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/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.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);
177 extern void calc_global_load(unsigned long ticks);
179 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
180 extern void update_cpu_load_nohz(int active);
182 static inline void update_cpu_load_nohz(int active) { }
185 extern void dump_cpu_task(int cpu);
190 #ifdef CONFIG_SCHED_DEBUG
191 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192 extern void proc_sched_set_task(struct task_struct *p);
196 * Task state bitmask. NOTE! These bits are also
197 * encoded in fs/proc/array.c: get_task_state().
199 * We have two separate sets of flags: task->state
200 * is about runnability, while task->exit_state are
201 * about the task exiting. Confusing, but this way
202 * modifying one set can't modify the other one by
205 #define TASK_RUNNING 0
206 #define TASK_INTERRUPTIBLE 1
207 #define TASK_UNINTERRUPTIBLE 2
208 #define __TASK_STOPPED 4
209 #define __TASK_TRACED 8
210 /* in tsk->exit_state */
212 #define EXIT_ZOMBIE 32
213 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
214 /* in tsk->state again */
216 #define TASK_WAKEKILL 128
217 #define TASK_WAKING 256
218 #define TASK_PARKED 512
219 #define TASK_NOLOAD 1024
220 #define TASK_STATE_MAX 2048
222 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
224 extern char ___assert_task_state[1 - 2*!!(
225 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
227 /* Convenience macros for the sake of set_task_state */
228 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
229 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
230 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
232 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
234 /* Convenience macros for the sake of wake_up */
235 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
236 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
238 /* get_task_state() */
239 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
240 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
241 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
243 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
244 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
245 #define task_is_stopped_or_traced(task) \
246 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
247 #define task_contributes_to_load(task) \
248 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
249 (task->flags & PF_FROZEN) == 0 && \
250 (task->state & TASK_NOLOAD) == 0)
252 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
254 #define __set_task_state(tsk, state_value) \
256 (tsk)->task_state_change = _THIS_IP_; \
257 (tsk)->state = (state_value); \
259 #define set_task_state(tsk, state_value) \
261 (tsk)->task_state_change = _THIS_IP_; \
262 smp_store_mb((tsk)->state, (state_value)); \
266 * set_current_state() includes a barrier so that the write of current->state
267 * is correctly serialised wrt the caller's subsequent test of whether to
270 * set_current_state(TASK_UNINTERRUPTIBLE);
271 * if (do_i_need_to_sleep())
274 * If the caller does not need such serialisation then use __set_current_state()
276 #define __set_current_state(state_value) \
278 current->task_state_change = _THIS_IP_; \
279 current->state = (state_value); \
281 #define set_current_state(state_value) \
283 current->task_state_change = _THIS_IP_; \
284 smp_store_mb(current->state, (state_value)); \
289 #define __set_task_state(tsk, state_value) \
290 do { (tsk)->state = (state_value); } while (0)
291 #define set_task_state(tsk, state_value) \
292 smp_store_mb((tsk)->state, (state_value))
295 * set_current_state() includes a barrier so that the write of current->state
296 * is correctly serialised wrt the caller's subsequent test of whether to
299 * set_current_state(TASK_UNINTERRUPTIBLE);
300 * if (do_i_need_to_sleep())
303 * If the caller does not need such serialisation then use __set_current_state()
305 #define __set_current_state(state_value) \
306 do { current->state = (state_value); } while (0)
307 #define set_current_state(state_value) \
308 smp_store_mb(current->state, (state_value))
312 /* Task command name length */
313 #define TASK_COMM_LEN 16
315 #include <linux/spinlock.h>
318 * This serializes "schedule()" and also protects
319 * the run-queue from deletions/modifications (but
320 * _adding_ to the beginning of the run-queue has
323 extern rwlock_t tasklist_lock;
324 extern spinlock_t mmlist_lock;
328 #ifdef CONFIG_PROVE_RCU
329 extern int lockdep_tasklist_lock_is_held(void);
330 #endif /* #ifdef CONFIG_PROVE_RCU */
332 extern void sched_init(void);
333 extern void sched_init_smp(void);
334 extern asmlinkage void schedule_tail(struct task_struct *prev);
335 extern void init_idle(struct task_struct *idle, int cpu);
336 extern void init_idle_bootup_task(struct task_struct *idle);
338 extern cpumask_var_t cpu_isolated_map;
340 extern int runqueue_is_locked(int cpu);
342 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
343 extern void nohz_balance_enter_idle(int cpu);
344 extern void set_cpu_sd_state_idle(void);
345 extern int get_nohz_timer_target(void);
347 static inline void nohz_balance_enter_idle(int cpu) { }
348 static inline void set_cpu_sd_state_idle(void) { }
352 * Only dump TASK_* tasks. (0 for all tasks)
354 extern void show_state_filter(unsigned long state_filter);
356 static inline void show_state(void)
358 show_state_filter(0);
361 extern void show_regs(struct pt_regs *);
364 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
365 * task), SP is the stack pointer of the first frame that should be shown in the back
366 * trace (or NULL if the entire call-chain of the task should be shown).
368 extern void show_stack(struct task_struct *task, unsigned long *sp);
370 extern void cpu_init (void);
371 extern void trap_init(void);
372 extern void update_process_times(int user);
373 extern void scheduler_tick(void);
375 extern void sched_show_task(struct task_struct *p);
377 #ifdef CONFIG_LOCKUP_DETECTOR
378 extern void touch_softlockup_watchdog_sched(void);
379 extern void touch_softlockup_watchdog(void);
380 extern void touch_softlockup_watchdog_sync(void);
381 extern void touch_all_softlockup_watchdogs(void);
382 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
384 size_t *lenp, loff_t *ppos);
385 extern unsigned int softlockup_panic;
386 extern unsigned int hardlockup_panic;
387 void lockup_detector_init(void);
389 static inline void touch_softlockup_watchdog_sched(void)
392 static inline void touch_softlockup_watchdog(void)
395 static inline void touch_softlockup_watchdog_sync(void)
398 static inline void touch_all_softlockup_watchdogs(void)
401 static inline void lockup_detector_init(void)
406 #ifdef CONFIG_DETECT_HUNG_TASK
407 void reset_hung_task_detector(void);
409 static inline void reset_hung_task_detector(void)
414 /* Attach to any functions which should be ignored in wchan output. */
415 #define __sched __attribute__((__section__(".sched.text")))
417 /* Linker adds these: start and end of __sched functions */
418 extern char __sched_text_start[], __sched_text_end[];
420 /* Is this address in the __sched functions? */
421 extern int in_sched_functions(unsigned long addr);
423 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
424 extern signed long schedule_timeout(signed long timeout);
425 extern signed long schedule_timeout_interruptible(signed long timeout);
426 extern signed long schedule_timeout_killable(signed long timeout);
427 extern signed long schedule_timeout_uninterruptible(signed long timeout);
428 asmlinkage void schedule(void);
429 extern void schedule_preempt_disabled(void);
431 extern long io_schedule_timeout(long timeout);
433 static inline void io_schedule(void)
435 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
439 struct user_namespace;
442 extern void arch_pick_mmap_layout(struct mm_struct *mm);
444 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
445 unsigned long, unsigned long);
447 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
448 unsigned long len, unsigned long pgoff,
449 unsigned long flags);
451 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
454 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
455 #define SUID_DUMP_USER 1 /* Dump as user of process */
456 #define SUID_DUMP_ROOT 2 /* Dump as root */
460 /* for SUID_DUMP_* above */
461 #define MMF_DUMPABLE_BITS 2
462 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
464 extern void set_dumpable(struct mm_struct *mm, int value);
466 * This returns the actual value of the suid_dumpable flag. For things
467 * that are using this for checking for privilege transitions, it must
468 * test against SUID_DUMP_USER rather than treating it as a boolean
471 static inline int __get_dumpable(unsigned long mm_flags)
473 return mm_flags & MMF_DUMPABLE_MASK;
476 static inline int get_dumpable(struct mm_struct *mm)
478 return __get_dumpable(mm->flags);
481 /* coredump filter bits */
482 #define MMF_DUMP_ANON_PRIVATE 2
483 #define MMF_DUMP_ANON_SHARED 3
484 #define MMF_DUMP_MAPPED_PRIVATE 4
485 #define MMF_DUMP_MAPPED_SHARED 5
486 #define MMF_DUMP_ELF_HEADERS 6
487 #define MMF_DUMP_HUGETLB_PRIVATE 7
488 #define MMF_DUMP_HUGETLB_SHARED 8
489 #define MMF_DUMP_DAX_PRIVATE 9
490 #define MMF_DUMP_DAX_SHARED 10
492 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
493 #define MMF_DUMP_FILTER_BITS 9
494 #define MMF_DUMP_FILTER_MASK \
495 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
496 #define MMF_DUMP_FILTER_DEFAULT \
497 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
498 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
500 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
501 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
503 # define MMF_DUMP_MASK_DEFAULT_ELF 0
505 /* leave room for more dump flags */
506 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
507 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
508 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
510 #define MMF_HAS_UPROBES 19 /* has uprobes */
511 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
513 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
515 struct sighand_struct {
517 struct k_sigaction action[_NSIG];
519 wait_queue_head_t signalfd_wqh;
522 struct pacct_struct {
525 unsigned long ac_mem;
526 cputime_t ac_utime, ac_stime;
527 unsigned long ac_minflt, ac_majflt;
538 * struct prev_cputime - snaphsot of system and user cputime
539 * @utime: time spent in user mode
540 * @stime: time spent in system mode
541 * @lock: protects the above two fields
543 * Stores previous user/system time values such that we can guarantee
546 struct prev_cputime {
547 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
554 static inline void prev_cputime_init(struct prev_cputime *prev)
556 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
557 prev->utime = prev->stime = 0;
558 raw_spin_lock_init(&prev->lock);
563 * struct task_cputime - collected CPU time counts
564 * @utime: time spent in user mode, in &cputime_t units
565 * @stime: time spent in kernel mode, in &cputime_t units
566 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
568 * This structure groups together three kinds of CPU time that are tracked for
569 * threads and thread groups. Most things considering CPU time want to group
570 * these counts together and treat all three of them in parallel.
572 struct task_cputime {
575 unsigned long long sum_exec_runtime;
578 /* Alternate field names when used to cache expirations. */
579 #define virt_exp utime
580 #define prof_exp stime
581 #define sched_exp sum_exec_runtime
583 #define INIT_CPUTIME \
584 (struct task_cputime) { \
587 .sum_exec_runtime = 0, \
591 * This is the atomic variant of task_cputime, which can be used for
592 * storing and updating task_cputime statistics without locking.
594 struct task_cputime_atomic {
597 atomic64_t sum_exec_runtime;
600 #define INIT_CPUTIME_ATOMIC \
601 (struct task_cputime_atomic) { \
602 .utime = ATOMIC64_INIT(0), \
603 .stime = ATOMIC64_INIT(0), \
604 .sum_exec_runtime = ATOMIC64_INIT(0), \
607 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
610 * Disable preemption until the scheduler is running -- use an unconditional
611 * value so that it also works on !PREEMPT_COUNT kernels.
613 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
615 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
618 * Initial preempt_count value; reflects the preempt_count schedule invariant
619 * which states that during context switches:
621 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
623 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
624 * Note: See finish_task_switch().
626 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
629 * struct thread_group_cputimer - thread group interval timer counts
630 * @cputime_atomic: atomic thread group interval timers.
631 * @running: true when there are timers running and
632 * @cputime_atomic receives updates.
633 * @checking_timer: true when a thread in the group is in the
634 * process of checking for thread group timers.
636 * This structure contains the version of task_cputime, above, that is
637 * used for thread group CPU timer calculations.
639 struct thread_group_cputimer {
640 struct task_cputime_atomic cputime_atomic;
645 #include <linux/rwsem.h>
649 * NOTE! "signal_struct" does not have its own
650 * locking, because a shared signal_struct always
651 * implies a shared sighand_struct, so locking
652 * sighand_struct is always a proper superset of
653 * the locking of signal_struct.
655 struct signal_struct {
659 struct list_head thread_head;
661 wait_queue_head_t wait_chldexit; /* for wait4() */
663 /* current thread group signal load-balancing target: */
664 struct task_struct *curr_target;
666 /* shared signal handling: */
667 struct sigpending shared_pending;
669 /* thread group exit support */
672 * - notify group_exit_task when ->count is equal to notify_count
673 * - everyone except group_exit_task is stopped during signal delivery
674 * of fatal signals, group_exit_task processes the signal.
677 struct task_struct *group_exit_task;
679 /* thread group stop support, overloads group_exit_code too */
680 int group_stop_count;
681 unsigned int flags; /* see SIGNAL_* flags below */
684 * PR_SET_CHILD_SUBREAPER marks a process, like a service
685 * manager, to re-parent orphan (double-forking) child processes
686 * to this process instead of 'init'. The service manager is
687 * able to receive SIGCHLD signals and is able to investigate
688 * the process until it calls wait(). All children of this
689 * process will inherit a flag if they should look for a
690 * child_subreaper process at exit.
692 unsigned int is_child_subreaper:1;
693 unsigned int has_child_subreaper:1;
695 /* POSIX.1b Interval Timers */
697 struct list_head posix_timers;
699 /* ITIMER_REAL timer for the process */
700 struct hrtimer real_timer;
701 struct pid *leader_pid;
702 ktime_t it_real_incr;
705 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
706 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
707 * values are defined to 0 and 1 respectively
709 struct cpu_itimer it[2];
712 * Thread group totals for process CPU timers.
713 * See thread_group_cputimer(), et al, for details.
715 struct thread_group_cputimer cputimer;
717 /* Earliest-expiration cache. */
718 struct task_cputime cputime_expires;
720 #ifdef CONFIG_NO_HZ_FULL
721 unsigned long tick_dep_mask;
724 struct list_head cpu_timers[3];
726 struct pid *tty_old_pgrp;
728 /* boolean value for session group leader */
731 struct tty_struct *tty; /* NULL if no tty */
733 #ifdef CONFIG_SCHED_AUTOGROUP
734 struct autogroup *autogroup;
737 * Cumulative resource counters for dead threads in the group,
738 * and for reaped dead child processes forked by this group.
739 * Live threads maintain their own counters and add to these
740 * in __exit_signal, except for the group leader.
742 seqlock_t stats_lock;
743 cputime_t utime, stime, cutime, cstime;
746 struct prev_cputime prev_cputime;
747 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
748 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
749 unsigned long inblock, oublock, cinblock, coublock;
750 unsigned long maxrss, cmaxrss;
751 struct task_io_accounting ioac;
754 * Cumulative ns of schedule CPU time fo dead threads in the
755 * group, not including a zombie group leader, (This only differs
756 * from jiffies_to_ns(utime + stime) if sched_clock uses something
757 * other than jiffies.)
759 unsigned long long sum_sched_runtime;
762 * We don't bother to synchronize most readers of this at all,
763 * because there is no reader checking a limit that actually needs
764 * to get both rlim_cur and rlim_max atomically, and either one
765 * alone is a single word that can safely be read normally.
766 * getrlimit/setrlimit use task_lock(current->group_leader) to
767 * protect this instead of the siglock, because they really
768 * have no need to disable irqs.
770 struct rlimit rlim[RLIM_NLIMITS];
772 #ifdef CONFIG_BSD_PROCESS_ACCT
773 struct pacct_struct pacct; /* per-process accounting information */
775 #ifdef CONFIG_TASKSTATS
776 struct taskstats *stats;
780 struct tty_audit_buf *tty_audit_buf;
783 oom_flags_t oom_flags;
784 short oom_score_adj; /* OOM kill score adjustment */
785 short oom_score_adj_min; /* OOM kill score adjustment min value.
786 * Only settable by CAP_SYS_RESOURCE. */
788 struct mutex cred_guard_mutex; /* guard against foreign influences on
789 * credential calculations
790 * (notably. ptrace) */
794 * Bits in flags field of signal_struct.
796 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
797 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
798 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
799 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
801 * Pending notifications to parent.
803 #define SIGNAL_CLD_STOPPED 0x00000010
804 #define SIGNAL_CLD_CONTINUED 0x00000020
805 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
807 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
809 /* If true, all threads except ->group_exit_task have pending SIGKILL */
810 static inline int signal_group_exit(const struct signal_struct *sig)
812 return (sig->flags & SIGNAL_GROUP_EXIT) ||
813 (sig->group_exit_task != NULL);
817 * Some day this will be a full-fledged user tracking system..
820 atomic_t __count; /* reference count */
821 atomic_t processes; /* How many processes does this user have? */
822 atomic_t sigpending; /* How many pending signals does this user have? */
823 #ifdef CONFIG_INOTIFY_USER
824 atomic_t inotify_watches; /* How many inotify watches does this user have? */
825 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
827 #ifdef CONFIG_FANOTIFY
828 atomic_t fanotify_listeners;
831 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
833 #ifdef CONFIG_POSIX_MQUEUE
834 /* protected by mq_lock */
835 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
837 unsigned long locked_shm; /* How many pages of mlocked shm ? */
838 unsigned long unix_inflight; /* How many files in flight in unix sockets */
839 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
842 struct key *uid_keyring; /* UID specific keyring */
843 struct key *session_keyring; /* UID's default session keyring */
846 /* Hash table maintenance information */
847 struct hlist_node uidhash_node;
850 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
851 atomic_long_t locked_vm;
855 extern int uids_sysfs_init(void);
857 extern struct user_struct *find_user(kuid_t);
859 extern struct user_struct root_user;
860 #define INIT_USER (&root_user)
863 struct backing_dev_info;
864 struct reclaim_state;
866 #ifdef CONFIG_SCHED_INFO
868 /* cumulative counters */
869 unsigned long pcount; /* # of times run on this cpu */
870 unsigned long long run_delay; /* time spent waiting on a runqueue */
873 unsigned long long last_arrival,/* when we last ran on a cpu */
874 last_queued; /* when we were last queued to run */
876 #endif /* CONFIG_SCHED_INFO */
878 #ifdef CONFIG_TASK_DELAY_ACCT
879 struct task_delay_info {
881 unsigned int flags; /* Private per-task flags */
883 /* For each stat XXX, add following, aligned appropriately
885 * struct timespec XXX_start, XXX_end;
889 * Atomicity of updates to XXX_delay, XXX_count protected by
890 * single lock above (split into XXX_lock if contention is an issue).
894 * XXX_count is incremented on every XXX operation, the delay
895 * associated with the operation is added to XXX_delay.
896 * XXX_delay contains the accumulated delay time in nanoseconds.
898 u64 blkio_start; /* Shared by blkio, swapin */
899 u64 blkio_delay; /* wait for sync block io completion */
900 u64 swapin_delay; /* wait for swapin block io completion */
901 u32 blkio_count; /* total count of the number of sync block */
902 /* io operations performed */
903 u32 swapin_count; /* total count of the number of swapin block */
904 /* io operations performed */
907 u64 freepages_delay; /* wait for memory reclaim */
908 u32 freepages_count; /* total count of memory reclaim */
910 #endif /* CONFIG_TASK_DELAY_ACCT */
912 static inline int sched_info_on(void)
914 #ifdef CONFIG_SCHEDSTATS
916 #elif defined(CONFIG_TASK_DELAY_ACCT)
917 extern int delayacct_on;
924 #ifdef CONFIG_SCHEDSTATS
925 void force_schedstat_enabled(void);
936 * Increase resolution of cpu_capacity calculations
938 #define SCHED_CAPACITY_SHIFT 10
939 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
942 * Wake-queues are lists of tasks with a pending wakeup, whose
943 * callers have already marked the task as woken internally,
944 * and can thus carry on. A common use case is being able to
945 * do the wakeups once the corresponding user lock as been
948 * We hold reference to each task in the list across the wakeup,
949 * thus guaranteeing that the memory is still valid by the time
950 * the actual wakeups are performed in wake_up_q().
952 * One per task suffices, because there's never a need for a task to be
953 * in two wake queues simultaneously; it is forbidden to abandon a task
954 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
955 * already in a wake queue, the wakeup will happen soon and the second
956 * waker can just skip it.
958 * The WAKE_Q macro declares and initializes the list head.
959 * wake_up_q() does NOT reinitialize the list; it's expected to be
960 * called near the end of a function, where the fact that the queue is
961 * not used again will be easy to see by inspection.
963 * Note that this can cause spurious wakeups. schedule() callers
964 * must ensure the call is done inside a loop, confirming that the
965 * wakeup condition has in fact occurred.
968 struct wake_q_node *next;
972 struct wake_q_node *first;
973 struct wake_q_node **lastp;
976 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
978 #define WAKE_Q(name) \
979 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
981 extern void wake_q_add(struct wake_q_head *head,
982 struct task_struct *task);
983 extern void wake_up_q(struct wake_q_head *head);
986 * sched-domains (multiprocessor balancing) declarations:
989 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
990 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
991 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
992 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
993 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
994 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
995 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
996 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
997 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
998 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
999 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1000 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1001 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1002 #define SD_NUMA 0x4000 /* cross-node balancing */
1004 #ifdef CONFIG_SCHED_SMT
1005 static inline int cpu_smt_flags(void)
1007 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1011 #ifdef CONFIG_SCHED_MC
1012 static inline int cpu_core_flags(void)
1014 return SD_SHARE_PKG_RESOURCES;
1019 static inline int cpu_numa_flags(void)
1025 struct sched_domain_attr {
1026 int relax_domain_level;
1029 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1030 .relax_domain_level = -1, \
1033 extern int sched_domain_level_max;
1037 struct sched_domain {
1038 /* These fields must be setup */
1039 struct sched_domain *parent; /* top domain must be null terminated */
1040 struct sched_domain *child; /* bottom domain must be null terminated */
1041 struct sched_group *groups; /* the balancing groups of the domain */
1042 unsigned long min_interval; /* Minimum balance interval ms */
1043 unsigned long max_interval; /* Maximum balance interval ms */
1044 unsigned int busy_factor; /* less balancing by factor if busy */
1045 unsigned int imbalance_pct; /* No balance until over watermark */
1046 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1047 unsigned int busy_idx;
1048 unsigned int idle_idx;
1049 unsigned int newidle_idx;
1050 unsigned int wake_idx;
1051 unsigned int forkexec_idx;
1052 unsigned int smt_gain;
1054 int nohz_idle; /* NOHZ IDLE status */
1055 int flags; /* See SD_* */
1058 /* Runtime fields. */
1059 unsigned long last_balance; /* init to jiffies. units in jiffies */
1060 unsigned int balance_interval; /* initialise to 1. units in ms. */
1061 unsigned int nr_balance_failed; /* initialise to 0 */
1063 /* idle_balance() stats */
1064 u64 max_newidle_lb_cost;
1065 unsigned long next_decay_max_lb_cost;
1067 #ifdef CONFIG_SCHEDSTATS
1068 /* load_balance() stats */
1069 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1070 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1071 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1072 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1073 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1074 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1075 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1076 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1078 /* Active load balancing */
1079 unsigned int alb_count;
1080 unsigned int alb_failed;
1081 unsigned int alb_pushed;
1083 /* SD_BALANCE_EXEC stats */
1084 unsigned int sbe_count;
1085 unsigned int sbe_balanced;
1086 unsigned int sbe_pushed;
1088 /* SD_BALANCE_FORK stats */
1089 unsigned int sbf_count;
1090 unsigned int sbf_balanced;
1091 unsigned int sbf_pushed;
1093 /* try_to_wake_up() stats */
1094 unsigned int ttwu_wake_remote;
1095 unsigned int ttwu_move_affine;
1096 unsigned int ttwu_move_balance;
1098 #ifdef CONFIG_SCHED_DEBUG
1102 void *private; /* used during construction */
1103 struct rcu_head rcu; /* used during destruction */
1106 unsigned int span_weight;
1108 * Span of all CPUs in this domain.
1110 * NOTE: this field is variable length. (Allocated dynamically
1111 * by attaching extra space to the end of the structure,
1112 * depending on how many CPUs the kernel has booted up with)
1114 unsigned long span[0];
1117 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1119 return to_cpumask(sd->span);
1122 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1123 struct sched_domain_attr *dattr_new);
1125 /* Allocate an array of sched domains, for partition_sched_domains(). */
1126 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1127 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1129 bool cpus_share_cache(int this_cpu, int that_cpu);
1131 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1132 typedef int (*sched_domain_flags_f)(void);
1134 #define SDTL_OVERLAP 0x01
1137 struct sched_domain **__percpu sd;
1138 struct sched_group **__percpu sg;
1139 struct sched_group_capacity **__percpu sgc;
1142 struct sched_domain_topology_level {
1143 sched_domain_mask_f mask;
1144 sched_domain_flags_f sd_flags;
1147 struct sd_data data;
1148 #ifdef CONFIG_SCHED_DEBUG
1153 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1154 extern void wake_up_if_idle(int cpu);
1156 #ifdef CONFIG_SCHED_DEBUG
1157 # define SD_INIT_NAME(type) .name = #type
1159 # define SD_INIT_NAME(type)
1162 #else /* CONFIG_SMP */
1164 struct sched_domain_attr;
1167 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1168 struct sched_domain_attr *dattr_new)
1172 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1177 #endif /* !CONFIG_SMP */
1180 struct io_context; /* See blkdev.h */
1183 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1184 extern void prefetch_stack(struct task_struct *t);
1186 static inline void prefetch_stack(struct task_struct *t) { }
1189 struct audit_context; /* See audit.c */
1191 struct pipe_inode_info;
1192 struct uts_namespace;
1194 struct load_weight {
1195 unsigned long weight;
1200 * The load_avg/util_avg accumulates an infinite geometric series.
1201 * 1) load_avg factors frequency scaling into the amount of time that a
1202 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1203 * aggregated such weights of all runnable and blocked sched_entities.
1204 * 2) util_avg factors frequency and cpu scaling into the amount of time
1205 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1206 * For cfs_rq, it is the aggregated such times of all runnable and
1207 * blocked sched_entities.
1208 * The 64 bit load_sum can:
1209 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1210 * the highest weight (=88761) always runnable, we should not overflow
1211 * 2) for entity, support any load.weight always runnable
1214 u64 last_update_time, load_sum;
1215 u32 util_sum, period_contrib;
1216 unsigned long load_avg, util_avg;
1219 #ifdef CONFIG_SCHEDSTATS
1220 struct sched_statistics {
1230 s64 sum_sleep_runtime;
1237 u64 nr_migrations_cold;
1238 u64 nr_failed_migrations_affine;
1239 u64 nr_failed_migrations_running;
1240 u64 nr_failed_migrations_hot;
1241 u64 nr_forced_migrations;
1244 u64 nr_wakeups_sync;
1245 u64 nr_wakeups_migrate;
1246 u64 nr_wakeups_local;
1247 u64 nr_wakeups_remote;
1248 u64 nr_wakeups_affine;
1249 u64 nr_wakeups_affine_attempts;
1250 u64 nr_wakeups_passive;
1251 u64 nr_wakeups_idle;
1255 struct sched_entity {
1256 struct load_weight load; /* for load-balancing */
1257 struct rb_node run_node;
1258 struct list_head group_node;
1262 u64 sum_exec_runtime;
1264 u64 prev_sum_exec_runtime;
1268 #ifdef CONFIG_SCHEDSTATS
1269 struct sched_statistics statistics;
1272 #ifdef CONFIG_FAIR_GROUP_SCHED
1274 struct sched_entity *parent;
1275 /* rq on which this entity is (to be) queued: */
1276 struct cfs_rq *cfs_rq;
1277 /* rq "owned" by this entity/group: */
1278 struct cfs_rq *my_q;
1283 * Per entity load average tracking.
1285 * Put into separate cache line so it does not
1286 * collide with read-mostly values above.
1288 struct sched_avg avg ____cacheline_aligned_in_smp;
1292 struct sched_rt_entity {
1293 struct list_head run_list;
1294 unsigned long timeout;
1295 unsigned long watchdog_stamp;
1296 unsigned int time_slice;
1297 unsigned short on_rq;
1298 unsigned short on_list;
1300 struct sched_rt_entity *back;
1301 #ifdef CONFIG_RT_GROUP_SCHED
1302 struct sched_rt_entity *parent;
1303 /* rq on which this entity is (to be) queued: */
1304 struct rt_rq *rt_rq;
1305 /* rq "owned" by this entity/group: */
1310 struct sched_dl_entity {
1311 struct rb_node rb_node;
1314 * Original scheduling parameters. Copied here from sched_attr
1315 * during sched_setattr(), they will remain the same until
1316 * the next sched_setattr().
1318 u64 dl_runtime; /* maximum runtime for each instance */
1319 u64 dl_deadline; /* relative deadline of each instance */
1320 u64 dl_period; /* separation of two instances (period) */
1321 u64 dl_bw; /* dl_runtime / dl_deadline */
1324 * Actual scheduling parameters. Initialized with the values above,
1325 * they are continously updated during task execution. Note that
1326 * the remaining runtime could be < 0 in case we are in overrun.
1328 s64 runtime; /* remaining runtime for this instance */
1329 u64 deadline; /* absolute deadline for this instance */
1330 unsigned int flags; /* specifying the scheduler behaviour */
1335 * @dl_throttled tells if we exhausted the runtime. If so, the
1336 * task has to wait for a replenishment to be performed at the
1337 * next firing of dl_timer.
1339 * @dl_boosted tells if we are boosted due to DI. If so we are
1340 * outside bandwidth enforcement mechanism (but only until we
1341 * exit the critical section);
1343 * @dl_yielded tells if task gave up the cpu before consuming
1344 * all its available runtime during the last job.
1346 int dl_throttled, dl_boosted, dl_yielded;
1349 * Bandwidth enforcement timer. Each -deadline task has its
1350 * own bandwidth to be enforced, thus we need one timer per task.
1352 struct hrtimer dl_timer;
1360 u8 pad; /* Otherwise the compiler can store garbage here. */
1362 u32 s; /* Set of bits. */
1366 enum perf_event_task_context {
1367 perf_invalid_context = -1,
1368 perf_hw_context = 0,
1370 perf_nr_task_contexts,
1373 /* Track pages that require TLB flushes */
1374 struct tlbflush_unmap_batch {
1376 * Each bit set is a CPU that potentially has a TLB entry for one of
1377 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1379 struct cpumask cpumask;
1381 /* True if any bit in cpumask is set */
1382 bool flush_required;
1385 * If true then the PTE was dirty when unmapped. The entry must be
1386 * flushed before IO is initiated or a stale TLB entry potentially
1387 * allows an update without redirtying the page.
1392 struct task_struct {
1393 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1396 unsigned int flags; /* per process flags, defined below */
1397 unsigned int ptrace;
1400 struct llist_node wake_entry;
1402 unsigned int wakee_flips;
1403 unsigned long wakee_flip_decay_ts;
1404 struct task_struct *last_wakee;
1410 int prio, static_prio, normal_prio;
1411 unsigned int rt_priority;
1412 const struct sched_class *sched_class;
1413 struct sched_entity se;
1414 struct sched_rt_entity rt;
1415 #ifdef CONFIG_CGROUP_SCHED
1416 struct task_group *sched_task_group;
1418 struct sched_dl_entity dl;
1420 #ifdef CONFIG_PREEMPT_NOTIFIERS
1421 /* list of struct preempt_notifier: */
1422 struct hlist_head preempt_notifiers;
1425 #ifdef CONFIG_BLK_DEV_IO_TRACE
1426 unsigned int btrace_seq;
1429 unsigned int policy;
1430 int nr_cpus_allowed;
1431 cpumask_t cpus_allowed;
1433 #ifdef CONFIG_PREEMPT_RCU
1434 int rcu_read_lock_nesting;
1435 union rcu_special rcu_read_unlock_special;
1436 struct list_head rcu_node_entry;
1437 struct rcu_node *rcu_blocked_node;
1438 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1439 #ifdef CONFIG_TASKS_RCU
1440 unsigned long rcu_tasks_nvcsw;
1441 bool rcu_tasks_holdout;
1442 struct list_head rcu_tasks_holdout_list;
1443 int rcu_tasks_idle_cpu;
1444 #endif /* #ifdef CONFIG_TASKS_RCU */
1446 #ifdef CONFIG_SCHED_INFO
1447 struct sched_info sched_info;
1450 struct list_head tasks;
1452 struct plist_node pushable_tasks;
1453 struct rb_node pushable_dl_tasks;
1456 struct mm_struct *mm, *active_mm;
1457 /* per-thread vma caching */
1458 u32 vmacache_seqnum;
1459 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1460 #if defined(SPLIT_RSS_COUNTING)
1461 struct task_rss_stat rss_stat;
1465 int exit_code, exit_signal;
1466 int pdeath_signal; /* The signal sent when the parent dies */
1467 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1469 /* Used for emulating ABI behavior of previous Linux versions */
1470 unsigned int personality;
1472 /* scheduler bits, serialized by scheduler locks */
1473 unsigned sched_reset_on_fork:1;
1474 unsigned sched_contributes_to_load:1;
1475 unsigned sched_migrated:1;
1476 unsigned :0; /* force alignment to the next boundary */
1478 /* unserialized, strictly 'current' */
1479 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1480 unsigned in_iowait:1;
1482 unsigned memcg_may_oom:1;
1484 unsigned memcg_kmem_skip_account:1;
1487 #ifdef CONFIG_COMPAT_BRK
1488 unsigned brk_randomized:1;
1491 unsigned long atomic_flags; /* Flags needing atomic access. */
1493 struct restart_block restart_block;
1498 #ifdef CONFIG_CC_STACKPROTECTOR
1499 /* Canary value for the -fstack-protector gcc feature */
1500 unsigned long stack_canary;
1503 * pointers to (original) parent process, youngest child, younger sibling,
1504 * older sibling, respectively. (p->father can be replaced with
1505 * p->real_parent->pid)
1507 struct task_struct __rcu *real_parent; /* real parent process */
1508 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1510 * children/sibling forms the list of my natural children
1512 struct list_head children; /* list of my children */
1513 struct list_head sibling; /* linkage in my parent's children list */
1514 struct task_struct *group_leader; /* threadgroup leader */
1517 * ptraced is the list of tasks this task is using ptrace on.
1518 * This includes both natural children and PTRACE_ATTACH targets.
1519 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1521 struct list_head ptraced;
1522 struct list_head ptrace_entry;
1524 /* PID/PID hash table linkage. */
1525 struct pid_link pids[PIDTYPE_MAX];
1526 struct list_head thread_group;
1527 struct list_head thread_node;
1529 struct completion *vfork_done; /* for vfork() */
1530 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1531 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1533 cputime_t utime, stime, utimescaled, stimescaled;
1535 struct prev_cputime prev_cputime;
1536 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1537 seqcount_t vtime_seqcount;
1538 unsigned long long vtime_snap;
1540 /* Task is sleeping or running in a CPU with VTIME inactive */
1542 /* Task runs in userspace in a CPU with VTIME active */
1544 /* Task runs in kernelspace in a CPU with VTIME active */
1546 } vtime_snap_whence;
1549 #ifdef CONFIG_NO_HZ_FULL
1550 unsigned long tick_dep_mask;
1552 unsigned long nvcsw, nivcsw; /* context switch counts */
1553 u64 start_time; /* monotonic time in nsec */
1554 u64 real_start_time; /* boot based time in nsec */
1555 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1556 unsigned long min_flt, maj_flt;
1558 struct task_cputime cputime_expires;
1559 struct list_head cpu_timers[3];
1561 /* process credentials */
1562 const struct cred __rcu *real_cred; /* objective and real subjective task
1563 * credentials (COW) */
1564 const struct cred __rcu *cred; /* effective (overridable) subjective task
1565 * credentials (COW) */
1566 char comm[TASK_COMM_LEN]; /* executable name excluding path
1567 - access with [gs]et_task_comm (which lock
1568 it with task_lock())
1569 - initialized normally by setup_new_exec */
1570 /* file system info */
1571 struct nameidata *nameidata;
1572 #ifdef CONFIG_SYSVIPC
1574 struct sysv_sem sysvsem;
1575 struct sysv_shm sysvshm;
1577 #ifdef CONFIG_DETECT_HUNG_TASK
1578 /* hung task detection */
1579 unsigned long last_switch_count;
1581 /* filesystem information */
1582 struct fs_struct *fs;
1583 /* open file information */
1584 struct files_struct *files;
1586 struct nsproxy *nsproxy;
1587 /* signal handlers */
1588 struct signal_struct *signal;
1589 struct sighand_struct *sighand;
1591 sigset_t blocked, real_blocked;
1592 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1593 struct sigpending pending;
1595 unsigned long sas_ss_sp;
1598 struct callback_head *task_works;
1600 struct audit_context *audit_context;
1601 #ifdef CONFIG_AUDITSYSCALL
1603 unsigned int sessionid;
1605 struct seccomp seccomp;
1607 /* Thread group tracking */
1610 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1612 spinlock_t alloc_lock;
1614 /* Protection of the PI data structures: */
1615 raw_spinlock_t pi_lock;
1617 struct wake_q_node wake_q;
1619 #ifdef CONFIG_RT_MUTEXES
1620 /* PI waiters blocked on a rt_mutex held by this task */
1621 struct rb_root pi_waiters;
1622 struct rb_node *pi_waiters_leftmost;
1623 /* Deadlock detection and priority inheritance handling */
1624 struct rt_mutex_waiter *pi_blocked_on;
1627 #ifdef CONFIG_DEBUG_MUTEXES
1628 /* mutex deadlock detection */
1629 struct mutex_waiter *blocked_on;
1631 #ifdef CONFIG_TRACE_IRQFLAGS
1632 unsigned int irq_events;
1633 unsigned long hardirq_enable_ip;
1634 unsigned long hardirq_disable_ip;
1635 unsigned int hardirq_enable_event;
1636 unsigned int hardirq_disable_event;
1637 int hardirqs_enabled;
1638 int hardirq_context;
1639 unsigned long softirq_disable_ip;
1640 unsigned long softirq_enable_ip;
1641 unsigned int softirq_disable_event;
1642 unsigned int softirq_enable_event;
1643 int softirqs_enabled;
1644 int softirq_context;
1646 #ifdef CONFIG_LOCKDEP
1647 # define MAX_LOCK_DEPTH 48UL
1650 unsigned int lockdep_recursion;
1651 struct held_lock held_locks[MAX_LOCK_DEPTH];
1652 gfp_t lockdep_reclaim_gfp;
1655 unsigned int in_ubsan;
1658 /* journalling filesystem info */
1661 /* stacked block device info */
1662 struct bio_list *bio_list;
1665 /* stack plugging */
1666 struct blk_plug *plug;
1670 struct reclaim_state *reclaim_state;
1672 struct backing_dev_info *backing_dev_info;
1674 struct io_context *io_context;
1676 unsigned long ptrace_message;
1677 siginfo_t *last_siginfo; /* For ptrace use. */
1678 struct task_io_accounting ioac;
1679 #if defined(CONFIG_TASK_XACCT)
1680 u64 acct_rss_mem1; /* accumulated rss usage */
1681 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1682 cputime_t acct_timexpd; /* stime + utime since last update */
1684 #ifdef CONFIG_CPUSETS
1685 nodemask_t mems_allowed; /* Protected by alloc_lock */
1686 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1687 int cpuset_mem_spread_rotor;
1688 int cpuset_slab_spread_rotor;
1690 #ifdef CONFIG_CGROUPS
1691 /* Control Group info protected by css_set_lock */
1692 struct css_set __rcu *cgroups;
1693 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1694 struct list_head cg_list;
1697 struct robust_list_head __user *robust_list;
1698 #ifdef CONFIG_COMPAT
1699 struct compat_robust_list_head __user *compat_robust_list;
1701 struct list_head pi_state_list;
1702 struct futex_pi_state *pi_state_cache;
1704 #ifdef CONFIG_PERF_EVENTS
1705 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1706 struct mutex perf_event_mutex;
1707 struct list_head perf_event_list;
1709 #ifdef CONFIG_DEBUG_PREEMPT
1710 unsigned long preempt_disable_ip;
1713 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1715 short pref_node_fork;
1717 #ifdef CONFIG_NUMA_BALANCING
1719 unsigned int numa_scan_period;
1720 unsigned int numa_scan_period_max;
1721 int numa_preferred_nid;
1722 unsigned long numa_migrate_retry;
1723 u64 node_stamp; /* migration stamp */
1724 u64 last_task_numa_placement;
1725 u64 last_sum_exec_runtime;
1726 struct callback_head numa_work;
1728 struct list_head numa_entry;
1729 struct numa_group *numa_group;
1732 * numa_faults is an array split into four regions:
1733 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1734 * in this precise order.
1736 * faults_memory: Exponential decaying average of faults on a per-node
1737 * basis. Scheduling placement decisions are made based on these
1738 * counts. The values remain static for the duration of a PTE scan.
1739 * faults_cpu: Track the nodes the process was running on when a NUMA
1740 * hinting fault was incurred.
1741 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1742 * during the current scan window. When the scan completes, the counts
1743 * in faults_memory and faults_cpu decay and these values are copied.
1745 unsigned long *numa_faults;
1746 unsigned long total_numa_faults;
1749 * numa_faults_locality tracks if faults recorded during the last
1750 * scan window were remote/local or failed to migrate. The task scan
1751 * period is adapted based on the locality of the faults with different
1752 * weights depending on whether they were shared or private faults
1754 unsigned long numa_faults_locality[3];
1756 unsigned long numa_pages_migrated;
1757 #endif /* CONFIG_NUMA_BALANCING */
1759 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1760 struct tlbflush_unmap_batch tlb_ubc;
1763 struct rcu_head rcu;
1766 * cache last used pipe for splice
1768 struct pipe_inode_info *splice_pipe;
1770 struct page_frag task_frag;
1772 #ifdef CONFIG_TASK_DELAY_ACCT
1773 struct task_delay_info *delays;
1775 #ifdef CONFIG_FAULT_INJECTION
1779 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1780 * balance_dirty_pages() for some dirty throttling pause
1783 int nr_dirtied_pause;
1784 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1786 #ifdef CONFIG_LATENCYTOP
1787 int latency_record_count;
1788 struct latency_record latency_record[LT_SAVECOUNT];
1791 * time slack values; these are used to round up poll() and
1792 * select() etc timeout values. These are in nanoseconds.
1795 u64 default_timer_slack_ns;
1798 unsigned int kasan_depth;
1800 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1801 /* Index of current stored address in ret_stack */
1803 /* Stack of return addresses for return function tracing */
1804 struct ftrace_ret_stack *ret_stack;
1805 /* time stamp for last schedule */
1806 unsigned long long ftrace_timestamp;
1808 * Number of functions that haven't been traced
1809 * because of depth overrun.
1811 atomic_t trace_overrun;
1812 /* Pause for the tracing */
1813 atomic_t tracing_graph_pause;
1815 #ifdef CONFIG_TRACING
1816 /* state flags for use by tracers */
1817 unsigned long trace;
1818 /* bitmask and counter of trace recursion */
1819 unsigned long trace_recursion;
1820 #endif /* CONFIG_TRACING */
1822 struct mem_cgroup *memcg_in_oom;
1823 gfp_t memcg_oom_gfp_mask;
1824 int memcg_oom_order;
1826 /* number of pages to reclaim on returning to userland */
1827 unsigned int memcg_nr_pages_over_high;
1829 #ifdef CONFIG_UPROBES
1830 struct uprobe_task *utask;
1832 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1833 unsigned int sequential_io;
1834 unsigned int sequential_io_avg;
1836 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1837 unsigned long task_state_change;
1839 int pagefault_disabled;
1840 /* CPU-specific state of this task */
1841 struct thread_struct thread;
1843 * WARNING: on x86, 'thread_struct' contains a variable-sized
1844 * structure. It *MUST* be at the end of 'task_struct'.
1846 * Do not put anything below here!
1850 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1851 extern int arch_task_struct_size __read_mostly;
1853 # define arch_task_struct_size (sizeof(struct task_struct))
1856 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1857 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1859 #define TNF_MIGRATED 0x01
1860 #define TNF_NO_GROUP 0x02
1861 #define TNF_SHARED 0x04
1862 #define TNF_FAULT_LOCAL 0x08
1863 #define TNF_MIGRATE_FAIL 0x10
1865 #ifdef CONFIG_NUMA_BALANCING
1866 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1867 extern pid_t task_numa_group_id(struct task_struct *p);
1868 extern void set_numabalancing_state(bool enabled);
1869 extern void task_numa_free(struct task_struct *p);
1870 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1871 int src_nid, int dst_cpu);
1873 static inline void task_numa_fault(int last_node, int node, int pages,
1877 static inline pid_t task_numa_group_id(struct task_struct *p)
1881 static inline void set_numabalancing_state(bool enabled)
1884 static inline void task_numa_free(struct task_struct *p)
1887 static inline bool should_numa_migrate_memory(struct task_struct *p,
1888 struct page *page, int src_nid, int dst_cpu)
1894 static inline struct pid *task_pid(struct task_struct *task)
1896 return task->pids[PIDTYPE_PID].pid;
1899 static inline struct pid *task_tgid(struct task_struct *task)
1901 return task->group_leader->pids[PIDTYPE_PID].pid;
1905 * Without tasklist or rcu lock it is not safe to dereference
1906 * the result of task_pgrp/task_session even if task == current,
1907 * we can race with another thread doing sys_setsid/sys_setpgid.
1909 static inline struct pid *task_pgrp(struct task_struct *task)
1911 return task->group_leader->pids[PIDTYPE_PGID].pid;
1914 static inline struct pid *task_session(struct task_struct *task)
1916 return task->group_leader->pids[PIDTYPE_SID].pid;
1919 struct pid_namespace;
1922 * the helpers to get the task's different pids as they are seen
1923 * from various namespaces
1925 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1926 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1928 * task_xid_nr_ns() : id seen from the ns specified;
1930 * set_task_vxid() : assigns a virtual id to a task;
1932 * see also pid_nr() etc in include/linux/pid.h
1934 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1935 struct pid_namespace *ns);
1937 static inline pid_t task_pid_nr(struct task_struct *tsk)
1942 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1943 struct pid_namespace *ns)
1945 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1948 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1950 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1954 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1959 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1961 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1963 return pid_vnr(task_tgid(tsk));
1967 static inline int pid_alive(const struct task_struct *p);
1968 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1974 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1980 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1982 return task_ppid_nr_ns(tsk, &init_pid_ns);
1985 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1986 struct pid_namespace *ns)
1988 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1991 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1993 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1997 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1998 struct pid_namespace *ns)
2000 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2003 static inline pid_t task_session_vnr(struct task_struct *tsk)
2005 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2008 /* obsolete, do not use */
2009 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2011 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2015 * pid_alive - check that a task structure is not stale
2016 * @p: Task structure to be checked.
2018 * Test if a process is not yet dead (at most zombie state)
2019 * If pid_alive fails, then pointers within the task structure
2020 * can be stale and must not be dereferenced.
2022 * Return: 1 if the process is alive. 0 otherwise.
2024 static inline int pid_alive(const struct task_struct *p)
2026 return p->pids[PIDTYPE_PID].pid != NULL;
2030 * is_global_init - check if a task structure is init. Since init
2031 * is free to have sub-threads we need to check tgid.
2032 * @tsk: Task structure to be checked.
2034 * Check if a task structure is the first user space task the kernel created.
2036 * Return: 1 if the task structure is init. 0 otherwise.
2038 static inline int is_global_init(struct task_struct *tsk)
2040 return task_tgid_nr(tsk) == 1;
2043 extern struct pid *cad_pid;
2045 extern void free_task(struct task_struct *tsk);
2046 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2048 extern void __put_task_struct(struct task_struct *t);
2050 static inline void put_task_struct(struct task_struct *t)
2052 if (atomic_dec_and_test(&t->usage))
2053 __put_task_struct(t);
2056 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2057 extern void task_cputime(struct task_struct *t,
2058 cputime_t *utime, cputime_t *stime);
2059 extern void task_cputime_scaled(struct task_struct *t,
2060 cputime_t *utimescaled, cputime_t *stimescaled);
2061 extern cputime_t task_gtime(struct task_struct *t);
2063 static inline void task_cputime(struct task_struct *t,
2064 cputime_t *utime, cputime_t *stime)
2072 static inline void task_cputime_scaled(struct task_struct *t,
2073 cputime_t *utimescaled,
2074 cputime_t *stimescaled)
2077 *utimescaled = t->utimescaled;
2079 *stimescaled = t->stimescaled;
2082 static inline cputime_t task_gtime(struct task_struct *t)
2087 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2088 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2093 #define PF_EXITING 0x00000004 /* getting shut down */
2094 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2095 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2096 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2097 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2098 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2099 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2100 #define PF_DUMPCORE 0x00000200 /* dumped core */
2101 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2102 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2103 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2104 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2105 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2106 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2107 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2108 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2109 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2110 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2111 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2112 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2113 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2114 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2115 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2116 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2117 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2118 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2119 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2122 * Only the _current_ task can read/write to tsk->flags, but other
2123 * tasks can access tsk->flags in readonly mode for example
2124 * with tsk_used_math (like during threaded core dumping).
2125 * There is however an exception to this rule during ptrace
2126 * or during fork: the ptracer task is allowed to write to the
2127 * child->flags of its traced child (same goes for fork, the parent
2128 * can write to the child->flags), because we're guaranteed the
2129 * child is not running and in turn not changing child->flags
2130 * at the same time the parent does it.
2132 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2133 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2134 #define clear_used_math() clear_stopped_child_used_math(current)
2135 #define set_used_math() set_stopped_child_used_math(current)
2136 #define conditional_stopped_child_used_math(condition, child) \
2137 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2138 #define conditional_used_math(condition) \
2139 conditional_stopped_child_used_math(condition, current)
2140 #define copy_to_stopped_child_used_math(child) \
2141 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2142 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2143 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2144 #define used_math() tsk_used_math(current)
2146 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2147 * __GFP_FS is also cleared as it implies __GFP_IO.
2149 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2151 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2152 flags &= ~(__GFP_IO | __GFP_FS);
2156 static inline unsigned int memalloc_noio_save(void)
2158 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2159 current->flags |= PF_MEMALLOC_NOIO;
2163 static inline void memalloc_noio_restore(unsigned int flags)
2165 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2168 /* Per-process atomic flags. */
2169 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2170 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2171 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2174 #define TASK_PFA_TEST(name, func) \
2175 static inline bool task_##func(struct task_struct *p) \
2176 { return test_bit(PFA_##name, &p->atomic_flags); }
2177 #define TASK_PFA_SET(name, func) \
2178 static inline void task_set_##func(struct task_struct *p) \
2179 { set_bit(PFA_##name, &p->atomic_flags); }
2180 #define TASK_PFA_CLEAR(name, func) \
2181 static inline void task_clear_##func(struct task_struct *p) \
2182 { clear_bit(PFA_##name, &p->atomic_flags); }
2184 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2185 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2187 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2188 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2189 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2191 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2192 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2193 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2196 * task->jobctl flags
2198 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2200 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2201 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2202 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2203 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2204 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2205 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2206 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2208 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2209 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2210 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2211 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2212 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2213 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2214 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2216 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2217 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2219 extern bool task_set_jobctl_pending(struct task_struct *task,
2220 unsigned long mask);
2221 extern void task_clear_jobctl_trapping(struct task_struct *task);
2222 extern void task_clear_jobctl_pending(struct task_struct *task,
2223 unsigned long mask);
2225 static inline void rcu_copy_process(struct task_struct *p)
2227 #ifdef CONFIG_PREEMPT_RCU
2228 p->rcu_read_lock_nesting = 0;
2229 p->rcu_read_unlock_special.s = 0;
2230 p->rcu_blocked_node = NULL;
2231 INIT_LIST_HEAD(&p->rcu_node_entry);
2232 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2233 #ifdef CONFIG_TASKS_RCU
2234 p->rcu_tasks_holdout = false;
2235 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2236 p->rcu_tasks_idle_cpu = -1;
2237 #endif /* #ifdef CONFIG_TASKS_RCU */
2240 static inline void tsk_restore_flags(struct task_struct *task,
2241 unsigned long orig_flags, unsigned long flags)
2243 task->flags &= ~flags;
2244 task->flags |= orig_flags & flags;
2247 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2248 const struct cpumask *trial);
2249 extern int task_can_attach(struct task_struct *p,
2250 const struct cpumask *cs_cpus_allowed);
2252 extern void do_set_cpus_allowed(struct task_struct *p,
2253 const struct cpumask *new_mask);
2255 extern int set_cpus_allowed_ptr(struct task_struct *p,
2256 const struct cpumask *new_mask);
2258 static inline void do_set_cpus_allowed(struct task_struct *p,
2259 const struct cpumask *new_mask)
2262 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2263 const struct cpumask *new_mask)
2265 if (!cpumask_test_cpu(0, new_mask))
2271 #ifdef CONFIG_NO_HZ_COMMON
2272 void calc_load_enter_idle(void);
2273 void calc_load_exit_idle(void);
2275 static inline void calc_load_enter_idle(void) { }
2276 static inline void calc_load_exit_idle(void) { }
2277 #endif /* CONFIG_NO_HZ_COMMON */
2280 * Do not use outside of architecture code which knows its limitations.
2282 * sched_clock() has no promise of monotonicity or bounded drift between
2283 * CPUs, use (which you should not) requires disabling IRQs.
2285 * Please use one of the three interfaces below.
2287 extern unsigned long long notrace sched_clock(void);
2289 * See the comment in kernel/sched/clock.c
2291 extern u64 cpu_clock(int cpu);
2292 extern u64 local_clock(void);
2293 extern u64 running_clock(void);
2294 extern u64 sched_clock_cpu(int cpu);
2297 extern void sched_clock_init(void);
2299 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2300 static inline void sched_clock_tick(void)
2304 static inline void sched_clock_idle_sleep_event(void)
2308 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2313 * Architectures can set this to 1 if they have specified
2314 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2315 * but then during bootup it turns out that sched_clock()
2316 * is reliable after all:
2318 extern int sched_clock_stable(void);
2319 extern void set_sched_clock_stable(void);
2320 extern void clear_sched_clock_stable(void);
2322 extern void sched_clock_tick(void);
2323 extern void sched_clock_idle_sleep_event(void);
2324 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2327 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2329 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2330 * The reason for this explicit opt-in is not to have perf penalty with
2331 * slow sched_clocks.
2333 extern void enable_sched_clock_irqtime(void);
2334 extern void disable_sched_clock_irqtime(void);
2336 static inline void enable_sched_clock_irqtime(void) {}
2337 static inline void disable_sched_clock_irqtime(void) {}
2340 extern unsigned long long
2341 task_sched_runtime(struct task_struct *task);
2343 /* sched_exec is called by processes performing an exec */
2345 extern void sched_exec(void);
2347 #define sched_exec() {}
2350 extern void sched_clock_idle_sleep_event(void);
2351 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2353 #ifdef CONFIG_HOTPLUG_CPU
2354 extern void idle_task_exit(void);
2356 static inline void idle_task_exit(void) {}
2359 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2360 extern void wake_up_nohz_cpu(int cpu);
2362 static inline void wake_up_nohz_cpu(int cpu) { }
2365 #ifdef CONFIG_NO_HZ_FULL
2366 extern u64 scheduler_tick_max_deferment(void);
2369 #ifdef CONFIG_SCHED_AUTOGROUP
2370 extern void sched_autogroup_create_attach(struct task_struct *p);
2371 extern void sched_autogroup_detach(struct task_struct *p);
2372 extern void sched_autogroup_fork(struct signal_struct *sig);
2373 extern void sched_autogroup_exit(struct signal_struct *sig);
2374 #ifdef CONFIG_PROC_FS
2375 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2376 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2379 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2380 static inline void sched_autogroup_detach(struct task_struct *p) { }
2381 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2382 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2385 extern int yield_to(struct task_struct *p, bool preempt);
2386 extern void set_user_nice(struct task_struct *p, long nice);
2387 extern int task_prio(const struct task_struct *p);
2389 * task_nice - return the nice value of a given task.
2390 * @p: the task in question.
2392 * Return: The nice value [ -20 ... 0 ... 19 ].
2394 static inline int task_nice(const struct task_struct *p)
2396 return PRIO_TO_NICE((p)->static_prio);
2398 extern int can_nice(const struct task_struct *p, const int nice);
2399 extern int task_curr(const struct task_struct *p);
2400 extern int idle_cpu(int cpu);
2401 extern int sched_setscheduler(struct task_struct *, int,
2402 const struct sched_param *);
2403 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2404 const struct sched_param *);
2405 extern int sched_setattr(struct task_struct *,
2406 const struct sched_attr *);
2407 extern struct task_struct *idle_task(int cpu);
2409 * is_idle_task - is the specified task an idle task?
2410 * @p: the task in question.
2412 * Return: 1 if @p is an idle task. 0 otherwise.
2414 static inline bool is_idle_task(const struct task_struct *p)
2418 extern struct task_struct *curr_task(int cpu);
2419 extern void set_curr_task(int cpu, struct task_struct *p);
2423 union thread_union {
2424 struct thread_info thread_info;
2425 unsigned long stack[THREAD_SIZE/sizeof(long)];
2428 #ifndef __HAVE_ARCH_KSTACK_END
2429 static inline int kstack_end(void *addr)
2431 /* Reliable end of stack detection:
2432 * Some APM bios versions misalign the stack
2434 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2438 extern union thread_union init_thread_union;
2439 extern struct task_struct init_task;
2441 extern struct mm_struct init_mm;
2443 extern struct pid_namespace init_pid_ns;
2446 * find a task by one of its numerical ids
2448 * find_task_by_pid_ns():
2449 * finds a task by its pid in the specified namespace
2450 * find_task_by_vpid():
2451 * finds a task by its virtual pid
2453 * see also find_vpid() etc in include/linux/pid.h
2456 extern struct task_struct *find_task_by_vpid(pid_t nr);
2457 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2458 struct pid_namespace *ns);
2460 /* per-UID process charging. */
2461 extern struct user_struct * alloc_uid(kuid_t);
2462 static inline struct user_struct *get_uid(struct user_struct *u)
2464 atomic_inc(&u->__count);
2467 extern void free_uid(struct user_struct *);
2469 #include <asm/current.h>
2471 extern void xtime_update(unsigned long ticks);
2473 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2474 extern int wake_up_process(struct task_struct *tsk);
2475 extern void wake_up_new_task(struct task_struct *tsk);
2477 extern void kick_process(struct task_struct *tsk);
2479 static inline void kick_process(struct task_struct *tsk) { }
2481 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2482 extern void sched_dead(struct task_struct *p);
2484 extern void proc_caches_init(void);
2485 extern void flush_signals(struct task_struct *);
2486 extern void ignore_signals(struct task_struct *);
2487 extern void flush_signal_handlers(struct task_struct *, int force_default);
2488 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2490 static inline int kernel_dequeue_signal(siginfo_t *info)
2492 struct task_struct *tsk = current;
2496 spin_lock_irq(&tsk->sighand->siglock);
2497 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2498 spin_unlock_irq(&tsk->sighand->siglock);
2503 static inline void kernel_signal_stop(void)
2505 spin_lock_irq(¤t->sighand->siglock);
2506 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2507 __set_current_state(TASK_STOPPED);
2508 spin_unlock_irq(¤t->sighand->siglock);
2513 extern void release_task(struct task_struct * p);
2514 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2515 extern int force_sigsegv(int, struct task_struct *);
2516 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2517 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2518 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2519 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2520 const struct cred *, u32);
2521 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2522 extern int kill_pid(struct pid *pid, int sig, int priv);
2523 extern int kill_proc_info(int, struct siginfo *, pid_t);
2524 extern __must_check bool do_notify_parent(struct task_struct *, int);
2525 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2526 extern void force_sig(int, struct task_struct *);
2527 extern int send_sig(int, struct task_struct *, int);
2528 extern int zap_other_threads(struct task_struct *p);
2529 extern struct sigqueue *sigqueue_alloc(void);
2530 extern void sigqueue_free(struct sigqueue *);
2531 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2532 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2534 static inline void restore_saved_sigmask(void)
2536 if (test_and_clear_restore_sigmask())
2537 __set_current_blocked(¤t->saved_sigmask);
2540 static inline sigset_t *sigmask_to_save(void)
2542 sigset_t *res = ¤t->blocked;
2543 if (unlikely(test_restore_sigmask()))
2544 res = ¤t->saved_sigmask;
2548 static inline int kill_cad_pid(int sig, int priv)
2550 return kill_pid(cad_pid, sig, priv);
2553 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2554 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2555 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2556 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2559 * True if we are on the alternate signal stack.
2561 static inline int on_sig_stack(unsigned long sp)
2563 #ifdef CONFIG_STACK_GROWSUP
2564 return sp >= current->sas_ss_sp &&
2565 sp - current->sas_ss_sp < current->sas_ss_size;
2567 return sp > current->sas_ss_sp &&
2568 sp - current->sas_ss_sp <= current->sas_ss_size;
2572 static inline int sas_ss_flags(unsigned long sp)
2574 if (!current->sas_ss_size)
2577 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2580 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2582 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2583 #ifdef CONFIG_STACK_GROWSUP
2584 return current->sas_ss_sp;
2586 return current->sas_ss_sp + current->sas_ss_size;
2592 * Routines for handling mm_structs
2594 extern struct mm_struct * mm_alloc(void);
2596 /* mmdrop drops the mm and the page tables */
2597 extern void __mmdrop(struct mm_struct *);
2598 static inline void mmdrop(struct mm_struct * mm)
2600 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2604 /* mmput gets rid of the mappings and all user-space */
2605 extern void mmput(struct mm_struct *);
2606 /* Grab a reference to a task's mm, if it is not already going away */
2607 extern struct mm_struct *get_task_mm(struct task_struct *task);
2609 * Grab a reference to a task's mm, if it is not already going away
2610 * and ptrace_may_access with the mode parameter passed to it
2613 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2614 /* Remove the current tasks stale references to the old mm_struct */
2615 extern void mm_release(struct task_struct *, struct mm_struct *);
2617 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2618 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2619 struct task_struct *, unsigned long);
2621 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2622 struct task_struct *);
2624 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2625 * via pt_regs, so ignore the tls argument passed via C. */
2626 static inline int copy_thread_tls(
2627 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2628 struct task_struct *p, unsigned long tls)
2630 return copy_thread(clone_flags, sp, arg, p);
2633 extern void flush_thread(void);
2634 extern void exit_thread(void);
2636 extern void exit_files(struct task_struct *);
2637 extern void __cleanup_sighand(struct sighand_struct *);
2639 extern void exit_itimers(struct signal_struct *);
2640 extern void flush_itimer_signals(void);
2642 extern void do_group_exit(int);
2644 extern int do_execve(struct filename *,
2645 const char __user * const __user *,
2646 const char __user * const __user *);
2647 extern int do_execveat(int, struct filename *,
2648 const char __user * const __user *,
2649 const char __user * const __user *,
2651 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2652 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2653 struct task_struct *fork_idle(int);
2654 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2656 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2657 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2659 __set_task_comm(tsk, from, false);
2661 extern char *get_task_comm(char *to, struct task_struct *tsk);
2664 void scheduler_ipi(void);
2665 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2667 static inline void scheduler_ipi(void) { }
2668 static inline unsigned long wait_task_inactive(struct task_struct *p,
2675 #define tasklist_empty() \
2676 list_empty(&init_task.tasks)
2678 #define next_task(p) \
2679 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2681 #define for_each_process(p) \
2682 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2684 extern bool current_is_single_threaded(void);
2687 * Careful: do_each_thread/while_each_thread is a double loop so
2688 * 'break' will not work as expected - use goto instead.
2690 #define do_each_thread(g, t) \
2691 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2693 #define while_each_thread(g, t) \
2694 while ((t = next_thread(t)) != g)
2696 #define __for_each_thread(signal, t) \
2697 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2699 #define for_each_thread(p, t) \
2700 __for_each_thread((p)->signal, t)
2702 /* Careful: this is a double loop, 'break' won't work as expected. */
2703 #define for_each_process_thread(p, t) \
2704 for_each_process(p) for_each_thread(p, t)
2706 static inline int get_nr_threads(struct task_struct *tsk)
2708 return tsk->signal->nr_threads;
2711 static inline bool thread_group_leader(struct task_struct *p)
2713 return p->exit_signal >= 0;
2716 /* Do to the insanities of de_thread it is possible for a process
2717 * to have the pid of the thread group leader without actually being
2718 * the thread group leader. For iteration through the pids in proc
2719 * all we care about is that we have a task with the appropriate
2720 * pid, we don't actually care if we have the right task.
2722 static inline bool has_group_leader_pid(struct task_struct *p)
2724 return task_pid(p) == p->signal->leader_pid;
2728 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2730 return p1->signal == p2->signal;
2733 static inline struct task_struct *next_thread(const struct task_struct *p)
2735 return list_entry_rcu(p->thread_group.next,
2736 struct task_struct, thread_group);
2739 static inline int thread_group_empty(struct task_struct *p)
2741 return list_empty(&p->thread_group);
2744 #define delay_group_leader(p) \
2745 (thread_group_leader(p) && !thread_group_empty(p))
2748 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2749 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2750 * pins the final release of task.io_context. Also protects ->cpuset and
2751 * ->cgroup.subsys[]. And ->vfork_done.
2753 * Nests both inside and outside of read_lock(&tasklist_lock).
2754 * It must not be nested with write_lock_irq(&tasklist_lock),
2755 * neither inside nor outside.
2757 static inline void task_lock(struct task_struct *p)
2759 spin_lock(&p->alloc_lock);
2762 static inline void task_unlock(struct task_struct *p)
2764 spin_unlock(&p->alloc_lock);
2767 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2768 unsigned long *flags);
2770 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2771 unsigned long *flags)
2773 struct sighand_struct *ret;
2775 ret = __lock_task_sighand(tsk, flags);
2776 (void)__cond_lock(&tsk->sighand->siglock, ret);
2780 static inline void unlock_task_sighand(struct task_struct *tsk,
2781 unsigned long *flags)
2783 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2787 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2788 * @tsk: task causing the changes
2790 * All operations which modify a threadgroup - a new thread joining the
2791 * group, death of a member thread (the assertion of PF_EXITING) and
2792 * exec(2) dethreading the process and replacing the leader - are wrapped
2793 * by threadgroup_change_{begin|end}(). This is to provide a place which
2794 * subsystems needing threadgroup stability can hook into for
2797 static inline void threadgroup_change_begin(struct task_struct *tsk)
2800 cgroup_threadgroup_change_begin(tsk);
2804 * threadgroup_change_end - mark the end of changes to a threadgroup
2805 * @tsk: task causing the changes
2807 * See threadgroup_change_begin().
2809 static inline void threadgroup_change_end(struct task_struct *tsk)
2811 cgroup_threadgroup_change_end(tsk);
2814 #ifndef __HAVE_THREAD_FUNCTIONS
2816 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2817 #define task_stack_page(task) ((task)->stack)
2819 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2821 *task_thread_info(p) = *task_thread_info(org);
2822 task_thread_info(p)->task = p;
2826 * Return the address of the last usable long on the stack.
2828 * When the stack grows down, this is just above the thread
2829 * info struct. Going any lower will corrupt the threadinfo.
2831 * When the stack grows up, this is the highest address.
2832 * Beyond that position, we corrupt data on the next page.
2834 static inline unsigned long *end_of_stack(struct task_struct *p)
2836 #ifdef CONFIG_STACK_GROWSUP
2837 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2839 return (unsigned long *)(task_thread_info(p) + 1);
2844 #define task_stack_end_corrupted(task) \
2845 (*(end_of_stack(task)) != STACK_END_MAGIC)
2847 static inline int object_is_on_stack(void *obj)
2849 void *stack = task_stack_page(current);
2851 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2854 extern void thread_info_cache_init(void);
2856 #ifdef CONFIG_DEBUG_STACK_USAGE
2857 static inline unsigned long stack_not_used(struct task_struct *p)
2859 unsigned long *n = end_of_stack(p);
2861 do { /* Skip over canary */
2865 return (unsigned long)n - (unsigned long)end_of_stack(p);
2868 extern void set_task_stack_end_magic(struct task_struct *tsk);
2870 /* set thread flags in other task's structures
2871 * - see asm/thread_info.h for TIF_xxxx flags available
2873 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2875 set_ti_thread_flag(task_thread_info(tsk), flag);
2878 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2880 clear_ti_thread_flag(task_thread_info(tsk), flag);
2883 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2885 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2888 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2890 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2893 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2895 return test_ti_thread_flag(task_thread_info(tsk), flag);
2898 static inline void set_tsk_need_resched(struct task_struct *tsk)
2900 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2903 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2905 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2908 static inline int test_tsk_need_resched(struct task_struct *tsk)
2910 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2913 static inline int restart_syscall(void)
2915 set_tsk_thread_flag(current, TIF_SIGPENDING);
2916 return -ERESTARTNOINTR;
2919 static inline int signal_pending(struct task_struct *p)
2921 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2924 static inline int __fatal_signal_pending(struct task_struct *p)
2926 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2929 static inline int fatal_signal_pending(struct task_struct *p)
2931 return signal_pending(p) && __fatal_signal_pending(p);
2934 static inline int signal_pending_state(long state, struct task_struct *p)
2936 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2938 if (!signal_pending(p))
2941 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2945 * cond_resched() and cond_resched_lock(): latency reduction via
2946 * explicit rescheduling in places that are safe. The return
2947 * value indicates whether a reschedule was done in fact.
2948 * cond_resched_lock() will drop the spinlock before scheduling,
2949 * cond_resched_softirq() will enable bhs before scheduling.
2951 extern int _cond_resched(void);
2953 #define cond_resched() ({ \
2954 ___might_sleep(__FILE__, __LINE__, 0); \
2958 extern int __cond_resched_lock(spinlock_t *lock);
2960 #define cond_resched_lock(lock) ({ \
2961 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2962 __cond_resched_lock(lock); \
2965 extern int __cond_resched_softirq(void);
2967 #define cond_resched_softirq() ({ \
2968 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2969 __cond_resched_softirq(); \
2972 static inline void cond_resched_rcu(void)
2974 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2982 * Does a critical section need to be broken due to another
2983 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2984 * but a general need for low latency)
2986 static inline int spin_needbreak(spinlock_t *lock)
2988 #ifdef CONFIG_PREEMPT
2989 return spin_is_contended(lock);
2996 * Idle thread specific functions to determine the need_resched
2999 #ifdef TIF_POLLING_NRFLAG
3000 static inline int tsk_is_polling(struct task_struct *p)
3002 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3005 static inline void __current_set_polling(void)
3007 set_thread_flag(TIF_POLLING_NRFLAG);
3010 static inline bool __must_check current_set_polling_and_test(void)
3012 __current_set_polling();
3015 * Polling state must be visible before we test NEED_RESCHED,
3016 * paired by resched_curr()
3018 smp_mb__after_atomic();
3020 return unlikely(tif_need_resched());
3023 static inline void __current_clr_polling(void)
3025 clear_thread_flag(TIF_POLLING_NRFLAG);
3028 static inline bool __must_check current_clr_polling_and_test(void)
3030 __current_clr_polling();
3033 * Polling state must be visible before we test NEED_RESCHED,
3034 * paired by resched_curr()
3036 smp_mb__after_atomic();
3038 return unlikely(tif_need_resched());
3042 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3043 static inline void __current_set_polling(void) { }
3044 static inline void __current_clr_polling(void) { }
3046 static inline bool __must_check current_set_polling_and_test(void)
3048 return unlikely(tif_need_resched());
3050 static inline bool __must_check current_clr_polling_and_test(void)
3052 return unlikely(tif_need_resched());
3056 static inline void current_clr_polling(void)
3058 __current_clr_polling();
3061 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3062 * Once the bit is cleared, we'll get IPIs with every new
3063 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3066 smp_mb(); /* paired with resched_curr() */
3068 preempt_fold_need_resched();
3071 static __always_inline bool need_resched(void)
3073 return unlikely(tif_need_resched());
3077 * Thread group CPU time accounting.
3079 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3080 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3083 * Reevaluate whether the task has signals pending delivery.
3084 * Wake the task if so.
3085 * This is required every time the blocked sigset_t changes.
3086 * callers must hold sighand->siglock.
3088 extern void recalc_sigpending_and_wake(struct task_struct *t);
3089 extern void recalc_sigpending(void);
3091 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3093 static inline void signal_wake_up(struct task_struct *t, bool resume)
3095 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3097 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3099 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3103 * Wrappers for p->thread_info->cpu access. No-op on UP.
3107 static inline unsigned int task_cpu(const struct task_struct *p)
3109 return task_thread_info(p)->cpu;
3112 static inline int task_node(const struct task_struct *p)
3114 return cpu_to_node(task_cpu(p));
3117 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3121 static inline unsigned int task_cpu(const struct task_struct *p)
3126 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3130 #endif /* CONFIG_SMP */
3132 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3133 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3135 #ifdef CONFIG_CGROUP_SCHED
3136 extern struct task_group root_task_group;
3137 #endif /* CONFIG_CGROUP_SCHED */
3139 extern int task_can_switch_user(struct user_struct *up,
3140 struct task_struct *tsk);
3142 #ifdef CONFIG_TASK_XACCT
3143 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3145 tsk->ioac.rchar += amt;
3148 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3150 tsk->ioac.wchar += amt;
3153 static inline void inc_syscr(struct task_struct *tsk)
3158 static inline void inc_syscw(struct task_struct *tsk)
3163 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3167 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3171 static inline void inc_syscr(struct task_struct *tsk)
3175 static inline void inc_syscw(struct task_struct *tsk)
3180 #ifndef TASK_SIZE_OF
3181 #define TASK_SIZE_OF(tsk) TASK_SIZE
3185 extern void mm_update_next_owner(struct mm_struct *mm);
3187 static inline void mm_update_next_owner(struct mm_struct *mm)
3190 #endif /* CONFIG_MEMCG */
3192 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3195 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3198 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3201 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3204 static inline unsigned long rlimit(unsigned int limit)
3206 return task_rlimit(current, limit);
3209 static inline unsigned long rlimit_max(unsigned int limit)
3211 return task_rlimit_max(current, limit);
3214 #ifdef CONFIG_CPU_FREQ
3215 struct update_util_data {
3216 void (*func)(struct update_util_data *data,
3217 u64 time, unsigned long util, unsigned long max);
3220 void cpufreq_set_update_util_data(int cpu, struct update_util_data *data);
3221 #endif /* CONFIG_CPU_FREQ */