigned long numa_pages_migrated;
1133 #endif /* CONFIG_NUMA_BALANCING */
1135 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1136 struct tlbflush_unmap_batch tlb_ubc;
1139 struct rcu_head rcu;
1142 * cache last used pipe for splice
1144 struct pipe_inode_info *splice_pipe;
1146 struct page_frag task_frag;
1148 #ifdef CONFIG_TASK_DELAY_ACCT
1149 struct task_delay_info *delays;
1152 #ifdef CONFIG_FAULT_INJECTION
1156 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1157 * balance_dirty_pages() for some dirty throttling pause
1160 int nr_dirtied_pause;
1161 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1163 #ifdef CONFIG_LATENCYTOP
1164 int latency_record_count;
1165 struct latency_record latency_record[LT_SAVECOUNT];
1168 * time slack values; these are used to round up poll() and
1169 * select() etc timeout values. These are in nanoseconds.
1172 u64 default_timer_slack_ns;
1175 unsigned int kasan_depth;
1177 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1178 /* Index of current stored address in ret_stack */
1180 /* Stack of return addresses for return function tracing */
1181 struct ftrace_ret_stack *ret_stack;
1182 /* time stamp for last schedule */
1183 unsigned long long ftrace_timestamp;
1185 * Number of functions that haven't been traced
1186 * because of depth overrun.
1188 atomic_t trace_overrun;
1189 /* Pause for the tracing */
1190 atomic_t tracing_graph_pause;
1192 #ifdef CONFIG_TRACING
1193 /* state flags for use by tracers */
1194 unsigned long trace;
1195 /* bitmask and counter of trace recursion */
1196 unsigned long trace_recursion;
1197 #endif /* CONFIG_TRACING */
1199 /* Coverage collection mode enabled for this task (0 if disabled). */
1200 enum kcov_mode kcov_mode;
1201 /* Size of the kcov_area. */
1203 /* Buffer for coverage collection. */
1205 /* kcov desciptor wired with this task or NULL. */
1209 struct mem_cgroup *memcg_in_oom;
1210 gfp_t memcg_oom_gfp_mask;
1211 int memcg_oom_order;
1213 /* number of pages to reclaim on returning to userland */
1214 unsigned int memcg_nr_pages_over_high;
1216 #ifdef CONFIG_UPROBES
1217 struct uprobe_task *utask;
1219 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1220 unsigned int sequential_io;
1221 unsigned int sequential_io_avg;
1223 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1224 unsigned long task_state_change;
1226 int pagefault_disabled;
1228 struct task_struct *oom_reaper_list;
1230 #ifdef CONFIG_VMAP_STACK
1231 struct vm_struct *stack_vm_area;
1233 #ifdef CONFIG_THREAD_INFO_IN_TASK
1234 /* A live task holds one reference. */
1235 atomic_t stack_refcount;
1237 /* CPU-specific state of this task */
1238 struct thread_struct thread;
1240 * WARNING: on x86, 'thread_struct' contains a variable-sized
1241 * structure. It *MUST* be at the end of 'task_struct'.
1243 * Do not put anything below here!
1247 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1248 extern int arch_task_struct_size __read_mostly;
1250 # define arch_task_struct_size (sizeof(struct task_struct))
1253 #ifdef CONFIG_VMAP_STACK
1254 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
1256 return t->stack_vm_area;
1259 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
1265 #define TNF_MIGRATED 0x01
1266 #define TNF_NO_GROUP 0x02
1267 #define TNF_SHARED 0x04
1268 #define TNF_FAULT_LOCAL 0x08
1269 #define TNF_MIGRATE_FAIL 0x10
1271 static inline bool in_vfork(struct task_struct *tsk)
1276 * need RCU to access ->real_parent if CLONE_VM was used along with
1279 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
1282 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
1283 * ->real_parent is not necessarily the task doing vfork(), so in
1284 * theory we can't rely on task_lock() if we want to dereference it.
1286 * And in this case we can't trust the real_parent->mm == tsk->mm
1287 * check, it can be false negative. But we do not care, if init or
1288 * another oom-unkillable task does this it should blame itself.
1291 ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
1297 #ifdef CONFIG_NUMA_BALANCING
1298 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1299 extern pid_t task_numa_group_id(struct task_struct *p);
1300 extern void set_numabalancing_state(bool enabled);
1301 extern void task_numa_free(struct task_struct *p);
1302 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1303 int src_nid, int dst_cpu);
1305 static inline void task_numa_fault(int last_node, int node, int pages,
1309 static inline pid_t task_numa_group_id(struct task_struct *p)
1313 static inline void set_numabalancing_state(bool enabled)
1316 static inline void task_numa_free(struct task_struct *p)
1319 static inline bool should_numa_migrate_memory(struct task_struct *p,
1320 struct page *page, int src_nid, int dst_cpu)
1326 static inline struct pid *task_pid(struct task_struct *task)
1328 return task->pids[PIDTYPE_PID].pid;
1331 static inline struct pid *task_tgid(struct task_struct *task)
1333 return task->group_leader->pids[PIDTYPE_PID].pid;
1337 * Without tasklist or rcu lock it is not safe to dereference
1338 * the result of task_pgrp/task_session even if task == current,
1339 * we can race with another thread doing sys_setsid/sys_setpgid.
1341 static inline struct pid *task_pgrp(struct task_struct *task)
1343 return task->group_leader->pids[PIDTYPE_PGID].pid;
1346 static inline struct pid *task_session(struct task_struct *task)
1348 return task->group_leader->pids[PIDTYPE_SID].pid;
1351 struct pid_namespace;
1354 * the helpers to get the task's different pids as they are seen
1355 * from various namespaces
1357 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1358 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1360 * task_xid_nr_ns() : id seen from the ns specified;
1362 * set_task_vxid() : assigns a virtual id to a task;
1364 * see also pid_nr() etc in include/linux/pid.h
1366 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1367 struct pid_namespace *ns);
1369 static inline pid_t task_pid_nr(struct task_struct *tsk)
1374 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1375 struct pid_namespace *ns)
1377 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1380 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1382 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1386 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1391 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1393 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1395 return pid_vnr(task_tgid(tsk));
1399 static inline int pid_alive(const struct task_struct *p);
1400 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1406 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1412 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1414 return task_ppid_nr_ns(tsk, &init_pid_ns);
1417 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1418 struct pid_namespace *ns)
1420 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1423 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1425 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1429 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1430 struct pid_namespace *ns)
1432 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1435 static inline pid_t task_session_vnr(struct task_struct *tsk)
1437 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1440 /* obsolete, do not use */
1441 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1443 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1447 * pid_alive - check that a task structure is not stale
1448 * @p: Task structure to be checked.
1450 * Test if a process is not yet dead (at most zombie state)
1451 * If pid_alive fails, then pointers within the task structure
1452 * can be stale and must not be dereferenced.
1454 * Return: 1 if the process is alive. 0 otherwise.
1456 static inline int pid_alive(const struct task_struct *p)
1458 return p->pids[PIDTYPE_PID].pid != NULL;
1462 * is_global_init - check if a task structure is init. Since init
1463 * is free to have sub-threads we need to check tgid.
1464 * @tsk: Task structure to be checked.
1466 * Check if a task structure is the first user space task the kernel created.
1468 * Return: 1 if the task structure is init. 0 otherwise.
1470 static inline int is_global_init(struct task_struct *tsk)
1472 return task_tgid_nr(tsk) == 1;
1475 extern struct pid *cad_pid;
1477 extern void free_task(struct task_struct *tsk);
1478 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1480 extern void __put_task_struct(struct task_struct *t);
1482 static inline void put_task_struct(struct task_struct *t)
1484 if (atomic_dec_and_test(&t->usage))
1485 __put_task_struct(t);
1488 struct task_struct *task_rcu_dereference(struct task_struct **ptask);
1489 struct task_struct *try_get_task_struct(struct task_struct **ptask);
1491 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1492 extern void task_cputime(struct task_struct *t,
1493 u64 *utime, u64 *stime);
1494 extern u64 task_gtime(struct task_struct *t);
1496 static inline void task_cputime(struct task_struct *t,
1497 u64 *utime, u64 *stime)
1503 static inline u64 task_gtime(struct task_struct *t)
1509 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
1510 static inline void task_cputime_scaled(struct task_struct *t,
1514 *utimescaled = t->utimescaled;
1515 *stimescaled = t->stimescaled;
1518 static inline void task_cputime_scaled(struct task_struct *t,
1522 task_cputime(t, utimescaled, stimescaled);
1526 extern void task_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
1527 extern void thread_group_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st);
1532 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1533 #define PF_EXITING 0x00000004 /* getting shut down */
1534 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1535 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1536 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1537 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1538 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1539 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1540 #define PF_DUMPCORE 0x00000200 /* dumped core */
1541 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1542 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1543 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1544 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1545 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1546 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1547 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1548 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1549 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1550 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1551 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1552 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1553 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1554 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1555 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1556 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1557 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1558 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1559 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1562 * Only the _current_ task can read/write to tsk->flags, but other
1563 * tasks can access tsk->flags in readonly mode for example
1564 * with tsk_used_math (like during threaded core dumping).
1565 * There is however an exception to this rule during ptrace
1566 * or during fork: the ptracer task is allowed to write to the
1567 * child->flags of its traced child (same goes for fork, the parent
1568 * can write to the child->flags), because we're guaranteed the
1569 * child is not running and in turn not changing child->flags
1570 * at the same time the parent does it.
1572 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1573 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1574 #define clear_used_math() clear_stopped_child_used_math(current)
1575 #define set_used_math() set_stopped_child_used_math(current)
1576 #define conditional_stopped_child_used_math(condition, child) \
1577 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1578 #define conditional_used_math(condition) \
1579 conditional_stopped_child_used_math(condition, current)
1580 #define copy_to_stopped_child_used_math(child) \
1581 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1582 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1583 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1584 #define used_math() tsk_used_math(current)
1586 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1587 * __GFP_FS is also cleared as it implies __GFP_IO.
1589 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1591 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1592 flags &= ~(__GFP_IO | __GFP_FS);
1596 static inline unsigned int memalloc_noio_save(void)
1598 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1599 current->flags |= PF_MEMALLOC_NOIO;
1603 static inline void memalloc_noio_restore(unsigned int flags)
1605 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1608 /* Per-process atomic flags. */
1609 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1610 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1611 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1612 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
1615 #define TASK_PFA_TEST(name, func) \
1616 static inline bool task_##func(struct task_struct *p) \
1617 { return test_bit(PFA_##name, &p->atomic_flags); }
1618 #define TASK_PFA_SET(name, func) \
1619 static inline void task_set_##func(struct task_struct *p) \
1620 { set_bit(PFA_##name, &p->atomic_flags); }
1621 #define TASK_PFA_CLEAR(name, func) \
1622 static inline void task_clear_##func(struct task_struct *p) \
1623 { clear_bit(PFA_##name, &p->atomic_flags); }
1625 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
1626 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
1628 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
1629 TASK_PFA_SET(SPREAD_PAGE, spread_page)
1630 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
1632 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
1633 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
1634 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
1636 TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
1637 TASK_PFA_SET(LMK_WAITING, lmk_waiting)
1640 * task->jobctl flags
1642 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1644 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1645 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1646 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1647 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1648 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1649 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1650 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1652 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
1653 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
1654 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
1655 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
1656 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
1657 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
1658 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
1660 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1661 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1663 extern bool task_set_jobctl_pending(struct task_struct *task,
1664 unsigned long mask);
1665 extern void task_clear_jobctl_trapping(struct task_struct *task);
1666 extern void task_clear_jobctl_pending(struct task_struct *task,
1667 unsigned long mask);
1669 static inline void rcu_copy_process(struct task_struct *p)
1671 #ifdef CONFIG_PREEMPT_RCU
1672 p->rcu_read_lock_nesting = 0;
1673 p->rcu_read_unlock_special.s = 0;
1674 p->rcu_blocked_node = NULL;
1675 INIT_LIST_HEAD(&p->rcu_node_entry);
1676 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1677 #ifdef CONFIG_TASKS_RCU
1678 p->rcu_tasks_holdout = false;
1679 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
1680 p->rcu_tasks_idle_cpu = -1;
1681 #endif /* #ifdef CONFIG_TASKS_RCU */
1684 static inline void tsk_restore_flags(struct task_struct *task,
1685 unsigned long orig_flags, unsigned long flags)
1687 task->flags &= ~flags;
1688 task->flags |= orig_flags & flags;
1691 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
1692 const struct cpumask *trial);
1693 extern int task_can_attach(struct task_struct *p,
1694 const struct cpumask *cs_cpus_allowed);
1696 extern void do_set_cpus_allowed(struct task_struct *p,
1697 const struct cpumask *new_mask);
1699 extern int set_cpus_allowed_ptr(struct task_struct *p,
1700 const struct cpumask *new_mask);
1702 static inline void do_set_cpus_allowed(struct task_struct *p,
1703 const struct cpumask *new_mask)
1706 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1707 const struct cpumask *new_mask)
1709 if (!cpumask_test_cpu(0, new_mask))
1715 #ifdef CONFIG_NO_HZ_COMMON
1716 void calc_load_enter_idle(void);
1717 void calc_load_exit_idle(void);
1719 static inline void calc_load_enter_idle(void) { }
1720 static inline void calc_load_exit_idle(void) { }
1721 #endif /* CONFIG_NO_HZ_COMMON */
1723 #ifndef cpu_relax_yield
1724 #define cpu_relax_yield() cpu_relax()
1727 extern unsigned long long
1728 task_sched_runtime(struct task_struct *task);
1730 /* sched_exec is called by processes performing an exec */
1732 extern void sched_exec(void);
1734 #define sched_exec() {}
1737 #ifdef CONFIG_HOTPLUG_CPU
1738 extern void idle_task_exit(void);
1740 static inline void idle_task_exit(void) {}
1743 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1744 extern void wake_up_nohz_cpu(int cpu);
1746 static inline void wake_up_nohz_cpu(int cpu) { }
1749 #ifdef CONFIG_NO_HZ_FULL
1750 extern u64 scheduler_tick_max_deferment(void);
1753 extern int yield_to(struct task_struct *p, bool preempt);
1754 extern void set_user_nice(struct task_struct *p, long nice);
1755 extern int task_prio(const struct task_struct *p);
1757 * task_nice - return the nice value of a given task.
1758 * @p: the task in question.
1760 * Return: The nice value [ -20 ... 0 ... 19 ].
1762 static inline int task_nice(const struct task_struct *p)
1764 return PRIO_TO_NICE((p)->static_prio);
1766 extern int can_nice(const struct task_struct *p, const int nice);
1767 extern int task_curr(const struct task_struct *p);
1768 extern int idle_cpu(int cpu);
1769 extern int sched_setscheduler(struct task_struct *, int,
1770 const struct sched_param *);
1771 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1772 const struct sched_param *);
1773 extern int sched_setattr(struct task_struct *,
1774 const struct sched_attr *);
1775 extern struct task_struct *idle_task(int cpu);
1777 * is_idle_task - is the specified task an idle task?
1778 * @p: the task in question.
1780 * Return: 1 if @p is an idle task. 0 otherwise.
1782 static inline bool is_idle_task(const struct task_struct *p)
1784 return !!(p->flags & PF_IDLE);
1786 extern struct task_struct *curr_task(int cpu);
1787 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
1791 union thread_union {
1792 #ifndef CONFIG_THREAD_INFO_IN_TASK
1793 struct thread_info thread_info;
1795 unsigned long stack[THREAD_SIZE/sizeof(long)];
1798 #ifndef __HAVE_ARCH_KSTACK_END
1799 static inline int kstack_end(void *addr)
1801 /* Reliable end of stack detection:
1802 * Some APM bios versions misalign the stack
1804 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1808 extern union thread_union init_thread_union;
1809 extern struct task_struct init_task;
1811 extern struct mm_struct init_mm;
1813 extern struct pid_namespace init_pid_ns;
1816 * find a task by one of its numerical ids
1818 * find_task_by_pid_ns():
1819 * finds a task by its pid in the specified namespace
1820 * find_task_by_vpid():
1821 * finds a task by its virtual pid
1823 * see also find_vpid() etc in include/linux/pid.h
1826 extern struct task_struct *find_task_by_vpid(pid_t nr);
1827 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1828 struct pid_namespace *ns);
1830 #include <asm/current.h>
1832 extern void xtime_update(unsigned long ticks);
1834 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
1835 extern int wake_up_process(struct task_struct *tsk);
1836 extern void wake_up_new_task(struct task_struct *tsk);
1838 extern void kick_process(struct task_struct *tsk);
1840 static inline void kick_process(struct task_struct *tsk) { }
1842 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
1843 extern void sched_dead(struct task_struct *p);
1845 extern void proc_caches_init(void);
1847 extern void release_task(struct task_struct * p);
1849 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
1850 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
1851 struct task_struct *, unsigned long);
1853 extern int copy_thread(unsigned long, unsigned long, unsigned long,
1854 struct task_struct *);
1856 /* Architectures that haven't opted into copy_thread_tls get the tls argument
1857 * via pt_regs, so ignore the tls argument passed via C. */
1858 static inline int copy_thread_tls(
1859 unsigned long clone_flags, unsigned long sp, unsigned long arg,
1860 struct task_struct *p, unsigned long tls)
1862 return copy_thread(clone_flags, sp, arg, p);
1865 extern void flush_thread(void);
1867 #ifdef CONFIG_HAVE_EXIT_THREAD
1868 extern void exit_thread(struct task_struct *tsk);
1870 static inline void exit_thread(struct task_struct *tsk)
1875 extern void exit_files(struct task_struct *);
1877 extern void exit_itimers(struct signal_struct *);
1879 extern void do_group_exit(int);
1881 extern int do_execve(struct filename *,
1882 const char __user * const __user *,
1883 const char __user * const __user *);
1884 extern int do_execveat(int, struct filename *,
1885 const char __user * const __user *,
1886 const char __user * const __user *,
1888 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
1889 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
1890 struct task_struct *fork_idle(int);
1891 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
1893 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
1894 static inline void set_task_comm(struct task_struct *tsk, const char *from)
1896 __set_task_comm(tsk, from, false);
1898 extern char *get_task_comm(char *to, struct task_struct *tsk);
1901 void scheduler_ipi(void);
1902 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
1904 static inline void scheduler_ipi(void) { }
1905 static inline unsigned long wait_task_inactive(struct task_struct *p,
1913 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
1914 * subscriptions and synchronises with wait4(). Also used in procfs. Also
1915 * pins the final release of task.io_context. Also protects ->cpuset and
1916 * ->cgroup.subsys[]. And ->vfork_done.
1918 * Nests both inside and outside of read_lock(&tasklist_lock).
1919 * It must not be nested with write_lock_irq(&tasklist_lock),
1920 * neither inside nor outside.
1922 static inline void task_lock(struct task_struct *p)
1924 spin_lock(&p->alloc_lock);
1927 static inline void task_unlock(struct task_struct *p)
1929 spin_unlock(&p->alloc_lock);
1932 #ifdef CONFIG_THREAD_INFO_IN_TASK
1934 static inline struct thread_info *task_thread_info(struct task_struct *task)
1936 return &task->thread_info;
1940 * When accessing the stack of a non-current task that might exit, use
1941 * try_get_task_stack() instead. task_stack_page will return a pointer
1942 * that could get freed out from under you.
1944 static inline void *task_stack_page(const struct task_struct *task)
1949 #define setup_thread_stack(new,old) do { } while(0)
1951 static inline unsigned long *end_of_stack(const struct task_struct *task)
1956 #elif !defined(__HAVE_THREAD_FUNCTIONS)
1958 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
1959 #define task_stack_page(task) ((void *)(task)->stack)
1961 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
1963 *task_thread_info(p) = *task_thread_info(org);
1964 task_thread_info(p)->task = p;
1968 * Return the address of the last usable long on the stack.
1970 * When the stack grows down, this is just above the thread
1971 * info struct. Going any lower will corrupt the threadinfo.
1973 * When the stack grows up, this is the highest address.
1974 * Beyond that position, we corrupt data on the next page.
1976 static inline unsigned long *end_of_stack(struct task_struct *p)
1978 #ifdef CONFIG_STACK_GROWSUP
1979 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
1981 return (unsigned long *)(task_thread_info(p) + 1);
1987 #ifdef CONFIG_THREAD_INFO_IN_TASK
1988 static inline void *try_get_task_stack(struct task_struct *tsk)
1990 return atomic_inc_not_zero(&tsk->stack_refcount) ?
1991 task_stack_page(tsk) : NULL;
1994 extern void put_task_stack(struct task_struct *tsk);
1996 static inline void *try_get_task_stack(struct task_struct *tsk)
1998 return task_stack_page(tsk);
2001 static inline void put_task_stack(struct task_struct *tsk) {}
2004 #define task_stack_end_corrupted(task) \
2005 (*(end_of_stack(task)) != STACK_END_MAGIC)
2007 static inline int object_is_on_stack(void *obj)
2009 void *stack = task_stack_page(current);
2011 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2014 extern void thread_stack_cache_init(void);
2016 #ifdef CONFIG_DEBUG_STACK_USAGE
2017 static inline unsigned long stack_not_used(struct task_struct *p)
2019 unsigned long *n = end_of_stack(p);
2021 do { /* Skip over canary */
2022 # ifdef CONFIG_STACK_GROWSUP
2029 # ifdef CONFIG_STACK_GROWSUP
2030 return (unsigned long)end_of_stack(p) - (unsigned long)n;
2032 return (unsigned long)n - (unsigned long)end_of_stack(p);
2036 extern void set_task_stack_end_magic(struct task_struct *tsk);
2038 /* set thread flags in other task's structures
2039 * - see asm/thread_info.h for TIF_xxxx flags available
2041 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2043 set_ti_thread_flag(task_thread_info(tsk), flag);
2046 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2048 clear_ti_thread_flag(task_thread_info(tsk), flag);
2051 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2053 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2056 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2058 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2061 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2063 return test_ti_thread_flag(task_thread_info(tsk), flag);
2066 static inline void set_tsk_need_resched(struct task_struct *tsk)
2068 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2071 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2073 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2076 static inline int test_tsk_need_resched(struct task_struct *tsk)
2078 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2081 static inline int restart_syscall(void)
2083 set_tsk_thread_flag(current, TIF_SIGPENDING);
2084 return -ERESTARTNOINTR;
2087 static inline int signal_pending(struct task_struct *p)
2089 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2092 static inline int __fatal_signal_pending(struct task_struct *p)
2094 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2097 static inline int fatal_signal_pending(struct task_struct *p)
2099 return signal_pending(p) && __fatal_signal_pending(p);
2102 static inline int signal_pending_state(long state, struct task_struct *p)
2104 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2106 if (!signal_pending(p))
2109 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2113 * cond_resched() and cond_resched_lock(): latency reduction via
2114 * explicit rescheduling in places that are safe. The return
2115 * value indicates whether a reschedule was done in fact.
2116 * cond_resched_lock() will drop the spinlock before scheduling,
2117 * cond_resched_softirq() will enable bhs before scheduling.
2119 #ifndef CONFIG_PREEMPT
2120 extern int _cond_resched(void);
2122 static inline int _cond_resched(void) { return 0; }
2125 #define cond_resched() ({ \
2126 ___might_sleep(__FILE__, __LINE__, 0); \
2130 extern int __cond_resched_lock(spinlock_t *lock);
2132 #define cond_resched_lock(lock) ({ \
2133 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2134 __cond_resched_lock(lock); \
2137 extern int __cond_resched_softirq(void);
2139 #define cond_resched_softirq() ({ \
2140 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2141 __cond_resched_softirq(); \
2144 static inline void cond_resched_rcu(void)
2146 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2154 * Does a critical section need to be broken due to another
2155 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2156 * but a general need for low latency)
2158 static inline int spin_needbreak(spinlock_t *lock)
2160 #ifdef CONFIG_PREEMPT
2161 return spin_is_contended(lock);
2167 static __always_inline bool need_resched(void)
2169 return unlikely(tif_need_resched());
2173 * Thread group CPU time accounting.
2175 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2176 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2179 * Reevaluate whether the task has signals pending delivery.
2180 * Wake the task if so.
2181 * This is required every time the blocked sigset_t changes.
2182 * callers must hold sighand->siglock.
2184 extern void recalc_sigpending_and_wake(struct task_struct *t);
2185 extern void recalc_sigpending(void);
2187 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2189 static inline void signal_wake_up(struct task_struct *t, bool resume)
2191 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2193 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2195 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2199 * Wrappers for p->thread_info->cpu access. No-op on UP.
2203 static inline unsigned int task_cpu(const struct task_struct *p)
2205 #ifdef CONFIG_THREAD_INFO_IN_TASK
2208 return task_thread_info(p)->cpu;
2212 static inline int task_node(const struct task_struct *p)
2214 return cpu_to_node(task_cpu(p));
2217 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2221 static inline unsigned int task_cpu(const struct task_struct *p)
2226 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2230 #endif /* CONFIG_SMP */
2233 * In order to reduce various lock holder preemption latencies provide an
2234 * interface to see if a vCPU is currently running or not.
2236 * This allows us to terminate optimistic spin loops and block, analogous to
2237 * the native optimistic spin heuristic of testing if the lock owner task is
2240 #ifndef vcpu_is_preempted
2241 # define vcpu_is_preempted(cpu) false
2244 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2245 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2247 #ifdef CONFIG_CGROUP_SCHED
2248 extern struct task_group root_task_group;
2249 #endif /* CONFIG_CGROUP_SCHED */
2251 extern int task_can_switch_user(struct user_struct *up,
2252 struct task_struct *tsk);
2254 #ifdef CONFIG_TASK_XACCT
2255 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2257 tsk->ioac.rchar += amt;
2260 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2262 tsk->ioac.wchar += amt;
2265 static inline void inc_syscr(struct task_struct *tsk)
2270 static inline void inc_syscw(struct task_struct *tsk)
2275 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2279 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2283 static inline void inc_syscr(struct task_struct *tsk)
2287 static inline void inc_syscw(struct task_struct *tsk)
2292 #ifndef TASK_SIZE_OF
2293 #define TASK_SIZE_OF(tsk) TASK_SIZE
2297 extern void mm_update_next_owner(struct mm_struct *mm);
2299 static inline void mm_update_next_owner(struct mm_struct *mm)
2302 #endif /* CONFIG_MEMCG */
2304 #define SCHED_CPUFREQ_RT (1U << 0)
2305 #define SCHED_CPUFREQ_DL (1U << 1)
2306 #define SCHED_CPUFREQ_IOWAIT (1U << 2)
2308 #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
2310 #ifdef CONFIG_CPU_FREQ
2311 struct update_util_data {
2312 void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
2315 void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
2316 void (*func)(struct update_util_data *data, u64 time,
2317 unsigned int flags));
2318 void cpufreq_remove_update_util_hook(int cpu);
2319 #endif /* CONFIG_CPU_FREQ */