4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/kcov.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
61 #include <linux/cgroup-defs.h>
63 #include <asm/processor.h>
65 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
68 * Extended scheduling parameters data structure.
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
91 * This is reflected by the actual fields of the sched_attr structure:
93 * @size size of the structure, for fwd/bwd compat.
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
117 /* SCHED_NORMAL, SCHED_BATCH */
120 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state;
130 struct robust_list_head;
133 struct perf_event_context;
138 #define VMACACHE_BITS 2
139 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
140 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
152 extern unsigned long avenrun[]; /* Load averages */
153 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
155 #define FSHIFT 11 /* nr of bits of precision */
156 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
157 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
158 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159 #define EXP_5 2014 /* 1/exp(5sec/5min) */
160 #define EXP_15 2037 /* 1/exp(5sec/15min) */
162 #define CALC_LOAD(load,exp,n) \
164 load += n*(FIXED_1-exp); \
167 extern unsigned long total_forks;
168 extern int nr_threads;
169 DECLARE_PER_CPU(unsigned long, process_counts);
170 extern int nr_processes(void);
171 extern unsigned long nr_running(void);
172 extern bool single_task_running(void);
173 extern unsigned long nr_iowait(void);
174 extern unsigned long nr_iowait_cpu(int cpu);
175 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
176 #ifdef CONFIG_CPU_QUIET
177 extern u64 nr_running_integral(unsigned int cpu);
180 extern void calc_global_load(unsigned long ticks);
182 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
183 extern void cpu_load_update_nohz_start(void);
184 extern void cpu_load_update_nohz_stop(void);
186 static inline void cpu_load_update_nohz_start(void) { }
187 static inline void cpu_load_update_nohz_stop(void) { }
190 extern void dump_cpu_task(int cpu);
195 #ifdef CONFIG_SCHED_DEBUG
196 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
197 extern void proc_sched_set_task(struct task_struct *p);
201 * Task state bitmask. NOTE! These bits are also
202 * encoded in fs/proc/array.c: get_task_state().
204 * We have two separate sets of flags: task->state
205 * is about runnability, while task->exit_state are
206 * about the task exiting. Confusing, but this way
207 * modifying one set can't modify the other one by
210 #define TASK_RUNNING 0
211 #define TASK_INTERRUPTIBLE 1
212 #define TASK_UNINTERRUPTIBLE 2
213 #define __TASK_STOPPED 4
214 #define __TASK_TRACED 8
215 /* in tsk->exit_state */
217 #define EXIT_ZOMBIE 32
218 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
219 /* in tsk->state again */
221 #define TASK_WAKEKILL 128
222 #define TASK_WAKING 256
223 #define TASK_PARKED 512
224 #define TASK_NOLOAD 1024
225 #define TASK_NEW 2048
226 #define TASK_STATE_MAX 4096
228 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
230 extern char ___assert_task_state[1 - 2*!!(
231 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
233 /* Convenience macros for the sake of set_task_state */
234 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
235 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
236 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
238 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
240 /* Convenience macros for the sake of wake_up */
241 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
242 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
244 /* get_task_state() */
245 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
246 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
247 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
249 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
250 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
251 #define task_is_stopped_or_traced(task) \
252 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
253 #define task_contributes_to_load(task) \
254 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
255 (task->flags & PF_FROZEN) == 0 && \
256 (task->state & TASK_NOLOAD) == 0)
258 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
260 #define __set_task_state(tsk, state_value) \
262 (tsk)->task_state_change = _THIS_IP_; \
263 (tsk)->state = (state_value); \
265 #define set_task_state(tsk, state_value) \
267 (tsk)->task_state_change = _THIS_IP_; \
268 smp_store_mb((tsk)->state, (state_value)); \
272 * set_current_state() includes a barrier so that the write of current->state
273 * is correctly serialised wrt the caller's subsequent test of whether to
276 * set_current_state(TASK_UNINTERRUPTIBLE);
277 * if (do_i_need_to_sleep())
280 * If the caller does not need such serialisation then use __set_current_state()
282 #define __set_current_state(state_value) \
284 current->task_state_change = _THIS_IP_; \
285 current->state = (state_value); \
287 #define set_current_state(state_value) \
289 current->task_state_change = _THIS_IP_; \
290 smp_store_mb(current->state, (state_value)); \
295 #define __set_task_state(tsk, state_value) \
296 do { (tsk)->state = (state_value); } while (0)
297 #define set_task_state(tsk, state_value) \
298 smp_store_mb((tsk)->state, (state_value))
301 * set_current_state() includes a barrier so that the write of current->state
302 * is correctly serialised wrt the caller's subsequent test of whether to
305 * set_current_state(TASK_UNINTERRUPTIBLE);
306 * if (do_i_need_to_sleep())
309 * If the caller does not need such serialisation then use __set_current_state()
311 #define __set_current_state(state_value) \
312 do { current->state = (state_value); } while (0)
313 #define set_current_state(state_value) \
314 smp_store_mb(current->state, (state_value))
318 /* Task command name length */
319 #define TASK_COMM_LEN 16
330 #include <linux/spinlock.h>
333 * This serializes "schedule()" and also protects
334 * the run-queue from deletions/modifications (but
335 * _adding_ to the beginning of the run-queue has
338 extern rwlock_t tasklist_lock;
339 extern spinlock_t mmlist_lock;
343 #ifdef CONFIG_PROVE_RCU
344 extern int lockdep_tasklist_lock_is_held(void);
345 #endif /* #ifdef CONFIG_PROVE_RCU */
347 extern void sched_init(void);
348 extern void sched_init_smp(void);
349 extern asmlinkage void schedule_tail(struct task_struct *prev);
350 extern void init_idle(struct task_struct *idle, int cpu);
351 extern void init_idle_bootup_task(struct task_struct *idle);
353 extern cpumask_var_t cpu_isolated_map;
355 extern int runqueue_is_locked(int cpu);
357 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
358 extern void nohz_balance_enter_idle(int cpu);
359 extern void set_cpu_sd_state_idle(void);
360 extern int get_nohz_timer_target(void);
362 static inline void nohz_balance_enter_idle(int cpu) { }
363 static inline void set_cpu_sd_state_idle(void) { }
367 * Only dump TASK_* tasks. (0 for all tasks)
369 extern void show_state_filter(unsigned long state_filter);
371 static inline void show_state(void)
373 show_state_filter(0);
376 extern void show_regs(struct pt_regs *);
379 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
380 * task), SP is the stack pointer of the first frame that should be shown in the back
381 * trace (or NULL if the entire call-chain of the task should be shown).
383 extern void show_stack(struct task_struct *task, unsigned long *sp);
385 extern void cpu_init (void);
386 extern void trap_init(void);
387 extern void update_process_times(int user);
388 extern void scheduler_tick(void);
389 extern int sched_cpu_starting(unsigned int cpu);
390 extern int sched_cpu_activate(unsigned int cpu);
391 extern int sched_cpu_deactivate(unsigned int cpu);
393 #ifdef CONFIG_HOTPLUG_CPU
394 extern int sched_cpu_dying(unsigned int cpu);
396 # define sched_cpu_dying NULL
399 extern void sched_show_task(struct task_struct *p);
401 #ifdef CONFIG_LOCKUP_DETECTOR
402 extern void touch_softlockup_watchdog_sched(void);
403 extern void touch_softlockup_watchdog(void);
404 extern void touch_softlockup_watchdog_sync(void);
405 extern void touch_all_softlockup_watchdogs(void);
406 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
408 size_t *lenp, loff_t *ppos);
409 extern unsigned int softlockup_panic;
410 extern unsigned int hardlockup_panic;
411 void lockup_detector_init(void);
413 static inline void touch_softlockup_watchdog_sched(void)
416 static inline void touch_softlockup_watchdog(void)
419 static inline void touch_softlockup_watchdog_sync(void)
422 static inline void touch_all_softlockup_watchdogs(void)
425 static inline void lockup_detector_init(void)
430 #ifdef CONFIG_DETECT_HUNG_TASK
431 void reset_hung_task_detector(void);
433 static inline void reset_hung_task_detector(void)
438 /* Attach to any functions which should be ignored in wchan output. */
439 #define __sched __attribute__((__section__(".sched.text")))
441 /* Linker adds these: start and end of __sched functions */
442 extern char __sched_text_start[], __sched_text_end[];
444 /* Is this address in the __sched functions? */
445 extern int in_sched_functions(unsigned long addr);
447 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
448 extern signed long schedule_timeout(signed long timeout);
449 extern signed long schedule_timeout_interruptible(signed long timeout);
450 extern signed long schedule_timeout_killable(signed long timeout);
451 extern signed long schedule_timeout_uninterruptible(signed long timeout);
452 extern signed long schedule_timeout_idle(signed long timeout);
453 asmlinkage void schedule(void);
454 extern void schedule_preempt_disabled(void);
456 extern long io_schedule_timeout(long timeout);
458 static inline void io_schedule(void)
460 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
463 void __noreturn do_task_dead(void);
466 struct user_namespace;
469 extern void arch_pick_mmap_layout(struct mm_struct *mm);
471 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
472 unsigned long, unsigned long);
474 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
475 unsigned long len, unsigned long pgoff,
476 unsigned long flags);
478 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
481 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
482 #define SUID_DUMP_USER 1 /* Dump as user of process */
483 #define SUID_DUMP_ROOT 2 /* Dump as root */
487 /* for SUID_DUMP_* above */
488 #define MMF_DUMPABLE_BITS 2
489 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
491 extern void set_dumpable(struct mm_struct *mm, int value);
493 * This returns the actual value of the suid_dumpable flag. For things
494 * that are using this for checking for privilege transitions, it must
495 * test against SUID_DUMP_USER rather than treating it as a boolean
498 static inline int __get_dumpable(unsigned long mm_flags)
500 return mm_flags & MMF_DUMPABLE_MASK;
503 static inline int get_dumpable(struct mm_struct *mm)
505 return __get_dumpable(mm->flags);
508 /* coredump filter bits */
509 #define MMF_DUMP_ANON_PRIVATE 2
510 #define MMF_DUMP_ANON_SHARED 3
511 #define MMF_DUMP_MAPPED_PRIVATE 4
512 #define MMF_DUMP_MAPPED_SHARED 5
513 #define MMF_DUMP_ELF_HEADERS 6
514 #define MMF_DUMP_HUGETLB_PRIVATE 7
515 #define MMF_DUMP_HUGETLB_SHARED 8
516 #define MMF_DUMP_DAX_PRIVATE 9
517 #define MMF_DUMP_DAX_SHARED 10
519 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
520 #define MMF_DUMP_FILTER_BITS 9
521 #define MMF_DUMP_FILTER_MASK \
522 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
523 #define MMF_DUMP_FILTER_DEFAULT \
524 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
525 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
527 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
528 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
530 # define MMF_DUMP_MASK_DEFAULT_ELF 0
532 /* leave room for more dump flags */
533 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
534 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
535 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
537 #define MMF_HAS_UPROBES 19 /* has uprobes */
538 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
539 #define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */
540 #define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */
541 #define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */
543 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
545 struct sighand_struct {
547 struct k_sigaction action[_NSIG];
549 wait_queue_head_t signalfd_wqh;
552 struct pacct_struct {
555 unsigned long ac_mem;
556 cputime_t ac_utime, ac_stime;
557 unsigned long ac_minflt, ac_majflt;
568 * struct prev_cputime - snaphsot of system and user cputime
569 * @utime: time spent in user mode
570 * @stime: time spent in system mode
571 * @lock: protects the above two fields
573 * Stores previous user/system time values such that we can guarantee
576 struct prev_cputime {
577 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
584 static inline void prev_cputime_init(struct prev_cputime *prev)
586 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
587 prev->utime = prev->stime = 0;
588 raw_spin_lock_init(&prev->lock);
593 * struct task_cputime - collected CPU time counts
594 * @utime: time spent in user mode, in &cputime_t units
595 * @stime: time spent in kernel mode, in &cputime_t units
596 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
598 * This structure groups together three kinds of CPU time that are tracked for
599 * threads and thread groups. Most things considering CPU time want to group
600 * these counts together and treat all three of them in parallel.
602 struct task_cputime {
605 unsigned long long sum_exec_runtime;
608 /* Alternate field names when used to cache expirations. */
609 #define virt_exp utime
610 #define prof_exp stime
611 #define sched_exp sum_exec_runtime
613 #define INIT_CPUTIME \
614 (struct task_cputime) { \
617 .sum_exec_runtime = 0, \
621 * This is the atomic variant of task_cputime, which can be used for
622 * storing and updating task_cputime statistics without locking.
624 struct task_cputime_atomic {
627 atomic64_t sum_exec_runtime;
630 #define INIT_CPUTIME_ATOMIC \
631 (struct task_cputime_atomic) { \
632 .utime = ATOMIC64_INIT(0), \
633 .stime = ATOMIC64_INIT(0), \
634 .sum_exec_runtime = ATOMIC64_INIT(0), \
637 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
640 * Disable preemption until the scheduler is running -- use an unconditional
641 * value so that it also works on !PREEMPT_COUNT kernels.
643 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
645 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
648 * Initial preempt_count value; reflects the preempt_count schedule invariant
649 * which states that during context switches:
651 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
653 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
654 * Note: See finish_task_switch().
656 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
659 * struct thread_group_cputimer - thread group interval timer counts
660 * @cputime_atomic: atomic thread group interval timers.
661 * @running: true when there are timers running and
662 * @cputime_atomic receives updates.
663 * @checking_timer: true when a thread in the group is in the
664 * process of checking for thread group timers.
666 * This structure contains the version of task_cputime, above, that is
667 * used for thread group CPU timer calculations.
669 struct thread_group_cputimer {
670 struct task_cputime_atomic cputime_atomic;
675 #include <linux/rwsem.h>
679 * NOTE! "signal_struct" does not have its own
680 * locking, because a shared signal_struct always
681 * implies a shared sighand_struct, so locking
682 * sighand_struct is always a proper superset of
683 * the locking of signal_struct.
685 struct signal_struct {
689 struct list_head thread_head;
691 wait_queue_head_t wait_chldexit; /* for wait4() */
693 /* current thread group signal load-balancing target: */
694 struct task_struct *curr_target;
696 /* shared signal handling: */
697 struct sigpending shared_pending;
699 /* thread group exit support */
702 * - notify group_exit_task when ->count is equal to notify_count
703 * - everyone except group_exit_task is stopped during signal delivery
704 * of fatal signals, group_exit_task processes the signal.
707 struct task_struct *group_exit_task;
709 /* thread group stop support, overloads group_exit_code too */
710 int group_stop_count;
711 unsigned int flags; /* see SIGNAL_* flags below */
714 * PR_SET_CHILD_SUBREAPER marks a process, like a service
715 * manager, to re-parent orphan (double-forking) child processes
716 * to this process instead of 'init'. The service manager is
717 * able to receive SIGCHLD signals and is able to investigate
718 * the process until it calls wait(). All children of this
719 * process will inherit a flag if they should look for a
720 * child_subreaper process at exit.
722 unsigned int is_child_subreaper:1;
723 unsigned int has_child_subreaper:1;
725 /* POSIX.1b Interval Timers */
727 struct list_head posix_timers;
729 /* ITIMER_REAL timer for the process */
730 struct hrtimer real_timer;
731 struct pid *leader_pid;
732 ktime_t it_real_incr;
735 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
736 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
737 * values are defined to 0 and 1 respectively
739 struct cpu_itimer it[2];
742 * Thread group totals for process CPU timers.
743 * See thread_group_cputimer(), et al, for details.
745 struct thread_group_cputimer cputimer;
747 /* Earliest-expiration cache. */
748 struct task_cputime cputime_expires;
750 #ifdef CONFIG_NO_HZ_FULL
751 atomic_t tick_dep_mask;
754 struct list_head cpu_timers[3];
756 struct pid *tty_old_pgrp;
758 /* boolean value for session group leader */
761 struct tty_struct *tty; /* NULL if no tty */
763 #ifdef CONFIG_SCHED_AUTOGROUP
764 struct autogroup *autogroup;
767 * Cumulative resource counters for dead threads in the group,
768 * and for reaped dead child processes forked by this group.
769 * Live threads maintain their own counters and add to these
770 * in __exit_signal, except for the group leader.
772 seqlock_t stats_lock;
773 cputime_t utime, stime, cutime, cstime;
776 struct prev_cputime prev_cputime;
777 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
778 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
779 unsigned long inblock, oublock, cinblock, coublock;
780 unsigned long maxrss, cmaxrss;
781 struct task_io_accounting ioac;
784 * Cumulative ns of schedule CPU time fo dead threads in the
785 * group, not including a zombie group leader, (This only differs
786 * from jiffies_to_ns(utime + stime) if sched_clock uses something
787 * other than jiffies.)
789 unsigned long long sum_sched_runtime;
792 * We don't bother to synchronize most readers of this at all,
793 * because there is no reader checking a limit that actually needs
794 * to get both rlim_cur and rlim_max atomically, and either one
795 * alone is a single word that can safely be read normally.
796 * getrlimit/setrlimit use task_lock(current->group_leader) to
797 * protect this instead of the siglock, because they really
798 * have no need to disable irqs.
800 struct rlimit rlim[RLIM_NLIMITS];
802 #ifdef CONFIG_BSD_PROCESS_ACCT
803 struct pacct_struct pacct; /* per-process accounting information */
805 #ifdef CONFIG_TASKSTATS
806 struct taskstats *stats;
810 struct tty_audit_buf *tty_audit_buf;
814 * Thread is the potential origin of an oom condition; kill first on
817 bool oom_flag_origin;
818 short oom_score_adj; /* OOM kill score adjustment */
819 short oom_score_adj_min; /* OOM kill score adjustment min value.
820 * Only settable by CAP_SYS_RESOURCE. */
821 struct mm_struct *oom_mm; /* recorded mm when the thread group got
822 * killed by the oom killer */
824 struct mutex cred_guard_mutex; /* guard against foreign influences on
825 * credential calculations
826 * (notably. ptrace) */
830 * Bits in flags field of signal_struct.
832 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
833 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
834 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
835 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
837 * Pending notifications to parent.
839 #define SIGNAL_CLD_STOPPED 0x00000010
840 #define SIGNAL_CLD_CONTINUED 0x00000020
841 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
843 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
845 #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
846 SIGNAL_STOP_CONTINUED)
848 static inline void signal_set_stop_flags(struct signal_struct *sig,
851 WARN_ON(sig->flags & (SIGNAL_GROUP_EXIT|SIGNAL_GROUP_COREDUMP));
852 sig->flags = (sig->flags & ~SIGNAL_STOP_MASK) | flags;
855 /* If true, all threads except ->group_exit_task have pending SIGKILL */
856 static inline int signal_group_exit(const struct signal_struct *sig)
858 return (sig->flags & SIGNAL_GROUP_EXIT) ||
859 (sig->group_exit_task != NULL);
863 * Some day this will be a full-fledged user tracking system..
866 atomic_t __count; /* reference count */
867 atomic_t processes; /* How many processes does this user have? */
868 atomic_t sigpending; /* How many pending signals does this user have? */
869 #ifdef CONFIG_INOTIFY_USER
870 atomic_t inotify_watches; /* How many inotify watches does this user have? */
871 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
873 #ifdef CONFIG_FANOTIFY
874 atomic_t fanotify_listeners;
877 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
879 #ifdef CONFIG_POSIX_MQUEUE
880 /* protected by mq_lock */
881 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
883 unsigned long locked_shm; /* How many pages of mlocked shm ? */
884 unsigned long unix_inflight; /* How many files in flight in unix sockets */
885 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
888 struct key *uid_keyring; /* UID specific keyring */
889 struct key *session_keyring; /* UID's default session keyring */
892 /* Hash table maintenance information */
893 struct hlist_node uidhash_node;
896 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
897 atomic_long_t locked_vm;
901 extern int uids_sysfs_init(void);
903 extern struct user_struct *find_user(kuid_t);
905 extern struct user_struct root_user;
906 #define INIT_USER (&root_user)
909 struct backing_dev_info;
910 struct reclaim_state;
912 #ifdef CONFIG_SCHED_INFO
914 /* cumulative counters */
915 unsigned long pcount; /* # of times run on this cpu */
916 unsigned long long run_delay; /* time spent waiting on a runqueue */
919 unsigned long long last_arrival,/* when we last ran on a cpu */
920 last_queued; /* when we were last queued to run */
922 #endif /* CONFIG_SCHED_INFO */
924 #ifdef CONFIG_TASK_DELAY_ACCT
925 struct task_delay_info {
927 unsigned int flags; /* Private per-task flags */
929 /* For each stat XXX, add following, aligned appropriately
931 * struct timespec XXX_start, XXX_end;
935 * Atomicity of updates to XXX_delay, XXX_count protected by
936 * single lock above (split into XXX_lock if contention is an issue).
940 * XXX_count is incremented on every XXX operation, the delay
941 * associated with the operation is added to XXX_delay.
942 * XXX_delay contains the accumulated delay time in nanoseconds.
944 u64 blkio_start; /* Shared by blkio, swapin */
945 u64 blkio_delay; /* wait for sync block io completion */
946 u64 swapin_delay; /* wait for swapin block io completion */
947 u32 blkio_count; /* total count of the number of sync block */
948 /* io operations performed */
949 u32 swapin_count; /* total count of the number of swapin block */
950 /* io operations performed */
953 u64 freepages_delay; /* wait for memory reclaim */
954 u32 freepages_count; /* total count of memory reclaim */
956 #endif /* CONFIG_TASK_DELAY_ACCT */
958 static inline int sched_info_on(void)
960 #ifdef CONFIG_SCHEDSTATS
962 #elif defined(CONFIG_TASK_DELAY_ACCT)
963 extern int delayacct_on;
970 #ifdef CONFIG_SCHEDSTATS
971 void force_schedstat_enabled(void);
982 * Integer metrics need fixed point arithmetic, e.g., sched/fair
983 * has a few: load, load_avg, util_avg, freq, and capacity.
985 * We define a basic fixed point arithmetic range, and then formalize
986 * all these metrics based on that basic range.
988 # define SCHED_FIXEDPOINT_SHIFT 10
989 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
992 * Increase resolution of cpu_capacity calculations
994 #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
995 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
997 struct sched_capacity_reqs {
1002 unsigned long total;
1006 * Wake-queues are lists of tasks with a pending wakeup, whose
1007 * callers have already marked the task as woken internally,
1008 * and can thus carry on. A common use case is being able to
1009 * do the wakeups once the corresponding user lock as been
1012 * We hold reference to each task in the list across the wakeup,
1013 * thus guaranteeing that the memory is still valid by the time
1014 * the actual wakeups are performed in wake_up_q().
1016 * One per task suffices, because there's never a need for a task to be
1017 * in two wake queues simultaneously; it is forbidden to abandon a task
1018 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
1019 * already in a wake queue, the wakeup will happen soon and the second
1020 * waker can just skip it.
1022 * The WAKE_Q macro declares and initializes the list head.
1023 * wake_up_q() does NOT reinitialize the list; it's expected to be
1024 * called near the end of a function, where the fact that the queue is
1025 * not used again will be easy to see by inspection.
1027 * Note that this can cause spurious wakeups. schedule() callers
1028 * must ensure the call is done inside a loop, confirming that the
1029 * wakeup condition has in fact occurred.
1031 struct wake_q_node {
1032 struct wake_q_node *next;
1035 struct wake_q_head {
1036 struct wake_q_node *first;
1037 struct wake_q_node **lastp;
1040 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1042 #define WAKE_Q(name) \
1043 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1045 extern void wake_q_add(struct wake_q_head *head,
1046 struct task_struct *task);
1047 extern void wake_up_q(struct wake_q_head *head);
1050 * sched-domains (multiprocessor balancing) declarations:
1053 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1054 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1055 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1056 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
1057 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
1058 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
1059 #define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
1060 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
1061 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1062 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1063 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1064 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1065 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1066 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1067 #define SD_NUMA 0x4000 /* cross-node balancing */
1068 #define SD_SHARE_CAP_STATES 0x8000 /* Domain members share capacity state */
1070 #ifdef CONFIG_SCHED_SMT
1071 static inline int cpu_smt_flags(void)
1073 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1077 #ifdef CONFIG_SCHED_MC
1078 static inline int cpu_core_flags(void)
1080 return SD_SHARE_PKG_RESOURCES;
1085 static inline int cpu_numa_flags(void)
1091 struct sched_domain_attr {
1092 int relax_domain_level;
1095 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1096 .relax_domain_level = -1, \
1099 extern int sched_domain_level_max;
1101 struct capacity_state {
1102 unsigned long cap; /* compute capacity */
1103 unsigned long power; /* power consumption at this compute capacity */
1107 unsigned long power; /* power consumption in this idle state */
1110 struct sched_group_energy {
1111 unsigned int nr_idle_states; /* number of idle states */
1112 struct idle_state *idle_states; /* ptr to idle state array */
1113 unsigned int nr_cap_states; /* number of capacity states */
1114 struct capacity_state *cap_states; /* ptr to capacity state array */
1117 unsigned long capacity_curr_of(int cpu);
1122 /* select_idle_sibling() stats */
1125 u64 sis_cache_affine;
1130 /* select_energy_cpu_brute() stats */
1134 u64 secb_insuff_cap;
1135 u64 secb_no_nrg_sav;
1139 /* find_best_target() stats */
1147 /* select_task_rq_fair() stats */
1152 struct sched_domain_shared {
1154 atomic_t nr_busy_cpus;
1158 struct sched_domain {
1159 /* These fields must be setup */
1160 struct sched_domain *parent; /* top domain must be null terminated */
1161 struct sched_domain *child; /* bottom domain must be null terminated */
1162 struct sched_group *groups; /* the balancing groups of the domain */
1163 unsigned long min_interval; /* Minimum balance interval ms */
1164 unsigned long max_interval; /* Maximum balance interval ms */
1165 unsigned int busy_factor; /* less balancing by factor if busy */
1166 unsigned int imbalance_pct; /* No balance until over watermark */
1167 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1168 unsigned int busy_idx;
1169 unsigned int idle_idx;
1170 unsigned int newidle_idx;
1171 unsigned int wake_idx;
1172 unsigned int forkexec_idx;
1173 unsigned int smt_gain;
1175 int nohz_idle; /* NOHZ IDLE status */
1176 int flags; /* See SD_* */
1179 /* Runtime fields. */
1180 unsigned long last_balance; /* init to jiffies. units in jiffies */
1181 unsigned int balance_interval; /* initialise to 1. units in ms. */
1182 unsigned int nr_balance_failed; /* initialise to 0 */
1184 /* idle_balance() stats */
1185 u64 max_newidle_lb_cost;
1186 unsigned long next_decay_max_lb_cost;
1188 u64 avg_scan_cost; /* select_idle_sibling */
1190 #ifdef CONFIG_SCHEDSTATS
1191 /* load_balance() stats */
1192 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1193 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1194 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1195 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1196 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1197 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1198 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1199 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1201 /* Active load balancing */
1202 unsigned int alb_count;
1203 unsigned int alb_failed;
1204 unsigned int alb_pushed;
1206 /* SD_BALANCE_EXEC stats */
1207 unsigned int sbe_count;
1208 unsigned int sbe_balanced;
1209 unsigned int sbe_pushed;
1211 /* SD_BALANCE_FORK stats */
1212 unsigned int sbf_count;
1213 unsigned int sbf_balanced;
1214 unsigned int sbf_pushed;
1216 /* try_to_wake_up() stats */
1217 unsigned int ttwu_wake_remote;
1218 unsigned int ttwu_move_affine;
1219 unsigned int ttwu_move_balance;
1221 struct eas_stats eas_stats;
1223 #ifdef CONFIG_SCHED_DEBUG
1227 void *private; /* used during construction */
1228 struct rcu_head rcu; /* used during destruction */
1230 struct sched_domain_shared *shared;
1232 unsigned int span_weight;
1234 * Span of all CPUs in this domain.
1236 * NOTE: this field is variable length. (Allocated dynamically
1237 * by attaching extra space to the end of the structure,
1238 * depending on how many CPUs the kernel has booted up with)
1240 unsigned long span[0];
1243 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1245 return to_cpumask(sd->span);
1248 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1249 struct sched_domain_attr *dattr_new);
1251 /* Allocate an array of sched domains, for partition_sched_domains(). */
1252 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1253 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1255 bool cpus_share_cache(int this_cpu, int that_cpu);
1257 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1258 typedef int (*sched_domain_flags_f)(void);
1260 const struct sched_group_energy * const(*sched_domain_energy_f)(int cpu);
1262 #define SDTL_OVERLAP 0x01
1265 struct sched_domain **__percpu sd;
1266 struct sched_domain_shared **__percpu sds;
1267 struct sched_group **__percpu sg;
1268 struct sched_group_capacity **__percpu sgc;
1271 struct sched_domain_topology_level {
1272 sched_domain_mask_f mask;
1273 sched_domain_flags_f sd_flags;
1274 sched_domain_energy_f energy;
1277 struct sd_data data;
1278 #ifdef CONFIG_SCHED_DEBUG
1283 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1284 extern void wake_up_if_idle(int cpu);
1286 #ifdef CONFIG_SCHED_DEBUG
1287 # define SD_INIT_NAME(type) .name = #type
1289 # define SD_INIT_NAME(type)
1292 #else /* CONFIG_SMP */
1294 struct sched_domain_attr;
1297 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1298 struct sched_domain_attr *dattr_new)
1302 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1307 #endif /* !CONFIG_SMP */
1310 struct io_context; /* See blkdev.h */
1313 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1314 extern void prefetch_stack(struct task_struct *t);
1316 static inline void prefetch_stack(struct task_struct *t) { }
1319 struct audit_context; /* See audit.c */
1321 struct pipe_inode_info;
1322 struct uts_namespace;
1324 struct load_weight {
1325 unsigned long weight;
1330 * The load_avg/util_avg accumulates an infinite geometric series
1331 * (see __update_load_avg() in kernel/sched/fair.c).
1333 * [load_avg definition]
1335 * load_avg = runnable% * scale_load_down(load)
1337 * where runnable% is the time ratio that a sched_entity is runnable.
1338 * For cfs_rq, it is the aggregated load_avg of all runnable and
1339 * blocked sched_entities.
1341 * load_avg may also take frequency scaling into account:
1343 * load_avg = runnable% * scale_load_down(load) * freq%
1345 * where freq% is the CPU frequency normalized to the highest frequency.
1347 * [util_avg definition]
1349 * util_avg = running% * SCHED_CAPACITY_SCALE
1351 * where running% is the time ratio that a sched_entity is running on
1352 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1353 * and blocked sched_entities.
1355 * util_avg may also factor frequency scaling and CPU capacity scaling:
1357 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1359 * where freq% is the same as above, and capacity% is the CPU capacity
1360 * normalized to the greatest capacity (due to uarch differences, etc).
1362 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1363 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1364 * we therefore scale them to as large a range as necessary. This is for
1365 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1369 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1370 * with the highest load (=88761), always runnable on a single cfs_rq,
1371 * and should not overflow as the number already hits PID_MAX_LIMIT.
1373 * For all other cases (including 32-bit kernels), struct load_weight's
1374 * weight will overflow first before we do, because:
1376 * Max(load_avg) <= Max(load.weight)
1378 * Then it is the load_weight's responsibility to consider overflow
1382 u64 last_update_time, load_sum;
1383 u32 util_sum, period_contrib;
1384 unsigned long load_avg, util_avg;
1387 #ifdef CONFIG_SCHEDSTATS
1388 struct sched_statistics {
1398 s64 sum_sleep_runtime;
1405 u64 nr_migrations_cold;
1406 u64 nr_failed_migrations_affine;
1407 u64 nr_failed_migrations_running;
1408 u64 nr_failed_migrations_hot;
1409 u64 nr_forced_migrations;
1412 u64 nr_wakeups_sync;
1413 u64 nr_wakeups_migrate;
1414 u64 nr_wakeups_local;
1415 u64 nr_wakeups_remote;
1416 u64 nr_wakeups_affine;
1417 u64 nr_wakeups_affine_attempts;
1418 u64 nr_wakeups_passive;
1419 u64 nr_wakeups_idle;
1421 /* select_idle_sibling() */
1422 u64 nr_wakeups_sis_attempts;
1423 u64 nr_wakeups_sis_idle;
1424 u64 nr_wakeups_sis_cache_affine;
1425 u64 nr_wakeups_sis_suff_cap;
1426 u64 nr_wakeups_sis_idle_cpu;
1427 u64 nr_wakeups_sis_count;
1429 /* energy_aware_wake_cpu() */
1430 u64 nr_wakeups_secb_attempts;
1431 u64 nr_wakeups_secb_sync;
1432 u64 nr_wakeups_secb_idle_bt;
1433 u64 nr_wakeups_secb_insuff_cap;
1434 u64 nr_wakeups_secb_no_nrg_sav;
1435 u64 nr_wakeups_secb_nrg_sav;
1436 u64 nr_wakeups_secb_count;
1438 /* find_best_target() */
1439 u64 nr_wakeups_fbt_attempts;
1440 u64 nr_wakeups_fbt_no_cpu;
1441 u64 nr_wakeups_fbt_no_sd;
1442 u64 nr_wakeups_fbt_pref_idle;
1443 u64 nr_wakeups_fbt_count;
1446 /* select_task_rq_fair() */
1447 u64 nr_wakeups_cas_attempts;
1448 u64 nr_wakeups_cas_count;
1452 #ifdef CONFIG_SCHED_WALT
1453 #define RAVG_HIST_SIZE_MAX 5
1455 /* ravg represents frequency scaled cpu-demand of tasks */
1458 * 'mark_start' marks the beginning of an event (task waking up, task
1459 * starting to execute, task being preempted) within a window
1461 * 'sum' represents how runnable a task has been within current
1462 * window. It incorporates both running time and wait time and is
1465 * 'sum_history' keeps track of history of 'sum' seen over previous
1466 * RAVG_HIST_SIZE windows. Windows where task was entirely sleeping are
1469 * 'demand' represents maximum sum seen over previous
1470 * sysctl_sched_ravg_hist_size windows. 'demand' could drive frequency
1473 * 'curr_window' represents task's contribution to cpu busy time
1474 * statistics (rq->curr_runnable_sum) in current window
1476 * 'prev_window' represents task's contribution to cpu busy time
1477 * statistics (rq->prev_runnable_sum) in previous window
1481 u32 sum_history[RAVG_HIST_SIZE_MAX];
1482 u32 curr_window, prev_window;
1487 struct sched_entity {
1488 struct load_weight load; /* for load-balancing */
1489 struct rb_node run_node;
1490 struct list_head group_node;
1494 u64 sum_exec_runtime;
1496 u64 prev_sum_exec_runtime;
1500 #ifdef CONFIG_SCHEDSTATS
1501 struct sched_statistics statistics;
1504 #ifdef CONFIG_FAIR_GROUP_SCHED
1506 struct sched_entity *parent;
1507 /* rq on which this entity is (to be) queued: */
1508 struct cfs_rq *cfs_rq;
1509 /* rq "owned" by this entity/group: */
1510 struct cfs_rq *my_q;
1515 * Per entity load average tracking.
1517 * Put into separate cache line so it does not
1518 * collide with read-mostly values above.
1520 struct sched_avg avg ____cacheline_aligned_in_smp;
1524 struct sched_rt_entity {
1525 struct list_head run_list;
1526 unsigned long timeout;
1527 unsigned long watchdog_stamp;
1528 unsigned int time_slice;
1529 unsigned short on_rq;
1530 unsigned short on_list;
1532 struct sched_rt_entity *back;
1533 #ifdef CONFIG_RT_GROUP_SCHED
1534 struct sched_rt_entity *parent;
1535 /* rq on which this entity is (to be) queued: */
1536 struct rt_rq *rt_rq;
1537 /* rq "owned" by this entity/group: */
1542 struct sched_dl_entity {
1543 struct rb_node rb_node;
1546 * Original scheduling parameters. Copied here from sched_attr
1547 * during sched_setattr(), they will remain the same until
1548 * the next sched_setattr().
1550 u64 dl_runtime; /* maximum runtime for each instance */
1551 u64 dl_deadline; /* relative deadline of each instance */
1552 u64 dl_period; /* separation of two instances (period) */
1553 u64 dl_bw; /* dl_runtime / dl_deadline */
1554 u64 dl_density; /* dl_runtime / dl_deadline */
1557 * Actual scheduling parameters. Initialized with the values above,
1558 * they are continously updated during task execution. Note that
1559 * the remaining runtime could be < 0 in case we are in overrun.
1561 s64 runtime; /* remaining runtime for this instance */
1562 u64 deadline; /* absolute deadline for this instance */
1563 unsigned int flags; /* specifying the scheduler behaviour */
1568 * @dl_throttled tells if we exhausted the runtime. If so, the
1569 * task has to wait for a replenishment to be performed at the
1570 * next firing of dl_timer.
1572 * @dl_boosted tells if we are boosted due to DI. If so we are
1573 * outside bandwidth enforcement mechanism (but only until we
1574 * exit the critical section);
1576 * @dl_yielded tells if task gave up the cpu before consuming
1577 * all its available runtime during the last job.
1579 int dl_throttled, dl_boosted, dl_yielded;
1582 * Bandwidth enforcement timer. Each -deadline task has its
1583 * own bandwidth to be enforced, thus we need one timer per task.
1585 struct hrtimer dl_timer;
1593 u8 pad; /* Otherwise the compiler can store garbage here. */
1595 u32 s; /* Set of bits. */
1599 enum perf_event_task_context {
1600 perf_invalid_context = -1,
1601 perf_hw_context = 0,
1603 perf_nr_task_contexts,
1606 /* Track pages that require TLB flushes */
1607 struct tlbflush_unmap_batch {
1609 * Each bit set is a CPU that potentially has a TLB entry for one of
1610 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1612 struct cpumask cpumask;
1614 /* True if any bit in cpumask is set */
1615 bool flush_required;
1618 * If true then the PTE was dirty when unmapped. The entry must be
1619 * flushed before IO is initiated or a stale TLB entry potentially
1620 * allows an update without redirtying the page.
1625 struct task_struct {
1626 #ifdef CONFIG_THREAD_INFO_IN_TASK
1628 * For reasons of header soup (see current_thread_info()), this
1629 * must be the first element of task_struct.
1631 struct thread_info thread_info;
1633 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1636 unsigned int flags; /* per process flags, defined below */
1637 unsigned int ptrace;
1640 struct llist_node wake_entry;
1642 #ifdef CONFIG_THREAD_INFO_IN_TASK
1643 unsigned int cpu; /* current CPU */
1645 unsigned int wakee_flips;
1646 unsigned long wakee_flip_decay_ts;
1647 struct task_struct *last_wakee;
1653 int prio, static_prio, normal_prio;
1654 unsigned int rt_priority;
1655 const struct sched_class *sched_class;
1656 struct sched_entity se;
1657 struct sched_rt_entity rt;
1658 #ifdef CONFIG_SCHED_WALT
1661 * 'init_load_pct' represents the initial task load assigned to children
1668 #ifdef CONFIG_CGROUP_SCHED
1669 struct task_group *sched_task_group;
1671 struct sched_dl_entity dl;
1673 #ifdef CONFIG_PREEMPT_NOTIFIERS
1674 /* list of struct preempt_notifier: */
1675 struct hlist_head preempt_notifiers;
1678 #ifdef CONFIG_BLK_DEV_IO_TRACE
1679 unsigned int btrace_seq;
1682 unsigned int policy;
1683 int nr_cpus_allowed;
1684 cpumask_t cpus_allowed;
1686 #ifdef CONFIG_PREEMPT_RCU
1687 int rcu_read_lock_nesting;
1688 union rcu_special rcu_read_unlock_special;
1689 struct list_head rcu_node_entry;
1690 struct rcu_node *rcu_blocked_node;
1691 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1692 #ifdef CONFIG_TASKS_RCU
1693 unsigned long rcu_tasks_nvcsw;
1694 bool rcu_tasks_holdout;
1695 struct list_head rcu_tasks_holdout_list;
1696 int rcu_tasks_idle_cpu;
1697 #endif /* #ifdef CONFIG_TASKS_RCU */
1699 #ifdef CONFIG_SCHED_INFO
1700 struct sched_info sched_info;
1703 struct list_head tasks;
1705 struct plist_node pushable_tasks;
1706 struct rb_node pushable_dl_tasks;
1709 struct mm_struct *mm, *active_mm;
1710 /* per-thread vma caching */
1711 u64 vmacache_seqnum;
1712 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1713 #if defined(SPLIT_RSS_COUNTING)
1714 struct task_rss_stat rss_stat;
1718 int exit_code, exit_signal;
1719 int pdeath_signal; /* The signal sent when the parent dies */
1720 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1722 /* Used for emulating ABI behavior of previous Linux versions */
1723 unsigned int personality;
1725 /* scheduler bits, serialized by scheduler locks */
1726 unsigned sched_reset_on_fork:1;
1727 unsigned sched_contributes_to_load:1;
1728 unsigned sched_migrated:1;
1729 unsigned sched_remote_wakeup:1;
1730 unsigned :0; /* force alignment to the next boundary */
1732 /* unserialized, strictly 'current' */
1733 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1734 unsigned in_iowait:1;
1735 #if !defined(TIF_RESTORE_SIGMASK)
1736 unsigned restore_sigmask:1;
1739 unsigned memcg_may_oom:1;
1741 unsigned memcg_kmem_skip_account:1;
1744 #ifdef CONFIG_COMPAT_BRK
1745 unsigned brk_randomized:1;
1747 #ifdef CONFIG_CGROUPS
1748 /* disallow userland-initiated cgroup migration */
1749 unsigned no_cgroup_migration:1;
1752 unsigned long atomic_flags; /* Flags needing atomic access. */
1754 struct restart_block restart_block;
1759 #ifdef CONFIG_CC_STACKPROTECTOR
1760 /* Canary value for the -fstack-protector gcc feature */
1761 unsigned long stack_canary;
1764 * pointers to (original) parent process, youngest child, younger sibling,
1765 * older sibling, respectively. (p->father can be replaced with
1766 * p->real_parent->pid)
1768 struct task_struct __rcu *real_parent; /* real parent process */
1769 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1771 * children/sibling forms the list of my natural children
1773 struct list_head children; /* list of my children */
1774 struct list_head sibling; /* linkage in my parent's children list */
1775 struct task_struct *group_leader; /* threadgroup leader */
1778 * ptraced is the list of tasks this task is using ptrace on.
1779 * This includes both natural children and PTRACE_ATTACH targets.
1780 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1782 struct list_head ptraced;
1783 struct list_head ptrace_entry;
1785 /* PID/PID hash table linkage. */
1786 struct pid_link pids[PIDTYPE_MAX];
1787 struct list_head thread_group;
1788 struct list_head thread_node;
1790 struct completion *vfork_done; /* for vfork() */
1791 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1792 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1794 cputime_t utime, stime, utimescaled, stimescaled;
1796 #ifdef CONFIG_CPU_FREQ_TIMES
1798 unsigned int max_state;
1800 struct prev_cputime prev_cputime;
1801 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1802 seqcount_t vtime_seqcount;
1803 unsigned long long vtime_snap;
1805 /* Task is sleeping or running in a CPU with VTIME inactive */
1807 /* Task runs in userspace in a CPU with VTIME active */
1809 /* Task runs in kernelspace in a CPU with VTIME active */
1811 } vtime_snap_whence;
1814 #ifdef CONFIG_NO_HZ_FULL
1815 atomic_t tick_dep_mask;
1817 unsigned long nvcsw, nivcsw; /* context switch counts */
1818 u64 start_time; /* monotonic time in nsec */
1819 u64 real_start_time; /* boot based time in nsec */
1820 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1821 unsigned long min_flt, maj_flt;
1823 struct task_cputime cputime_expires;
1824 struct list_head cpu_timers[3];
1826 /* process credentials */
1827 const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
1828 const struct cred __rcu *real_cred; /* objective and real subjective task
1829 * credentials (COW) */
1830 const struct cred __rcu *cred; /* effective (overridable) subjective task
1831 * credentials (COW) */
1832 char comm[TASK_COMM_LEN]; /* executable name excluding path
1833 - access with [gs]et_task_comm (which lock
1834 it with task_lock())
1835 - initialized normally by setup_new_exec */
1836 /* file system info */
1837 struct nameidata *nameidata;
1838 #ifdef CONFIG_SYSVIPC
1840 struct sysv_sem sysvsem;
1841 struct sysv_shm sysvshm;
1843 #ifdef CONFIG_DETECT_HUNG_TASK
1844 /* hung task detection */
1845 unsigned long last_switch_count;
1847 /* filesystem information */
1848 struct fs_struct *fs;
1849 /* open file information */
1850 struct files_struct *files;
1852 struct nsproxy *nsproxy;
1853 /* signal handlers */
1854 struct signal_struct *signal;
1855 struct sighand_struct *sighand;
1857 sigset_t blocked, real_blocked;
1858 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1859 struct sigpending pending;
1861 unsigned long sas_ss_sp;
1863 unsigned sas_ss_flags;
1865 struct callback_head *task_works;
1867 struct audit_context *audit_context;
1868 #ifdef CONFIG_AUDITSYSCALL
1870 unsigned int sessionid;
1872 struct seccomp seccomp;
1874 /* Thread group tracking */
1877 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1879 spinlock_t alloc_lock;
1881 /* Protection of the PI data structures: */
1882 raw_spinlock_t pi_lock;
1884 struct wake_q_node wake_q;
1886 #ifdef CONFIG_RT_MUTEXES
1887 /* PI waiters blocked on a rt_mutex held by this task */
1888 struct rb_root pi_waiters;
1889 struct rb_node *pi_waiters_leftmost;
1890 /* Deadlock detection and priority inheritance handling */
1891 struct rt_mutex_waiter *pi_blocked_on;
1894 #ifdef CONFIG_DEBUG_MUTEXES
1895 /* mutex deadlock detection */
1896 struct mutex_waiter *blocked_on;
1898 #ifdef CONFIG_TRACE_IRQFLAGS
1899 unsigned int irq_events;
1900 unsigned long hardirq_enable_ip;
1901 unsigned long hardirq_disable_ip;
1902 unsigned int hardirq_enable_event;
1903 unsigned int hardirq_disable_event;
1904 int hardirqs_enabled;
1905 int hardirq_context;
1906 unsigned long softirq_disable_ip;
1907 unsigned long softirq_enable_ip;
1908 unsigned int softirq_disable_event;
1909 unsigned int softirq_enable_event;
1910 int softirqs_enabled;
1911 int softirq_context;
1913 #ifdef CONFIG_LOCKDEP
1914 # define MAX_LOCK_DEPTH 48UL
1917 unsigned int lockdep_recursion;
1918 struct held_lock held_locks[MAX_LOCK_DEPTH];
1919 gfp_t lockdep_reclaim_gfp;
1922 unsigned int in_ubsan;
1925 /* journalling filesystem info */
1928 /* stacked block device info */
1929 struct bio_list *bio_list;
1932 /* stack plugging */
1933 struct blk_plug *plug;
1937 struct reclaim_state *reclaim_state;
1939 struct backing_dev_info *backing_dev_info;
1941 struct io_context *io_context;
1943 unsigned long ptrace_message;
1944 siginfo_t *last_siginfo; /* For ptrace use. */
1945 struct task_io_accounting ioac;
1946 #if defined(CONFIG_TASK_XACCT)
1947 u64 acct_rss_mem1; /* accumulated rss usage */
1948 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1949 cputime_t acct_timexpd; /* stime + utime since last update */
1951 #ifdef CONFIG_CPUSETS
1952 nodemask_t mems_allowed; /* Protected by alloc_lock */
1953 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1954 int cpuset_mem_spread_rotor;
1955 int cpuset_slab_spread_rotor;
1957 #ifdef CONFIG_CGROUPS
1958 /* Control Group info protected by css_set_lock */
1959 struct css_set __rcu *cgroups;
1960 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1961 struct list_head cg_list;
1964 struct robust_list_head __user *robust_list;
1965 #ifdef CONFIG_COMPAT
1966 struct compat_robust_list_head __user *compat_robust_list;
1968 struct list_head pi_state_list;
1969 struct futex_pi_state *pi_state_cache;
1971 #ifdef CONFIG_PERF_EVENTS
1972 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1973 struct mutex perf_event_mutex;
1974 struct list_head perf_event_list;
1976 #ifdef CONFIG_DEBUG_PREEMPT
1977 unsigned long preempt_disable_ip;
1980 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1982 short pref_node_fork;
1984 #ifdef CONFIG_NUMA_BALANCING
1986 unsigned int numa_scan_period;
1987 unsigned int numa_scan_period_max;
1988 int numa_preferred_nid;
1989 unsigned long numa_migrate_retry;
1990 u64 node_stamp; /* migration stamp */
1991 u64 last_task_numa_placement;
1992 u64 last_sum_exec_runtime;
1993 struct callback_head numa_work;
1995 struct list_head numa_entry;
1996 struct numa_group *numa_group;
1999 * numa_faults is an array split into four regions:
2000 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
2001 * in this precise order.
2003 * faults_memory: Exponential decaying average of faults on a per-node
2004 * basis. Scheduling placement decisions are made based on these
2005 * counts. The values remain static for the duration of a PTE scan.
2006 * faults_cpu: Track the nodes the process was running on when a NUMA
2007 * hinting fault was incurred.
2008 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
2009 * during the current scan window. When the scan completes, the counts
2010 * in faults_memory and faults_cpu decay and these values are copied.
2012 unsigned long *numa_faults;
2013 unsigned long total_numa_faults;
2016 * numa_faults_locality tracks if faults recorded during the last
2017 * scan window were remote/local or failed to migrate. The task scan
2018 * period is adapted based on the locality of the faults with different
2019 * weights depending on whether they were shared or private faults
2021 unsigned long numa_faults_locality[3];
2023 unsigned long numa_pages_migrated;
2024 #endif /* CONFIG_NUMA_BALANCING */
2026 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
2027 struct tlbflush_unmap_batch tlb_ubc;
2030 struct rcu_head rcu;
2033 * cache last used pipe for splice
2035 struct pipe_inode_info *splice_pipe;
2037 struct page_frag task_frag;
2039 #ifdef CONFIG_TASK_DELAY_ACCT
2040 struct task_delay_info *delays;
2042 #ifdef CONFIG_FAULT_INJECTION
2046 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
2047 * balance_dirty_pages() for some dirty throttling pause
2050 int nr_dirtied_pause;
2051 unsigned long dirty_paused_when; /* start of a write-and-pause period */
2053 #ifdef CONFIG_LATENCYTOP
2054 int latency_record_count;
2055 struct latency_record latency_record[LT_SAVECOUNT];
2058 * time slack values; these are used to round up poll() and
2059 * select() etc timeout values. These are in nanoseconds.
2062 u64 default_timer_slack_ns;
2065 unsigned int kasan_depth;
2067 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
2068 /* Index of current stored address in ret_stack */
2070 /* Stack of return addresses for return function tracing */
2071 struct ftrace_ret_stack *ret_stack;
2072 /* time stamp for last schedule */
2073 unsigned long long ftrace_timestamp;
2075 * Number of functions that haven't been traced
2076 * because of depth overrun.
2078 atomic_t trace_overrun;
2079 /* Pause for the tracing */
2080 atomic_t tracing_graph_pause;
2082 #ifdef CONFIG_TRACING
2083 /* state flags for use by tracers */
2084 unsigned long trace;
2085 /* bitmask and counter of trace recursion */
2086 unsigned long trace_recursion;
2087 #endif /* CONFIG_TRACING */
2089 /* Coverage collection mode enabled for this task (0 if disabled). */
2090 enum kcov_mode kcov_mode;
2091 /* Size of the kcov_area. */
2093 /* Buffer for coverage collection. */
2095 /* kcov desciptor wired with this task or NULL. */
2099 struct mem_cgroup *memcg_in_oom;
2100 gfp_t memcg_oom_gfp_mask;
2101 int memcg_oom_order;
2103 /* number of pages to reclaim on returning to userland */
2104 unsigned int memcg_nr_pages_over_high;
2106 #ifdef CONFIG_UPROBES
2107 struct uprobe_task *utask;
2109 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
2110 unsigned int sequential_io;
2111 unsigned int sequential_io_avg;
2113 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
2114 unsigned long task_state_change;
2116 int pagefault_disabled;
2118 struct task_struct *oom_reaper_list;
2120 #ifdef CONFIG_VMAP_STACK
2121 struct vm_struct *stack_vm_area;
2123 #ifdef CONFIG_THREAD_INFO_IN_TASK
2124 /* A live task holds one reference. */
2125 atomic_t stack_refcount;
2127 /* CPU-specific state of this task */
2128 struct thread_struct thread;
2130 * WARNING: on x86, 'thread_struct' contains a variable-sized
2131 * structure. It *MUST* be at the end of 'task_struct'.
2133 * Do not put anything below here!
2137 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
2138 extern int arch_task_struct_size __read_mostly;
2140 # define arch_task_struct_size (sizeof(struct task_struct))
2143 #ifdef CONFIG_VMAP_STACK
2144 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
2146 return t->stack_vm_area;
2149 static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t)
2155 /* Future-safe accessor for struct task_struct's cpus_allowed. */
2156 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
2158 static inline int tsk_nr_cpus_allowed(struct task_struct *p)
2160 return p->nr_cpus_allowed;
2163 #define TNF_MIGRATED 0x01
2164 #define TNF_NO_GROUP 0x02
2165 #define TNF_SHARED 0x04
2166 #define TNF_FAULT_LOCAL 0x08
2167 #define TNF_MIGRATE_FAIL 0x10
2169 static inline bool in_vfork(struct task_struct *tsk)
2174 * need RCU to access ->real_parent if CLONE_VM was used along with
2177 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
2180 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
2181 * ->real_parent is not necessarily the task doing vfork(), so in
2182 * theory we can't rely on task_lock() if we want to dereference it.
2184 * And in this case we can't trust the real_parent->mm == tsk->mm
2185 * check, it can be false negative. But we do not care, if init or
2186 * another oom-unkillable task does this it should blame itself.
2189 ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
2195 #ifdef CONFIG_NUMA_BALANCING
2196 extern void task_numa_fault(int last_node, int node, int pages, int flags);
2197 extern pid_t task_numa_group_id(struct task_struct *p);
2198 extern void set_numabalancing_state(bool enabled);
2199 extern void task_numa_free(struct task_struct *p);
2200 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
2201 int src_nid, int dst_cpu);
2203 static inline void task_numa_fault(int last_node, int node, int pages,
2207 static inline pid_t task_numa_group_id(struct task_struct *p)
2211 static inline void set_numabalancing_state(bool enabled)
2214 static inline void task_numa_free(struct task_struct *p)
2217 static inline bool should_numa_migrate_memory(struct task_struct *p,
2218 struct page *page, int src_nid, int dst_cpu)
2224 static inline struct pid *task_pid(struct task_struct *task)
2226 return task->pids[PIDTYPE_PID].pid;
2229 static inline struct pid *task_tgid(struct task_struct *task)
2231 return task->group_leader->pids[PIDTYPE_PID].pid;
2235 * Without tasklist or rcu lock it is not safe to dereference
2236 * the result of task_pgrp/task_session even if task == current,
2237 * we can race with another thread doing sys_setsid/sys_setpgid.
2239 static inline struct pid *task_pgrp(struct task_struct *task)
2241 return task->group_leader->pids[PIDTYPE_PGID].pid;
2244 static inline struct pid *task_session(struct task_struct *task)
2246 return task->group_leader->pids[PIDTYPE_SID].pid;
2249 struct pid_namespace;
2252 * the helpers to get the task's different pids as they are seen
2253 * from various namespaces
2255 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
2256 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2258 * task_xid_nr_ns() : id seen from the ns specified;
2260 * set_task_vxid() : assigns a virtual id to a task;
2262 * see also pid_nr() etc in include/linux/pid.h
2264 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
2265 struct pid_namespace *ns);
2267 static inline pid_t task_pid_nr(struct task_struct *tsk)
2272 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
2273 struct pid_namespace *ns)
2275 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
2278 static inline pid_t task_pid_vnr(struct task_struct *tsk)
2280 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
2284 static inline pid_t task_tgid_nr(struct task_struct *tsk)
2290 static inline int pid_alive(const struct task_struct *p);
2292 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2293 struct pid_namespace *ns)
2295 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2298 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2300 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2304 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2305 struct pid_namespace *ns)
2307 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2310 static inline pid_t task_session_vnr(struct task_struct *tsk)
2312 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2315 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
2317 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, ns);
2320 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
2322 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, NULL);
2325 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
2331 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
2337 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
2339 return task_ppid_nr_ns(tsk, &init_pid_ns);
2342 /* obsolete, do not use */
2343 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2345 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2349 * pid_alive - check that a task structure is not stale
2350 * @p: Task structure to be checked.
2352 * Test if a process is not yet dead (at most zombie state)
2353 * If pid_alive fails, then pointers within the task structure
2354 * can be stale and must not be dereferenced.
2356 * Return: 1 if the process is alive. 0 otherwise.
2358 static inline int pid_alive(const struct task_struct *p)
2360 return p->pids[PIDTYPE_PID].pid != NULL;
2364 * is_global_init - check if a task structure is init. Since init
2365 * is free to have sub-threads we need to check tgid.
2366 * @tsk: Task structure to be checked.
2368 * Check if a task structure is the first user space task the kernel created.
2370 * Return: 1 if the task structure is init. 0 otherwise.
2372 static inline int is_global_init(struct task_struct *tsk)
2374 return task_tgid_nr(tsk) == 1;
2377 extern struct pid *cad_pid;
2379 extern void free_task(struct task_struct *tsk);
2380 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2382 extern void __put_task_struct(struct task_struct *t);
2384 static inline void put_task_struct(struct task_struct *t)
2386 if (atomic_dec_and_test(&t->usage))
2387 __put_task_struct(t);
2390 struct task_struct *task_rcu_dereference(struct task_struct **ptask);
2391 struct task_struct *try_get_task_struct(struct task_struct **ptask);
2393 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2394 extern void task_cputime(struct task_struct *t,
2395 cputime_t *utime, cputime_t *stime);
2396 extern void task_cputime_scaled(struct task_struct *t,
2397 cputime_t *utimescaled, cputime_t *stimescaled);
2398 extern cputime_t task_gtime(struct task_struct *t);
2400 static inline void task_cputime(struct task_struct *t,
2401 cputime_t *utime, cputime_t *stime)
2409 static inline void task_cputime_scaled(struct task_struct *t,
2410 cputime_t *utimescaled,
2411 cputime_t *stimescaled)
2414 *utimescaled = t->utimescaled;
2416 *stimescaled = t->stimescaled;
2419 static inline cputime_t task_gtime(struct task_struct *t)
2424 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2425 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2430 #define PF_EXITING 0x00000004 /* getting shut down */
2431 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2432 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2433 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2434 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2435 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2436 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2437 #define PF_DUMPCORE 0x00000200 /* dumped core */
2438 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2439 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2440 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2441 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2442 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2443 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2444 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2445 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2446 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2447 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2448 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2449 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2450 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2451 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2452 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2453 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2454 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2455 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2456 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2459 * Only the _current_ task can read/write to tsk->flags, but other
2460 * tasks can access tsk->flags in readonly mode for example
2461 * with tsk_used_math (like during threaded core dumping).
2462 * There is however an exception to this rule during ptrace
2463 * or during fork: the ptracer task is allowed to write to the
2464 * child->flags of its traced child (same goes for fork, the parent
2465 * can write to the child->flags), because we're guaranteed the
2466 * child is not running and in turn not changing child->flags
2467 * at the same time the parent does it.
2469 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2470 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2471 #define clear_used_math() clear_stopped_child_used_math(current)
2472 #define set_used_math() set_stopped_child_used_math(current)
2473 #define conditional_stopped_child_used_math(condition, child) \
2474 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2475 #define conditional_used_math(condition) \
2476 conditional_stopped_child_used_math(condition, current)
2477 #define copy_to_stopped_child_used_math(child) \
2478 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2479 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2480 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2481 #define used_math() tsk_used_math(current)
2483 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2484 * __GFP_FS is also cleared as it implies __GFP_IO.
2486 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2488 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2489 flags &= ~(__GFP_IO | __GFP_FS);
2493 static inline unsigned int memalloc_noio_save(void)
2495 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2496 current->flags |= PF_MEMALLOC_NOIO;
2500 static inline void memalloc_noio_restore(unsigned int flags)
2502 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2505 /* Per-process atomic flags. */
2506 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2507 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2508 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2509 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
2510 #define PFA_SPEC_SSB_DISABLE 4 /* Speculative Store Bypass disabled */
2511 #define PFA_SPEC_SSB_FORCE_DISABLE 5 /* Speculative Store Bypass force disabled*/
2514 #define TASK_PFA_TEST(name, func) \
2515 static inline bool task_##func(struct task_struct *p) \
2516 { return test_bit(PFA_##name, &p->atomic_flags); }
2517 #define TASK_PFA_SET(name, func) \
2518 static inline void task_set_##func(struct task_struct *p) \
2519 { set_bit(PFA_##name, &p->atomic_flags); }
2520 #define TASK_PFA_CLEAR(name, func) \
2521 static inline void task_clear_##func(struct task_struct *p) \
2522 { clear_bit(PFA_##name, &p->atomic_flags); }
2524 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2525 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2527 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2528 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2529 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2531 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2532 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2533 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2535 TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
2536 TASK_PFA_SET(LMK_WAITING, lmk_waiting)
2538 TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable)
2539 TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable)
2540 TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable)
2542 TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
2543 TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable)
2546 * task->jobctl flags
2548 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2550 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2551 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2552 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2553 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2554 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2555 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2556 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2558 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2559 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2560 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2561 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2562 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2563 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2564 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2566 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2567 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2569 extern bool task_set_jobctl_pending(struct task_struct *task,
2570 unsigned long mask);
2571 extern void task_clear_jobctl_trapping(struct task_struct *task);
2572 extern void task_clear_jobctl_pending(struct task_struct *task,
2573 unsigned long mask);
2575 static inline void rcu_copy_process(struct task_struct *p)
2577 #ifdef CONFIG_PREEMPT_RCU
2578 p->rcu_read_lock_nesting = 0;
2579 p->rcu_read_unlock_special.s = 0;
2580 p->rcu_blocked_node = NULL;
2581 INIT_LIST_HEAD(&p->rcu_node_entry);
2582 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2583 #ifdef CONFIG_TASKS_RCU
2584 p->rcu_tasks_holdout = false;
2585 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2586 p->rcu_tasks_idle_cpu = -1;
2587 #endif /* #ifdef CONFIG_TASKS_RCU */
2590 static inline void tsk_restore_flags(struct task_struct *task,
2591 unsigned long orig_flags, unsigned long flags)
2593 task->flags &= ~flags;
2594 task->flags |= orig_flags & flags;
2597 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2598 const struct cpumask *trial);
2599 extern int task_can_attach(struct task_struct *p,
2600 const struct cpumask *cs_cpus_allowed);
2602 extern void do_set_cpus_allowed(struct task_struct *p,
2603 const struct cpumask *new_mask);
2605 extern int set_cpus_allowed_ptr(struct task_struct *p,
2606 const struct cpumask *new_mask);
2608 static inline void do_set_cpus_allowed(struct task_struct *p,
2609 const struct cpumask *new_mask)
2612 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2613 const struct cpumask *new_mask)
2615 if (!cpumask_test_cpu(0, new_mask))
2621 #ifdef CONFIG_NO_HZ_COMMON
2622 void calc_load_enter_idle(void);
2623 void calc_load_exit_idle(void);
2625 static inline void calc_load_enter_idle(void) { }
2626 static inline void calc_load_exit_idle(void) { }
2627 #endif /* CONFIG_NO_HZ_COMMON */
2630 * Do not use outside of architecture code which knows its limitations.
2632 * sched_clock() has no promise of monotonicity or bounded drift between
2633 * CPUs, use (which you should not) requires disabling IRQs.
2635 * Please use one of the three interfaces below.
2637 extern unsigned long long notrace sched_clock(void);
2639 * See the comment in kernel/sched/clock.c
2641 extern u64 running_clock(void);
2642 extern u64 sched_clock_cpu(int cpu);
2645 extern void sched_clock_init(void);
2647 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2648 static inline void sched_clock_tick(void)
2652 static inline void sched_clock_idle_sleep_event(void)
2656 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2660 static inline u64 cpu_clock(int cpu)
2662 return sched_clock();
2665 static inline u64 local_clock(void)
2667 return sched_clock();
2671 * Architectures can set this to 1 if they have specified
2672 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2673 * but then during bootup it turns out that sched_clock()
2674 * is reliable after all:
2676 extern int sched_clock_stable(void);
2677 extern void set_sched_clock_stable(void);
2678 extern void clear_sched_clock_stable(void);
2680 extern void sched_clock_tick(void);
2681 extern void sched_clock_idle_sleep_event(void);
2682 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2685 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2686 * time source that is monotonic per cpu argument and has bounded drift
2689 * ######################### BIG FAT WARNING ##########################
2690 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2691 * # go backwards !! #
2692 * ####################################################################
2694 static inline u64 cpu_clock(int cpu)
2696 return sched_clock_cpu(cpu);
2699 static inline u64 local_clock(void)
2701 return sched_clock_cpu(raw_smp_processor_id());
2705 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2707 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2708 * The reason for this explicit opt-in is not to have perf penalty with
2709 * slow sched_clocks.
2711 extern void enable_sched_clock_irqtime(void);
2712 extern void disable_sched_clock_irqtime(void);
2714 static inline void enable_sched_clock_irqtime(void) {}
2715 static inline void disable_sched_clock_irqtime(void) {}
2718 extern unsigned long long
2719 task_sched_runtime(struct task_struct *task);
2721 /* sched_exec is called by processes performing an exec */
2723 extern void sched_exec(void);
2725 #define sched_exec() {}
2728 extern void sched_clock_idle_sleep_event(void);
2729 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2731 #ifdef CONFIG_HOTPLUG_CPU
2732 extern void idle_task_exit(void);
2734 static inline void idle_task_exit(void) {}
2737 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2738 extern void wake_up_nohz_cpu(int cpu);
2740 static inline void wake_up_nohz_cpu(int cpu) { }
2743 #ifdef CONFIG_NO_HZ_FULL
2744 extern u64 scheduler_tick_max_deferment(void);
2747 #ifdef CONFIG_SCHED_AUTOGROUP
2748 extern void sched_autogroup_create_attach(struct task_struct *p);
2749 extern void sched_autogroup_detach(struct task_struct *p);
2750 extern void sched_autogroup_fork(struct signal_struct *sig);
2751 extern void sched_autogroup_exit(struct signal_struct *sig);
2752 extern void sched_autogroup_exit_task(struct task_struct *p);
2753 #ifdef CONFIG_PROC_FS
2754 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2755 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2758 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2759 static inline void sched_autogroup_detach(struct task_struct *p) { }
2760 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2761 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2762 static inline void sched_autogroup_exit_task(struct task_struct *p) { }
2765 extern int yield_to(struct task_struct *p, bool preempt);
2766 extern void set_user_nice(struct task_struct *p, long nice);
2767 extern int task_prio(const struct task_struct *p);
2769 * task_nice - return the nice value of a given task.
2770 * @p: the task in question.
2772 * Return: The nice value [ -20 ... 0 ... 19 ].
2774 static inline int task_nice(const struct task_struct *p)
2776 return PRIO_TO_NICE((p)->static_prio);
2778 extern int can_nice(const struct task_struct *p, const int nice);
2779 extern int task_curr(const struct task_struct *p);
2780 extern int idle_cpu(int cpu);
2781 extern int sched_setscheduler(struct task_struct *, int,
2782 const struct sched_param *);
2783 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2784 const struct sched_param *);
2785 extern int sched_setattr(struct task_struct *,
2786 const struct sched_attr *);
2787 extern struct task_struct *idle_task(int cpu);
2789 * is_idle_task - is the specified task an idle task?
2790 * @p: the task in question.
2792 * Return: 1 if @p is an idle task. 0 otherwise.
2794 static inline bool is_idle_task(const struct task_struct *p)
2798 extern struct task_struct *curr_task(int cpu);
2799 extern void ia64_set_curr_task(int cpu, struct task_struct *p);
2803 union thread_union {
2804 #ifndef CONFIG_THREAD_INFO_IN_TASK
2805 struct thread_info thread_info;
2807 unsigned long stack[THREAD_SIZE/sizeof(long)];
2810 #ifndef __HAVE_ARCH_KSTACK_END
2811 static inline int kstack_end(void *addr)
2813 /* Reliable end of stack detection:
2814 * Some APM bios versions misalign the stack
2816 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2820 extern union thread_union init_thread_union;
2821 extern struct task_struct init_task;
2823 extern struct mm_struct init_mm;
2825 extern struct pid_namespace init_pid_ns;
2828 * find a task by one of its numerical ids
2830 * find_task_by_pid_ns():
2831 * finds a task by its pid in the specified namespace
2832 * find_task_by_vpid():
2833 * finds a task by its virtual pid
2835 * see also find_vpid() etc in include/linux/pid.h
2838 extern struct task_struct *find_task_by_vpid(pid_t nr);
2839 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2840 struct pid_namespace *ns);
2842 /* per-UID process charging. */
2843 extern struct user_struct * alloc_uid(kuid_t);
2844 static inline struct user_struct *get_uid(struct user_struct *u)
2846 atomic_inc(&u->__count);
2849 extern void free_uid(struct user_struct *);
2851 #include <asm/current.h>
2853 extern void xtime_update(unsigned long ticks);
2855 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2856 extern int wake_up_process(struct task_struct *tsk);
2857 extern void wake_up_new_task(struct task_struct *tsk);
2859 extern void kick_process(struct task_struct *tsk);
2861 static inline void kick_process(struct task_struct *tsk) { }
2863 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2864 extern void sched_dead(struct task_struct *p);
2866 extern void proc_caches_init(void);
2867 extern void flush_signals(struct task_struct *);
2868 extern void ignore_signals(struct task_struct *);
2869 extern void flush_signal_handlers(struct task_struct *, int force_default);
2870 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2872 static inline int kernel_dequeue_signal(siginfo_t *info)
2874 struct task_struct *tsk = current;
2878 spin_lock_irq(&tsk->sighand->siglock);
2879 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2880 spin_unlock_irq(&tsk->sighand->siglock);
2885 static inline void kernel_signal_stop(void)
2887 spin_lock_irq(¤t->sighand->siglock);
2888 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2889 __set_current_state(TASK_STOPPED);
2890 spin_unlock_irq(¤t->sighand->siglock);
2895 extern void release_task(struct task_struct * p);
2896 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2897 extern int force_sigsegv(int, struct task_struct *);
2898 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2899 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2900 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2901 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2902 const struct cred *, u32);
2903 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2904 extern int kill_pid(struct pid *pid, int sig, int priv);
2905 extern int kill_proc_info(int, struct siginfo *, pid_t);
2906 extern __must_check bool do_notify_parent(struct task_struct *, int);
2907 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2908 extern void force_sig(int, struct task_struct *);
2909 extern int send_sig(int, struct task_struct *, int);
2910 extern int zap_other_threads(struct task_struct *p);
2911 extern struct sigqueue *sigqueue_alloc(void);
2912 extern void sigqueue_free(struct sigqueue *);
2913 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2914 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2916 #ifdef TIF_RESTORE_SIGMASK
2918 * Legacy restore_sigmask accessors. These are inefficient on
2919 * SMP architectures because they require atomic operations.
2923 * set_restore_sigmask() - make sure saved_sigmask processing gets done
2925 * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2926 * will run before returning to user mode, to process the flag. For
2927 * all callers, TIF_SIGPENDING is already set or it's no harm to set
2928 * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
2929 * arch code will notice on return to user mode, in case those bits
2930 * are scarce. We set TIF_SIGPENDING here to ensure that the arch
2931 * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2933 static inline void set_restore_sigmask(void)
2935 set_thread_flag(TIF_RESTORE_SIGMASK);
2936 WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2938 static inline void clear_restore_sigmask(void)
2940 clear_thread_flag(TIF_RESTORE_SIGMASK);
2942 static inline bool test_restore_sigmask(void)
2944 return test_thread_flag(TIF_RESTORE_SIGMASK);
2946 static inline bool test_and_clear_restore_sigmask(void)
2948 return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK);
2951 #else /* TIF_RESTORE_SIGMASK */
2953 /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2954 static inline void set_restore_sigmask(void)
2956 current->restore_sigmask = true;
2957 WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2959 static inline void clear_restore_sigmask(void)
2961 current->restore_sigmask = false;
2963 static inline bool test_restore_sigmask(void)
2965 return current->restore_sigmask;
2967 static inline bool test_and_clear_restore_sigmask(void)
2969 if (!current->restore_sigmask)
2971 current->restore_sigmask = false;
2976 static inline void restore_saved_sigmask(void)
2978 if (test_and_clear_restore_sigmask())
2979 __set_current_blocked(¤t->saved_sigmask);
2982 static inline sigset_t *sigmask_to_save(void)
2984 sigset_t *res = ¤t->blocked;
2985 if (unlikely(test_restore_sigmask()))
2986 res = ¤t->saved_sigmask;
2990 static inline int kill_cad_pid(int sig, int priv)
2992 return kill_pid(cad_pid, sig, priv);
2995 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2996 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2997 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2998 #define SEND_SIG_FORCED ((struct siginfo *) 2)
3001 * True if we are on the alternate signal stack.
3003 static inline int on_sig_stack(unsigned long sp)
3006 * If the signal stack is SS_AUTODISARM then, by construction, we
3007 * can't be on the signal stack unless user code deliberately set
3008 * SS_AUTODISARM when we were already on it.
3010 * This improves reliability: if user state gets corrupted such that
3011 * the stack pointer points very close to the end of the signal stack,
3012 * then this check will enable the signal to be handled anyway.
3014 if (current->sas_ss_flags & SS_AUTODISARM)
3017 #ifdef CONFIG_STACK_GROWSUP
3018 return sp >= current->sas_ss_sp &&
3019 sp - current->sas_ss_sp < current->sas_ss_size;
3021 return sp > current->sas_ss_sp &&
3022 sp - current->sas_ss_sp <= current->sas_ss_size;
3026 static inline int sas_ss_flags(unsigned long sp)
3028 if (!current->sas_ss_size)
3031 return on_sig_stack(sp) ? SS_ONSTACK : 0;
3034 static inline void sas_ss_reset(struct task_struct *p)
3038 p->sas_ss_flags = SS_DISABLE;
3041 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
3043 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
3044 #ifdef CONFIG_STACK_GROWSUP
3045 return current->sas_ss_sp;
3047 return current->sas_ss_sp + current->sas_ss_size;
3053 * Routines for handling mm_structs
3055 extern struct mm_struct * mm_alloc(void);
3057 /* mmdrop drops the mm and the page tables */
3058 extern void __mmdrop(struct mm_struct *);
3059 static inline void mmdrop(struct mm_struct *mm)
3061 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
3065 static inline void mmdrop_async_fn(struct work_struct *work)
3067 struct mm_struct *mm = container_of(work, struct mm_struct, async_put_work);
3071 static inline void mmdrop_async(struct mm_struct *mm)
3073 if (unlikely(atomic_dec_and_test(&mm->mm_count))) {
3074 INIT_WORK(&mm->async_put_work, mmdrop_async_fn);
3075 schedule_work(&mm->async_put_work);
3079 static inline bool mmget_not_zero(struct mm_struct *mm)
3081 return atomic_inc_not_zero(&mm->mm_users);
3084 /* mmput gets rid of the mappings and all user-space */
3085 extern void mmput(struct mm_struct *);
3087 /* same as above but performs the slow path from the async context. Can
3088 * be called from the atomic context as well
3090 extern void mmput_async(struct mm_struct *);
3093 /* Grab a reference to a task's mm, if it is not already going away */
3094 extern struct mm_struct *get_task_mm(struct task_struct *task);
3096 * Grab a reference to a task's mm, if it is not already going away
3097 * and ptrace_may_access with the mode parameter passed to it
3100 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
3101 /* Remove the current tasks stale references to the old mm_struct */
3102 extern void mm_release(struct task_struct *, struct mm_struct *);
3104 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
3105 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
3106 struct task_struct *, unsigned long);
3108 extern int copy_thread(unsigned long, unsigned long, unsigned long,
3109 struct task_struct *);
3111 /* Architectures that haven't opted into copy_thread_tls get the tls argument
3112 * via pt_regs, so ignore the tls argument passed via C. */
3113 static inline int copy_thread_tls(
3114 unsigned long clone_flags, unsigned long sp, unsigned long arg,
3115 struct task_struct *p, unsigned long tls)
3117 return copy_thread(clone_flags, sp, arg, p);
3120 extern void flush_thread(void);
3122 #ifdef CONFIG_HAVE_EXIT_THREAD
3123 extern void exit_thread(struct task_struct *tsk);
3125 static inline void exit_thread(struct task_struct *tsk)
3130 extern void exit_files(struct task_struct *);
3131 extern void __cleanup_sighand(struct sighand_struct *);
3133 extern void exit_itimers(struct signal_struct *);
3134 extern void flush_itimer_signals(void);
3136 extern void do_group_exit(int);
3138 extern int do_execve(struct filename *,
3139 const char __user * const __user *,
3140 const char __user * const __user *);
3141 extern int do_execveat(int, struct filename *,
3142 const char __user * const __user *,
3143 const char __user * const __user *,
3145 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
3146 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
3147 struct task_struct *fork_idle(int);
3148 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
3150 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
3151 static inline void set_task_comm(struct task_struct *tsk, const char *from)
3153 __set_task_comm(tsk, from, false);
3155 extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk);
3156 #define get_task_comm(buf, tsk) ({ \
3157 BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \
3158 __get_task_comm(buf, sizeof(buf), tsk); \
3162 void scheduler_ipi(void);
3163 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
3165 static inline void scheduler_ipi(void) { }
3166 static inline unsigned long wait_task_inactive(struct task_struct *p,
3173 #define tasklist_empty() \
3174 list_empty(&init_task.tasks)
3176 #define next_task(p) \
3177 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
3179 #define for_each_process(p) \
3180 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
3182 extern bool current_is_single_threaded(void);
3185 * Careful: do_each_thread/while_each_thread is a double loop so
3186 * 'break' will not work as expected - use goto instead.
3188 #define do_each_thread(g, t) \
3189 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
3191 #define while_each_thread(g, t) \
3192 while ((t = next_thread(t)) != g)
3194 #define __for_each_thread(signal, t) \
3195 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
3197 #define for_each_thread(p, t) \
3198 __for_each_thread((p)->signal, t)
3200 /* Careful: this is a double loop, 'break' won't work as expected. */
3201 #define for_each_process_thread(p, t) \
3202 for_each_process(p) for_each_thread(p, t)
3204 static inline int get_nr_threads(struct task_struct *tsk)
3206 return tsk->signal->nr_threads;
3209 static inline bool thread_group_leader(struct task_struct *p)
3211 return p->exit_signal >= 0;
3214 /* Do to the insanities of de_thread it is possible for a process
3215 * to have the pid of the thread group leader without actually being
3216 * the thread group leader. For iteration through the pids in proc
3217 * all we care about is that we have a task with the appropriate
3218 * pid, we don't actually care if we have the right task.
3220 static inline bool has_group_leader_pid(struct task_struct *p)
3222 return task_pid(p) == p->signal->leader_pid;
3226 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
3228 return p1->signal == p2->signal;
3231 static inline struct task_struct *next_thread(const struct task_struct *p)
3233 return list_entry_rcu(p->thread_group.next,
3234 struct task_struct, thread_group);
3237 static inline int thread_group_empty(struct task_struct *p)
3239 return list_empty(&p->thread_group);
3242 #define delay_group_leader(p) \
3243 (thread_group_leader(p) && !thread_group_empty(p))
3246 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
3247 * subscriptions and synchronises with wait4(). Also used in procfs. Also
3248 * pins the final release of task.io_context. Also protects ->cpuset and
3249 * ->cgroup.subsys[]. And ->vfork_done.
3251 * Nests both inside and outside of read_lock(&tasklist_lock).
3252 * It must not be nested with write_lock_irq(&tasklist_lock),
3253 * neither inside nor outside.
3255 static inline void task_lock(struct task_struct *p)
3257 spin_lock(&p->alloc_lock);
3260 static inline void task_unlock(struct task_struct *p)
3262 spin_unlock(&p->alloc_lock);
3265 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
3266 unsigned long *flags);
3268 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
3269 unsigned long *flags)
3271 struct sighand_struct *ret;
3273 ret = __lock_task_sighand(tsk, flags);
3274 (void)__cond_lock(&tsk->sighand->siglock, ret);
3278 static inline void unlock_task_sighand(struct task_struct *tsk,
3279 unsigned long *flags)
3281 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
3285 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
3286 * @tsk: task causing the changes
3288 * All operations which modify a threadgroup - a new thread joining the
3289 * group, death of a member thread (the assertion of PF_EXITING) and
3290 * exec(2) dethreading the process and replacing the leader - are wrapped
3291 * by threadgroup_change_{begin|end}(). This is to provide a place which
3292 * subsystems needing threadgroup stability can hook into for
3295 static inline void threadgroup_change_begin(struct task_struct *tsk)
3298 cgroup_threadgroup_change_begin(tsk);
3302 * threadgroup_change_end - mark the end of changes to a threadgroup
3303 * @tsk: task causing the changes
3305 * See threadgroup_change_begin().
3307 static inline void threadgroup_change_end(struct task_struct *tsk)
3309 cgroup_threadgroup_change_end(tsk);
3312 #ifdef CONFIG_THREAD_INFO_IN_TASK
3314 static inline struct thread_info *task_thread_info(struct task_struct *task)
3316 return &task->thread_info;
3320 * When accessing the stack of a non-current task that might exit, use
3321 * try_get_task_stack() instead. task_stack_page will return a pointer
3322 * that could get freed out from under you.
3324 static inline void *task_stack_page(const struct task_struct *task)
3329 #define setup_thread_stack(new,old) do { } while(0)
3331 static inline unsigned long *end_of_stack(const struct task_struct *task)
3336 #elif !defined(__HAVE_THREAD_FUNCTIONS)
3338 #ifdef CONFIG_AMLOGIC_VMAP
3339 #define task_thread_info(task) \
3340 ((struct thread_info *)(((unsigned long)(task)->stack) + \
3341 THREAD_INFO_OFFSET))
3343 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
3345 #define task_stack_page(task) ((void *)(task)->stack)
3347 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
3349 *task_thread_info(p) = *task_thread_info(org);
3350 task_thread_info(p)->task = p;
3354 * Return the address of the last usable long on the stack.
3356 * When the stack grows down, this is just above the thread
3357 * info struct. Going any lower will corrupt the threadinfo.
3359 * When the stack grows up, this is the highest address.
3360 * Beyond that position, we corrupt data on the next page.
3362 static inline unsigned long *end_of_stack(struct task_struct *p)
3364 #ifdef CONFIG_AMLOGIC_VMAP
3366 #else /* CONFIG_AMLOGIC_VMAP */
3367 #ifdef CONFIG_STACK_GROWSUP
3368 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
3370 return (unsigned long *)(task_thread_info(p) + 1);
3372 #endif /* CONFIG_AMLOGIC_VMAP */
3377 #ifdef CONFIG_THREAD_INFO_IN_TASK
3378 static inline void *try_get_task_stack(struct task_struct *tsk)
3380 return atomic_inc_not_zero(&tsk->stack_refcount) ?
3381 task_stack_page(tsk) : NULL;
3384 extern void put_task_stack(struct task_struct *tsk);
3386 static inline void *try_get_task_stack(struct task_struct *tsk)
3388 return task_stack_page(tsk);
3391 static inline void put_task_stack(struct task_struct *tsk) {}
3394 #ifdef CONFIG_AMLOGIC_VMAP
3395 #define task_stack_end_corrupted(task) (false)
3397 #define task_stack_end_corrupted(task) \
3398 (*(end_of_stack(task)) != STACK_END_MAGIC)
3401 static inline int object_is_on_stack(void *obj)
3403 void *stack = task_stack_page(current);
3405 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
3408 extern void thread_stack_cache_init(void);
3410 #ifdef CONFIG_DEBUG_STACK_USAGE
3411 static inline unsigned long stack_not_used(struct task_struct *p)
3413 unsigned long *n = end_of_stack(p);
3415 do { /* Skip over canary */
3416 # ifdef CONFIG_STACK_GROWSUP
3423 # ifdef CONFIG_STACK_GROWSUP
3424 return (unsigned long)end_of_stack(p) - (unsigned long)n;
3426 return (unsigned long)n - (unsigned long)end_of_stack(p);
3430 extern void set_task_stack_end_magic(struct task_struct *tsk);
3432 /* set thread flags in other task's structures
3433 * - see asm/thread_info.h for TIF_xxxx flags available
3435 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
3437 set_ti_thread_flag(task_thread_info(tsk), flag);
3440 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3442 clear_ti_thread_flag(task_thread_info(tsk), flag);
3445 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
3447 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
3450 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3452 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
3455 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
3457 return test_ti_thread_flag(task_thread_info(tsk), flag);
3460 static inline void set_tsk_need_resched(struct task_struct *tsk)
3462 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3465 static inline void clear_tsk_need_resched(struct task_struct *tsk)
3467 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3470 static inline int test_tsk_need_resched(struct task_struct *tsk)
3472 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
3475 static inline int restart_syscall(void)
3477 set_tsk_thread_flag(current, TIF_SIGPENDING);
3478 return -ERESTARTNOINTR;
3481 static inline int signal_pending(struct task_struct *p)
3483 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
3486 static inline int __fatal_signal_pending(struct task_struct *p)
3488 return unlikely(sigismember(&p->pending.signal, SIGKILL));
3491 static inline int fatal_signal_pending(struct task_struct *p)
3493 return signal_pending(p) && __fatal_signal_pending(p);
3496 static inline int signal_pending_state(long state, struct task_struct *p)
3498 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
3500 if (!signal_pending(p))
3503 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
3507 * cond_resched() and cond_resched_lock(): latency reduction via
3508 * explicit rescheduling in places that are safe. The return
3509 * value indicates whether a reschedule was done in fact.
3510 * cond_resched_lock() will drop the spinlock before scheduling,
3511 * cond_resched_softirq() will enable bhs before scheduling.
3513 #ifndef CONFIG_PREEMPT
3514 extern int _cond_resched(void);
3516 static inline int _cond_resched(void) { return 0; }
3519 #define cond_resched() ({ \
3520 ___might_sleep(__FILE__, __LINE__, 0); \
3524 extern int __cond_resched_lock(spinlock_t *lock);
3526 #define cond_resched_lock(lock) ({ \
3527 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3528 __cond_resched_lock(lock); \
3531 extern int __cond_resched_softirq(void);
3533 #define cond_resched_softirq() ({ \
3534 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3535 __cond_resched_softirq(); \
3538 static inline void cond_resched_rcu(void)
3540 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3547 static inline unsigned long get_preempt_disable_ip(struct task_struct *p)
3549 #ifdef CONFIG_DEBUG_PREEMPT
3550 return p->preempt_disable_ip;
3557 * Does a critical section need to be broken due to another
3558 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3559 * but a general need for low latency)
3561 static inline int spin_needbreak(spinlock_t *lock)
3563 #ifdef CONFIG_PREEMPT
3564 return spin_is_contended(lock);
3571 * Idle thread specific functions to determine the need_resched
3574 #ifdef TIF_POLLING_NRFLAG
3575 static inline int tsk_is_polling(struct task_struct *p)
3577 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3580 static inline void __current_set_polling(void)
3582 set_thread_flag(TIF_POLLING_NRFLAG);
3585 static inline bool __must_check current_set_polling_and_test(void)
3587 __current_set_polling();
3590 * Polling state must be visible before we test NEED_RESCHED,
3591 * paired by resched_curr()
3593 smp_mb__after_atomic();
3595 return unlikely(tif_need_resched());
3598 static inline void __current_clr_polling(void)
3600 clear_thread_flag(TIF_POLLING_NRFLAG);
3603 static inline bool __must_check current_clr_polling_and_test(void)
3605 __current_clr_polling();
3608 * Polling state must be visible before we test NEED_RESCHED,
3609 * paired by resched_curr()
3611 smp_mb__after_atomic();
3613 return unlikely(tif_need_resched());
3617 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3618 static inline void __current_set_polling(void) { }
3619 static inline void __current_clr_polling(void) { }
3621 static inline bool __must_check current_set_polling_and_test(void)
3623 return unlikely(tif_need_resched());
3625 static inline bool __must_check current_clr_polling_and_test(void)
3627 return unlikely(tif_need_resched());
3631 static inline void current_clr_polling(void)
3633 __current_clr_polling();
3636 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3637 * Once the bit is cleared, we'll get IPIs with every new
3638 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3641 smp_mb(); /* paired with resched_curr() */
3643 preempt_fold_need_resched();
3646 static __always_inline bool need_resched(void)
3648 return unlikely(tif_need_resched());
3652 * Thread group CPU time accounting.
3654 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3655 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3658 * Reevaluate whether the task has signals pending delivery.
3659 * Wake the task if so.
3660 * This is required every time the blocked sigset_t changes.
3661 * callers must hold sighand->siglock.
3663 extern void recalc_sigpending_and_wake(struct task_struct *t);
3664 extern void recalc_sigpending(void);
3666 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3668 static inline void signal_wake_up(struct task_struct *t, bool resume)
3670 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3672 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3674 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3678 * Wrappers for p->thread_info->cpu access. No-op on UP.
3682 static inline unsigned int task_cpu(const struct task_struct *p)
3684 #ifdef CONFIG_THREAD_INFO_IN_TASK
3687 return task_thread_info(p)->cpu;
3691 static inline int task_node(const struct task_struct *p)
3693 return cpu_to_node(task_cpu(p));
3696 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3700 static inline unsigned int task_cpu(const struct task_struct *p)
3705 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3709 #endif /* CONFIG_SMP */
3711 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3712 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3714 #ifdef CONFIG_CGROUP_SCHED
3715 extern struct task_group root_task_group;
3716 #endif /* CONFIG_CGROUP_SCHED */
3718 extern int task_can_switch_user(struct user_struct *up,
3719 struct task_struct *tsk);
3721 #ifdef CONFIG_TASK_XACCT
3722 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3724 tsk->ioac.rchar += amt;
3727 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3729 tsk->ioac.wchar += amt;
3732 static inline void inc_syscr(struct task_struct *tsk)
3737 static inline void inc_syscw(struct task_struct *tsk)
3742 static inline void inc_syscfs(struct task_struct *tsk)
3747 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3751 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3755 static inline void inc_syscr(struct task_struct *tsk)
3759 static inline void inc_syscw(struct task_struct *tsk)
3762 static inline void inc_syscfs(struct task_struct *tsk)
3767 #ifndef TASK_SIZE_OF
3768 #define TASK_SIZE_OF(tsk) TASK_SIZE
3772 extern void mm_update_next_owner(struct mm_struct *mm);
3774 static inline void mm_update_next_owner(struct mm_struct *mm)
3777 #endif /* CONFIG_MEMCG */
3779 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3782 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3785 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3788 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3791 static inline unsigned long rlimit(unsigned int limit)
3793 return task_rlimit(current, limit);
3796 static inline unsigned long rlimit_max(unsigned int limit)
3798 return task_rlimit_max(current, limit);
3801 #define SCHED_CPUFREQ_RT (1U << 0)
3802 #define SCHED_CPUFREQ_DL (1U << 1)
3803 #define SCHED_CPUFREQ_IOWAIT (1U << 2)
3805 #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
3807 #ifdef CONFIG_CPU_FREQ
3808 struct update_util_data {
3809 void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
3812 void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
3813 void (*func)(struct update_util_data *data, u64 time,
3814 unsigned int flags));
3815 void cpufreq_remove_update_util_hook(int cpu);
3816 #endif /* CONFIG_CPU_FREQ */