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(void);
182 static inline void update_cpu_load_nohz(void) { }
185 extern unsigned long get_parent_ip(unsigned long addr);
187 extern void dump_cpu_task(int cpu);
192 #ifdef CONFIG_SCHED_DEBUG
193 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
194 extern void proc_sched_set_task(struct task_struct *p);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_NOLOAD 1024
222 #define TASK_STATE_MAX 2048
224 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
226 extern char ___assert_task_state[1 - 2*!!(
227 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
229 /* Convenience macros for the sake of set_task_state */
230 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
231 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
232 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
234 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
236 /* Convenience macros for the sake of wake_up */
237 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
238 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
240 /* get_task_state() */
241 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
242 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
243 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
245 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
246 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
247 #define task_is_stopped_or_traced(task) \
248 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
249 #define task_contributes_to_load(task) \
250 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
251 (task->flags & PF_FROZEN) == 0 && \
252 (task->state & TASK_NOLOAD) == 0)
254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256 #define __set_task_state(tsk, state_value) \
258 (tsk)->task_state_change = _THIS_IP_; \
259 (tsk)->state = (state_value); \
261 #define set_task_state(tsk, state_value) \
263 (tsk)->task_state_change = _THIS_IP_; \
264 smp_store_mb((tsk)->state, (state_value)); \
268 * set_current_state() includes a barrier so that the write of current->state
269 * is correctly serialised wrt the caller's subsequent test of whether to
272 * set_current_state(TASK_UNINTERRUPTIBLE);
273 * if (do_i_need_to_sleep())
276 * If the caller does not need such serialisation then use __set_current_state()
278 #define __set_current_state(state_value) \
280 current->task_state_change = _THIS_IP_; \
281 current->state = (state_value); \
283 #define set_current_state(state_value) \
285 current->task_state_change = _THIS_IP_; \
286 smp_store_mb(current->state, (state_value)); \
291 #define __set_task_state(tsk, state_value) \
292 do { (tsk)->state = (state_value); } while (0)
293 #define set_task_state(tsk, state_value) \
294 smp_store_mb((tsk)->state, (state_value))
297 * set_current_state() includes a barrier so that the write of current->state
298 * is correctly serialised wrt the caller's subsequent test of whether to
301 * set_current_state(TASK_UNINTERRUPTIBLE);
302 * if (do_i_need_to_sleep())
305 * If the caller does not need such serialisation then use __set_current_state()
307 #define __set_current_state(state_value) \
308 do { current->state = (state_value); } while (0)
309 #define set_current_state(state_value) \
310 smp_store_mb(current->state, (state_value))
314 /* Task command name length */
315 #define TASK_COMM_LEN 16
317 #include <linux/spinlock.h>
320 * This serializes "schedule()" and also protects
321 * the run-queue from deletions/modifications (but
322 * _adding_ to the beginning of the run-queue has
325 extern rwlock_t tasklist_lock;
326 extern spinlock_t mmlist_lock;
330 #ifdef CONFIG_PROVE_RCU
331 extern int lockdep_tasklist_lock_is_held(void);
332 #endif /* #ifdef CONFIG_PROVE_RCU */
334 extern void sched_init(void);
335 extern void sched_init_smp(void);
336 extern asmlinkage void schedule_tail(struct task_struct *prev);
337 extern void init_idle(struct task_struct *idle, int cpu);
338 extern void init_idle_bootup_task(struct task_struct *idle);
340 extern cpumask_var_t cpu_isolated_map;
342 extern int runqueue_is_locked(int cpu);
344 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
345 extern void nohz_balance_enter_idle(int cpu);
346 extern void set_cpu_sd_state_idle(void);
347 extern int get_nohz_timer_target(void);
349 static inline void nohz_balance_enter_idle(int cpu) { }
350 static inline void set_cpu_sd_state_idle(void) { }
354 * Only dump TASK_* tasks. (0 for all tasks)
356 extern void show_state_filter(unsigned long state_filter);
358 static inline void show_state(void)
360 show_state_filter(0);
363 extern void show_regs(struct pt_regs *);
366 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
367 * task), SP is the stack pointer of the first frame that should be shown in the back
368 * trace (or NULL if the entire call-chain of the task should be shown).
370 extern void show_stack(struct task_struct *task, unsigned long *sp);
372 extern void cpu_init (void);
373 extern void trap_init(void);
374 extern void update_process_times(int user);
375 extern void scheduler_tick(void);
377 extern void sched_show_task(struct task_struct *p);
379 #ifdef CONFIG_LOCKUP_DETECTOR
380 extern void touch_softlockup_watchdog(void);
381 extern void touch_softlockup_watchdog_sync(void);
382 extern void touch_all_softlockup_watchdogs(void);
383 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
385 size_t *lenp, loff_t *ppos);
386 extern unsigned int softlockup_panic;
387 void lockup_detector_init(void);
389 static inline void touch_softlockup_watchdog(void)
392 static inline void touch_softlockup_watchdog_sync(void)
395 static inline void touch_all_softlockup_watchdogs(void)
398 static inline void lockup_detector_init(void)
403 #ifdef CONFIG_DETECT_HUNG_TASK
404 void reset_hung_task_detector(void);
406 static inline void reset_hung_task_detector(void)
411 /* Attach to any functions which should be ignored in wchan output. */
412 #define __sched __attribute__((__section__(".sched.text")))
414 /* Linker adds these: start and end of __sched functions */
415 extern char __sched_text_start[], __sched_text_end[];
417 /* Is this address in the __sched functions? */
418 extern int in_sched_functions(unsigned long addr);
420 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
421 extern signed long schedule_timeout(signed long timeout);
422 extern signed long schedule_timeout_interruptible(signed long timeout);
423 extern signed long schedule_timeout_killable(signed long timeout);
424 extern signed long schedule_timeout_uninterruptible(signed long timeout);
425 asmlinkage void schedule(void);
426 extern void schedule_preempt_disabled(void);
428 extern long io_schedule_timeout(long timeout);
430 static inline void io_schedule(void)
432 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
436 struct user_namespace;
439 extern void arch_pick_mmap_layout(struct mm_struct *mm);
441 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
442 unsigned long, unsigned long);
444 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
445 unsigned long len, unsigned long pgoff,
446 unsigned long flags);
448 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
451 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
452 #define SUID_DUMP_USER 1 /* Dump as user of process */
453 #define SUID_DUMP_ROOT 2 /* Dump as root */
457 /* for SUID_DUMP_* above */
458 #define MMF_DUMPABLE_BITS 2
459 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
461 extern void set_dumpable(struct mm_struct *mm, int value);
463 * This returns the actual value of the suid_dumpable flag. For things
464 * that are using this for checking for privilege transitions, it must
465 * test against SUID_DUMP_USER rather than treating it as a boolean
468 static inline int __get_dumpable(unsigned long mm_flags)
470 return mm_flags & MMF_DUMPABLE_MASK;
473 static inline int get_dumpable(struct mm_struct *mm)
475 return __get_dumpable(mm->flags);
478 /* coredump filter bits */
479 #define MMF_DUMP_ANON_PRIVATE 2
480 #define MMF_DUMP_ANON_SHARED 3
481 #define MMF_DUMP_MAPPED_PRIVATE 4
482 #define MMF_DUMP_MAPPED_SHARED 5
483 #define MMF_DUMP_ELF_HEADERS 6
484 #define MMF_DUMP_HUGETLB_PRIVATE 7
485 #define MMF_DUMP_HUGETLB_SHARED 8
487 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
488 #define MMF_DUMP_FILTER_BITS 7
489 #define MMF_DUMP_FILTER_MASK \
490 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
491 #define MMF_DUMP_FILTER_DEFAULT \
492 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
493 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
495 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
496 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
498 # define MMF_DUMP_MASK_DEFAULT_ELF 0
500 /* leave room for more dump flags */
501 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
502 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
503 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
505 #define MMF_HAS_UPROBES 19 /* has uprobes */
506 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
508 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
510 struct sighand_struct {
512 struct k_sigaction action[_NSIG];
514 wait_queue_head_t signalfd_wqh;
517 struct pacct_struct {
520 unsigned long ac_mem;
521 cputime_t ac_utime, ac_stime;
522 unsigned long ac_minflt, ac_majflt;
533 * struct prev_cputime - snaphsot of system and user cputime
534 * @utime: time spent in user mode
535 * @stime: time spent in system mode
536 * @lock: protects the above two fields
538 * Stores previous user/system time values such that we can guarantee
541 struct prev_cputime {
542 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
549 static inline void prev_cputime_init(struct prev_cputime *prev)
551 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
552 prev->utime = prev->stime = 0;
553 raw_spin_lock_init(&prev->lock);
558 * struct task_cputime - collected CPU time counts
559 * @utime: time spent in user mode, in &cputime_t units
560 * @stime: time spent in kernel mode, in &cputime_t units
561 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
563 * This structure groups together three kinds of CPU time that are tracked for
564 * threads and thread groups. Most things considering CPU time want to group
565 * these counts together and treat all three of them in parallel.
567 struct task_cputime {
570 unsigned long long sum_exec_runtime;
573 /* Alternate field names when used to cache expirations. */
574 #define virt_exp utime
575 #define prof_exp stime
576 #define sched_exp sum_exec_runtime
578 #define INIT_CPUTIME \
579 (struct task_cputime) { \
582 .sum_exec_runtime = 0, \
586 * This is the atomic variant of task_cputime, which can be used for
587 * storing and updating task_cputime statistics without locking.
589 struct task_cputime_atomic {
592 atomic64_t sum_exec_runtime;
595 #define INIT_CPUTIME_ATOMIC \
596 (struct task_cputime_atomic) { \
597 .utime = ATOMIC64_INIT(0), \
598 .stime = ATOMIC64_INIT(0), \
599 .sum_exec_runtime = ATOMIC64_INIT(0), \
602 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
605 * Disable preemption until the scheduler is running -- use an unconditional
606 * value so that it also works on !PREEMPT_COUNT kernels.
608 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
610 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
613 * Initial preempt_count value; reflects the preempt_count schedule invariant
614 * which states that during context switches:
616 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
618 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
619 * Note: See finish_task_switch().
621 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
624 * struct thread_group_cputimer - thread group interval timer counts
625 * @cputime_atomic: atomic thread group interval timers.
626 * @running: true when there are timers running and
627 * @cputime_atomic receives updates.
628 * @checking_timer: true when a thread in the group is in the
629 * process of checking for thread group timers.
631 * This structure contains the version of task_cputime, above, that is
632 * used for thread group CPU timer calculations.
634 struct thread_group_cputimer {
635 struct task_cputime_atomic cputime_atomic;
640 #include <linux/rwsem.h>
644 * NOTE! "signal_struct" does not have its own
645 * locking, because a shared signal_struct always
646 * implies a shared sighand_struct, so locking
647 * sighand_struct is always a proper superset of
648 * the locking of signal_struct.
650 struct signal_struct {
654 struct list_head thread_head;
656 wait_queue_head_t wait_chldexit; /* for wait4() */
658 /* current thread group signal load-balancing target: */
659 struct task_struct *curr_target;
661 /* shared signal handling: */
662 struct sigpending shared_pending;
664 /* thread group exit support */
667 * - notify group_exit_task when ->count is equal to notify_count
668 * - everyone except group_exit_task is stopped during signal delivery
669 * of fatal signals, group_exit_task processes the signal.
672 struct task_struct *group_exit_task;
674 /* thread group stop support, overloads group_exit_code too */
675 int group_stop_count;
676 unsigned int flags; /* see SIGNAL_* flags below */
679 * PR_SET_CHILD_SUBREAPER marks a process, like a service
680 * manager, to re-parent orphan (double-forking) child processes
681 * to this process instead of 'init'. The service manager is
682 * able to receive SIGCHLD signals and is able to investigate
683 * the process until it calls wait(). All children of this
684 * process will inherit a flag if they should look for a
685 * child_subreaper process at exit.
687 unsigned int is_child_subreaper:1;
688 unsigned int has_child_subreaper:1;
690 /* POSIX.1b Interval Timers */
692 struct list_head posix_timers;
694 /* ITIMER_REAL timer for the process */
695 struct hrtimer real_timer;
696 struct pid *leader_pid;
697 ktime_t it_real_incr;
700 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
701 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
702 * values are defined to 0 and 1 respectively
704 struct cpu_itimer it[2];
707 * Thread group totals for process CPU timers.
708 * See thread_group_cputimer(), et al, for details.
710 struct thread_group_cputimer cputimer;
712 /* Earliest-expiration cache. */
713 struct task_cputime cputime_expires;
715 struct list_head cpu_timers[3];
717 struct pid *tty_old_pgrp;
719 /* boolean value for session group leader */
722 struct tty_struct *tty; /* NULL if no tty */
724 #ifdef CONFIG_SCHED_AUTOGROUP
725 struct autogroup *autogroup;
728 * Cumulative resource counters for dead threads in the group,
729 * and for reaped dead child processes forked by this group.
730 * Live threads maintain their own counters and add to these
731 * in __exit_signal, except for the group leader.
733 seqlock_t stats_lock;
734 cputime_t utime, stime, cutime, cstime;
737 struct prev_cputime prev_cputime;
738 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
739 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
740 unsigned long inblock, oublock, cinblock, coublock;
741 unsigned long maxrss, cmaxrss;
742 struct task_io_accounting ioac;
745 * Cumulative ns of schedule CPU time fo dead threads in the
746 * group, not including a zombie group leader, (This only differs
747 * from jiffies_to_ns(utime + stime) if sched_clock uses something
748 * other than jiffies.)
750 unsigned long long sum_sched_runtime;
753 * We don't bother to synchronize most readers of this at all,
754 * because there is no reader checking a limit that actually needs
755 * to get both rlim_cur and rlim_max atomically, and either one
756 * alone is a single word that can safely be read normally.
757 * getrlimit/setrlimit use task_lock(current->group_leader) to
758 * protect this instead of the siglock, because they really
759 * have no need to disable irqs.
761 struct rlimit rlim[RLIM_NLIMITS];
763 #ifdef CONFIG_BSD_PROCESS_ACCT
764 struct pacct_struct pacct; /* per-process accounting information */
766 #ifdef CONFIG_TASKSTATS
767 struct taskstats *stats;
771 unsigned audit_tty_log_passwd;
772 struct tty_audit_buf *tty_audit_buf;
775 oom_flags_t oom_flags;
776 short oom_score_adj; /* OOM kill score adjustment */
777 short oom_score_adj_min; /* OOM kill score adjustment min value.
778 * Only settable by CAP_SYS_RESOURCE. */
780 struct mutex cred_guard_mutex; /* guard against foreign influences on
781 * credential calculations
782 * (notably. ptrace) */
786 * Bits in flags field of signal_struct.
788 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
789 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
790 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
791 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
793 * Pending notifications to parent.
795 #define SIGNAL_CLD_STOPPED 0x00000010
796 #define SIGNAL_CLD_CONTINUED 0x00000020
797 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
799 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
801 /* If true, all threads except ->group_exit_task have pending SIGKILL */
802 static inline int signal_group_exit(const struct signal_struct *sig)
804 return (sig->flags & SIGNAL_GROUP_EXIT) ||
805 (sig->group_exit_task != NULL);
809 * Some day this will be a full-fledged user tracking system..
812 atomic_t __count; /* reference count */
813 atomic_t processes; /* How many processes does this user have? */
814 atomic_t sigpending; /* How many pending signals does this user have? */
815 #ifdef CONFIG_INOTIFY_USER
816 atomic_t inotify_watches; /* How many inotify watches does this user have? */
817 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
819 #ifdef CONFIG_FANOTIFY
820 atomic_t fanotify_listeners;
823 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
825 #ifdef CONFIG_POSIX_MQUEUE
826 /* protected by mq_lock */
827 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
829 unsigned long locked_shm; /* How many pages of mlocked shm ? */
832 struct key *uid_keyring; /* UID specific keyring */
833 struct key *session_keyring; /* UID's default session keyring */
836 /* Hash table maintenance information */
837 struct hlist_node uidhash_node;
840 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
841 atomic_long_t locked_vm;
845 extern int uids_sysfs_init(void);
847 extern struct user_struct *find_user(kuid_t);
849 extern struct user_struct root_user;
850 #define INIT_USER (&root_user)
853 struct backing_dev_info;
854 struct reclaim_state;
856 #ifdef CONFIG_SCHED_INFO
858 /* cumulative counters */
859 unsigned long pcount; /* # of times run on this cpu */
860 unsigned long long run_delay; /* time spent waiting on a runqueue */
863 unsigned long long last_arrival,/* when we last ran on a cpu */
864 last_queued; /* when we were last queued to run */
866 #endif /* CONFIG_SCHED_INFO */
868 #ifdef CONFIG_TASK_DELAY_ACCT
869 struct task_delay_info {
871 unsigned int flags; /* Private per-task flags */
873 /* For each stat XXX, add following, aligned appropriately
875 * struct timespec XXX_start, XXX_end;
879 * Atomicity of updates to XXX_delay, XXX_count protected by
880 * single lock above (split into XXX_lock if contention is an issue).
884 * XXX_count is incremented on every XXX operation, the delay
885 * associated with the operation is added to XXX_delay.
886 * XXX_delay contains the accumulated delay time in nanoseconds.
888 u64 blkio_start; /* Shared by blkio, swapin */
889 u64 blkio_delay; /* wait for sync block io completion */
890 u64 swapin_delay; /* wait for swapin block io completion */
891 u32 blkio_count; /* total count of the number of sync block */
892 /* io operations performed */
893 u32 swapin_count; /* total count of the number of swapin block */
894 /* io operations performed */
897 u64 freepages_delay; /* wait for memory reclaim */
898 u32 freepages_count; /* total count of memory reclaim */
900 #endif /* CONFIG_TASK_DELAY_ACCT */
902 static inline int sched_info_on(void)
904 #ifdef CONFIG_SCHEDSTATS
906 #elif defined(CONFIG_TASK_DELAY_ACCT)
907 extern int delayacct_on;
922 * Increase resolution of cpu_capacity calculations
924 #define SCHED_CAPACITY_SHIFT 10
925 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
928 * Wake-queues are lists of tasks with a pending wakeup, whose
929 * callers have already marked the task as woken internally,
930 * and can thus carry on. A common use case is being able to
931 * do the wakeups once the corresponding user lock as been
934 * We hold reference to each task in the list across the wakeup,
935 * thus guaranteeing that the memory is still valid by the time
936 * the actual wakeups are performed in wake_up_q().
938 * One per task suffices, because there's never a need for a task to be
939 * in two wake queues simultaneously; it is forbidden to abandon a task
940 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
941 * already in a wake queue, the wakeup will happen soon and the second
942 * waker can just skip it.
944 * The WAKE_Q macro declares and initializes the list head.
945 * wake_up_q() does NOT reinitialize the list; it's expected to be
946 * called near the end of a function, where the fact that the queue is
947 * not used again will be easy to see by inspection.
949 * Note that this can cause spurious wakeups. schedule() callers
950 * must ensure the call is done inside a loop, confirming that the
951 * wakeup condition has in fact occurred.
954 struct wake_q_node *next;
958 struct wake_q_node *first;
959 struct wake_q_node **lastp;
962 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
964 #define WAKE_Q(name) \
965 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
967 extern void wake_q_add(struct wake_q_head *head,
968 struct task_struct *task);
969 extern void wake_up_q(struct wake_q_head *head);
972 * sched-domains (multiprocessor balancing) declarations:
975 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
976 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
977 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
978 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
979 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
980 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
981 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
982 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
983 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
984 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
985 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
986 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
987 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
988 #define SD_NUMA 0x4000 /* cross-node balancing */
990 #ifdef CONFIG_SCHED_SMT
991 static inline int cpu_smt_flags(void)
993 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
997 #ifdef CONFIG_SCHED_MC
998 static inline int cpu_core_flags(void)
1000 return SD_SHARE_PKG_RESOURCES;
1005 static inline int cpu_numa_flags(void)
1011 struct sched_domain_attr {
1012 int relax_domain_level;
1015 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1016 .relax_domain_level = -1, \
1019 extern int sched_domain_level_max;
1023 struct sched_domain {
1024 /* These fields must be setup */
1025 struct sched_domain *parent; /* top domain must be null terminated */
1026 struct sched_domain *child; /* bottom domain must be null terminated */
1027 struct sched_group *groups; /* the balancing groups of the domain */
1028 unsigned long min_interval; /* Minimum balance interval ms */
1029 unsigned long max_interval; /* Maximum balance interval ms */
1030 unsigned int busy_factor; /* less balancing by factor if busy */
1031 unsigned int imbalance_pct; /* No balance until over watermark */
1032 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1033 unsigned int busy_idx;
1034 unsigned int idle_idx;
1035 unsigned int newidle_idx;
1036 unsigned int wake_idx;
1037 unsigned int forkexec_idx;
1038 unsigned int smt_gain;
1040 int nohz_idle; /* NOHZ IDLE status */
1041 int flags; /* See SD_* */
1044 /* Runtime fields. */
1045 unsigned long last_balance; /* init to jiffies. units in jiffies */
1046 unsigned int balance_interval; /* initialise to 1. units in ms. */
1047 unsigned int nr_balance_failed; /* initialise to 0 */
1049 /* idle_balance() stats */
1050 u64 max_newidle_lb_cost;
1051 unsigned long next_decay_max_lb_cost;
1053 #ifdef CONFIG_SCHEDSTATS
1054 /* load_balance() stats */
1055 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1056 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1057 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1058 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1059 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1060 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1061 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1062 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1064 /* Active load balancing */
1065 unsigned int alb_count;
1066 unsigned int alb_failed;
1067 unsigned int alb_pushed;
1069 /* SD_BALANCE_EXEC stats */
1070 unsigned int sbe_count;
1071 unsigned int sbe_balanced;
1072 unsigned int sbe_pushed;
1074 /* SD_BALANCE_FORK stats */
1075 unsigned int sbf_count;
1076 unsigned int sbf_balanced;
1077 unsigned int sbf_pushed;
1079 /* try_to_wake_up() stats */
1080 unsigned int ttwu_wake_remote;
1081 unsigned int ttwu_move_affine;
1082 unsigned int ttwu_move_balance;
1084 #ifdef CONFIG_SCHED_DEBUG
1088 void *private; /* used during construction */
1089 struct rcu_head rcu; /* used during destruction */
1092 unsigned int span_weight;
1094 * Span of all CPUs in this domain.
1096 * NOTE: this field is variable length. (Allocated dynamically
1097 * by attaching extra space to the end of the structure,
1098 * depending on how many CPUs the kernel has booted up with)
1100 unsigned long span[0];
1103 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1105 return to_cpumask(sd->span);
1108 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1109 struct sched_domain_attr *dattr_new);
1111 /* Allocate an array of sched domains, for partition_sched_domains(). */
1112 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1113 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1115 bool cpus_share_cache(int this_cpu, int that_cpu);
1117 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1118 typedef int (*sched_domain_flags_f)(void);
1120 #define SDTL_OVERLAP 0x01
1123 struct sched_domain **__percpu sd;
1124 struct sched_group **__percpu sg;
1125 struct sched_group_capacity **__percpu sgc;
1128 struct sched_domain_topology_level {
1129 sched_domain_mask_f mask;
1130 sched_domain_flags_f sd_flags;
1133 struct sd_data data;
1134 #ifdef CONFIG_SCHED_DEBUG
1139 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1140 extern void wake_up_if_idle(int cpu);
1142 #ifdef CONFIG_SCHED_DEBUG
1143 # define SD_INIT_NAME(type) .name = #type
1145 # define SD_INIT_NAME(type)
1148 #else /* CONFIG_SMP */
1150 struct sched_domain_attr;
1153 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1154 struct sched_domain_attr *dattr_new)
1158 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1163 #endif /* !CONFIG_SMP */
1166 struct io_context; /* See blkdev.h */
1169 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1170 extern void prefetch_stack(struct task_struct *t);
1172 static inline void prefetch_stack(struct task_struct *t) { }
1175 struct audit_context; /* See audit.c */
1177 struct pipe_inode_info;
1178 struct uts_namespace;
1180 struct load_weight {
1181 unsigned long weight;
1186 * The load_avg/util_avg accumulates an infinite geometric series.
1187 * 1) load_avg factors frequency scaling into the amount of time that a
1188 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1189 * aggregated such weights of all runnable and blocked sched_entities.
1190 * 2) util_avg factors frequency and cpu scaling into the amount of time
1191 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1192 * For cfs_rq, it is the aggregated such times of all runnable and
1193 * blocked sched_entities.
1194 * The 64 bit load_sum can:
1195 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1196 * the highest weight (=88761) always runnable, we should not overflow
1197 * 2) for entity, support any load.weight always runnable
1200 u64 last_update_time, load_sum;
1201 u32 util_sum, period_contrib;
1202 unsigned long load_avg, util_avg;
1205 #ifdef CONFIG_SCHEDSTATS
1206 struct sched_statistics {
1216 s64 sum_sleep_runtime;
1223 u64 nr_migrations_cold;
1224 u64 nr_failed_migrations_affine;
1225 u64 nr_failed_migrations_running;
1226 u64 nr_failed_migrations_hot;
1227 u64 nr_forced_migrations;
1230 u64 nr_wakeups_sync;
1231 u64 nr_wakeups_migrate;
1232 u64 nr_wakeups_local;
1233 u64 nr_wakeups_remote;
1234 u64 nr_wakeups_affine;
1235 u64 nr_wakeups_affine_attempts;
1236 u64 nr_wakeups_passive;
1237 u64 nr_wakeups_idle;
1241 struct sched_entity {
1242 struct load_weight load; /* for load-balancing */
1243 struct rb_node run_node;
1244 struct list_head group_node;
1248 u64 sum_exec_runtime;
1250 u64 prev_sum_exec_runtime;
1254 #ifdef CONFIG_SCHEDSTATS
1255 struct sched_statistics statistics;
1258 #ifdef CONFIG_FAIR_GROUP_SCHED
1260 struct sched_entity *parent;
1261 /* rq on which this entity is (to be) queued: */
1262 struct cfs_rq *cfs_rq;
1263 /* rq "owned" by this entity/group: */
1264 struct cfs_rq *my_q;
1268 /* Per entity load average tracking */
1269 struct sched_avg avg;
1273 struct sched_rt_entity {
1274 struct list_head run_list;
1275 unsigned long timeout;
1276 unsigned long watchdog_stamp;
1277 unsigned int time_slice;
1279 struct sched_rt_entity *back;
1280 #ifdef CONFIG_RT_GROUP_SCHED
1281 struct sched_rt_entity *parent;
1282 /* rq on which this entity is (to be) queued: */
1283 struct rt_rq *rt_rq;
1284 /* rq "owned" by this entity/group: */
1289 struct sched_dl_entity {
1290 struct rb_node rb_node;
1293 * Original scheduling parameters. Copied here from sched_attr
1294 * during sched_setattr(), they will remain the same until
1295 * the next sched_setattr().
1297 u64 dl_runtime; /* maximum runtime for each instance */
1298 u64 dl_deadline; /* relative deadline of each instance */
1299 u64 dl_period; /* separation of two instances (period) */
1300 u64 dl_bw; /* dl_runtime / dl_deadline */
1303 * Actual scheduling parameters. Initialized with the values above,
1304 * they are continously updated during task execution. Note that
1305 * the remaining runtime could be < 0 in case we are in overrun.
1307 s64 runtime; /* remaining runtime for this instance */
1308 u64 deadline; /* absolute deadline for this instance */
1309 unsigned int flags; /* specifying the scheduler behaviour */
1314 * @dl_throttled tells if we exhausted the runtime. If so, the
1315 * task has to wait for a replenishment to be performed at the
1316 * next firing of dl_timer.
1318 * @dl_new tells if a new instance arrived. If so we must
1319 * start executing it with full runtime and reset its absolute
1322 * @dl_boosted tells if we are boosted due to DI. If so we are
1323 * outside bandwidth enforcement mechanism (but only until we
1324 * exit the critical section);
1326 * @dl_yielded tells if task gave up the cpu before consuming
1327 * all its available runtime during the last job.
1329 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1332 * Bandwidth enforcement timer. Each -deadline task has its
1333 * own bandwidth to be enforced, thus we need one timer per task.
1335 struct hrtimer dl_timer;
1343 u8 pad; /* Otherwise the compiler can store garbage here. */
1345 u32 s; /* Set of bits. */
1349 enum perf_event_task_context {
1350 perf_invalid_context = -1,
1351 perf_hw_context = 0,
1353 perf_nr_task_contexts,
1356 /* Track pages that require TLB flushes */
1357 struct tlbflush_unmap_batch {
1359 * Each bit set is a CPU that potentially has a TLB entry for one of
1360 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1362 struct cpumask cpumask;
1364 /* True if any bit in cpumask is set */
1365 bool flush_required;
1368 * If true then the PTE was dirty when unmapped. The entry must be
1369 * flushed before IO is initiated or a stale TLB entry potentially
1370 * allows an update without redirtying the page.
1375 struct task_struct {
1376 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1379 unsigned int flags; /* per process flags, defined below */
1380 unsigned int ptrace;
1383 struct llist_node wake_entry;
1385 unsigned int wakee_flips;
1386 unsigned long wakee_flip_decay_ts;
1387 struct task_struct *last_wakee;
1393 int prio, static_prio, normal_prio;
1394 unsigned int rt_priority;
1395 const struct sched_class *sched_class;
1396 struct sched_entity se;
1397 struct sched_rt_entity rt;
1398 #ifdef CONFIG_CGROUP_SCHED
1399 struct task_group *sched_task_group;
1401 struct sched_dl_entity dl;
1403 #ifdef CONFIG_PREEMPT_NOTIFIERS
1404 /* list of struct preempt_notifier: */
1405 struct hlist_head preempt_notifiers;
1408 #ifdef CONFIG_BLK_DEV_IO_TRACE
1409 unsigned int btrace_seq;
1412 unsigned int policy;
1413 int nr_cpus_allowed;
1414 cpumask_t cpus_allowed;
1416 #ifdef CONFIG_PREEMPT_RCU
1417 int rcu_read_lock_nesting;
1418 union rcu_special rcu_read_unlock_special;
1419 struct list_head rcu_node_entry;
1420 struct rcu_node *rcu_blocked_node;
1421 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1422 #ifdef CONFIG_TASKS_RCU
1423 unsigned long rcu_tasks_nvcsw;
1424 bool rcu_tasks_holdout;
1425 struct list_head rcu_tasks_holdout_list;
1426 int rcu_tasks_idle_cpu;
1427 #endif /* #ifdef CONFIG_TASKS_RCU */
1429 #ifdef CONFIG_SCHED_INFO
1430 struct sched_info sched_info;
1433 struct list_head tasks;
1435 struct plist_node pushable_tasks;
1436 struct rb_node pushable_dl_tasks;
1439 struct mm_struct *mm, *active_mm;
1440 /* per-thread vma caching */
1441 u32 vmacache_seqnum;
1442 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1443 #if defined(SPLIT_RSS_COUNTING)
1444 struct task_rss_stat rss_stat;
1448 int exit_code, exit_signal;
1449 int pdeath_signal; /* The signal sent when the parent dies */
1450 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1452 /* Used for emulating ABI behavior of previous Linux versions */
1453 unsigned int personality;
1455 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1457 unsigned in_iowait:1;
1459 /* Revert to default priority/policy when forking */
1460 unsigned sched_reset_on_fork:1;
1461 unsigned sched_contributes_to_load:1;
1462 unsigned sched_migrated:1;
1464 #ifdef CONFIG_MEMCG_KMEM
1465 unsigned memcg_kmem_skip_account:1;
1467 #ifdef CONFIG_COMPAT_BRK
1468 unsigned brk_randomized:1;
1471 unsigned long atomic_flags; /* Flags needing atomic access. */
1473 struct restart_block restart_block;
1478 #ifdef CONFIG_CC_STACKPROTECTOR
1479 /* Canary value for the -fstack-protector gcc feature */
1480 unsigned long stack_canary;
1483 * pointers to (original) parent process, youngest child, younger sibling,
1484 * older sibling, respectively. (p->father can be replaced with
1485 * p->real_parent->pid)
1487 struct task_struct __rcu *real_parent; /* real parent process */
1488 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1490 * children/sibling forms the list of my natural children
1492 struct list_head children; /* list of my children */
1493 struct list_head sibling; /* linkage in my parent's children list */
1494 struct task_struct *group_leader; /* threadgroup leader */
1497 * ptraced is the list of tasks this task is using ptrace on.
1498 * This includes both natural children and PTRACE_ATTACH targets.
1499 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1501 struct list_head ptraced;
1502 struct list_head ptrace_entry;
1504 /* PID/PID hash table linkage. */
1505 struct pid_link pids[PIDTYPE_MAX];
1506 struct list_head thread_group;
1507 struct list_head thread_node;
1509 struct completion *vfork_done; /* for vfork() */
1510 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1511 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1513 cputime_t utime, stime, utimescaled, stimescaled;
1515 struct prev_cputime prev_cputime;
1516 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1517 seqlock_t vtime_seqlock;
1518 unsigned long long vtime_snap;
1523 } vtime_snap_whence;
1525 unsigned long nvcsw, nivcsw; /* context switch counts */
1526 u64 start_time; /* monotonic time in nsec */
1527 u64 real_start_time; /* boot based time in nsec */
1528 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1529 unsigned long min_flt, maj_flt;
1531 struct task_cputime cputime_expires;
1532 struct list_head cpu_timers[3];
1534 /* process credentials */
1535 const struct cred __rcu *real_cred; /* objective and real subjective task
1536 * credentials (COW) */
1537 const struct cred __rcu *cred; /* effective (overridable) subjective task
1538 * credentials (COW) */
1539 char comm[TASK_COMM_LEN]; /* executable name excluding path
1540 - access with [gs]et_task_comm (which lock
1541 it with task_lock())
1542 - initialized normally by setup_new_exec */
1543 /* file system info */
1544 struct nameidata *nameidata;
1545 #ifdef CONFIG_SYSVIPC
1547 struct sysv_sem sysvsem;
1548 struct sysv_shm sysvshm;
1550 #ifdef CONFIG_DETECT_HUNG_TASK
1551 /* hung task detection */
1552 unsigned long last_switch_count;
1554 /* filesystem information */
1555 struct fs_struct *fs;
1556 /* open file information */
1557 struct files_struct *files;
1559 struct nsproxy *nsproxy;
1560 /* signal handlers */
1561 struct signal_struct *signal;
1562 struct sighand_struct *sighand;
1564 sigset_t blocked, real_blocked;
1565 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1566 struct sigpending pending;
1568 unsigned long sas_ss_sp;
1570 int (*notifier)(void *priv);
1571 void *notifier_data;
1572 sigset_t *notifier_mask;
1573 struct callback_head *task_works;
1575 struct audit_context *audit_context;
1576 #ifdef CONFIG_AUDITSYSCALL
1578 unsigned int sessionid;
1580 struct seccomp seccomp;
1582 /* Thread group tracking */
1585 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1587 spinlock_t alloc_lock;
1589 /* Protection of the PI data structures: */
1590 raw_spinlock_t pi_lock;
1592 struct wake_q_node wake_q;
1594 #ifdef CONFIG_RT_MUTEXES
1595 /* PI waiters blocked on a rt_mutex held by this task */
1596 struct rb_root pi_waiters;
1597 struct rb_node *pi_waiters_leftmost;
1598 /* Deadlock detection and priority inheritance handling */
1599 struct rt_mutex_waiter *pi_blocked_on;
1602 #ifdef CONFIG_DEBUG_MUTEXES
1603 /* mutex deadlock detection */
1604 struct mutex_waiter *blocked_on;
1606 #ifdef CONFIG_TRACE_IRQFLAGS
1607 unsigned int irq_events;
1608 unsigned long hardirq_enable_ip;
1609 unsigned long hardirq_disable_ip;
1610 unsigned int hardirq_enable_event;
1611 unsigned int hardirq_disable_event;
1612 int hardirqs_enabled;
1613 int hardirq_context;
1614 unsigned long softirq_disable_ip;
1615 unsigned long softirq_enable_ip;
1616 unsigned int softirq_disable_event;
1617 unsigned int softirq_enable_event;
1618 int softirqs_enabled;
1619 int softirq_context;
1621 #ifdef CONFIG_LOCKDEP
1622 # define MAX_LOCK_DEPTH 48UL
1625 unsigned int lockdep_recursion;
1626 struct held_lock held_locks[MAX_LOCK_DEPTH];
1627 gfp_t lockdep_reclaim_gfp;
1630 /* journalling filesystem info */
1633 /* stacked block device info */
1634 struct bio_list *bio_list;
1637 /* stack plugging */
1638 struct blk_plug *plug;
1642 struct reclaim_state *reclaim_state;
1644 struct backing_dev_info *backing_dev_info;
1646 struct io_context *io_context;
1648 unsigned long ptrace_message;
1649 siginfo_t *last_siginfo; /* For ptrace use. */
1650 struct task_io_accounting ioac;
1651 #if defined(CONFIG_TASK_XACCT)
1652 u64 acct_rss_mem1; /* accumulated rss usage */
1653 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1654 cputime_t acct_timexpd; /* stime + utime since last update */
1656 #ifdef CONFIG_CPUSETS
1657 nodemask_t mems_allowed; /* Protected by alloc_lock */
1658 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1659 int cpuset_mem_spread_rotor;
1660 int cpuset_slab_spread_rotor;
1662 #ifdef CONFIG_CGROUPS
1663 /* Control Group info protected by css_set_lock */
1664 struct css_set __rcu *cgroups;
1665 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1666 struct list_head cg_list;
1669 struct robust_list_head __user *robust_list;
1670 #ifdef CONFIG_COMPAT
1671 struct compat_robust_list_head __user *compat_robust_list;
1673 struct list_head pi_state_list;
1674 struct futex_pi_state *pi_state_cache;
1676 #ifdef CONFIG_PERF_EVENTS
1677 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1678 struct mutex perf_event_mutex;
1679 struct list_head perf_event_list;
1681 #ifdef CONFIG_DEBUG_PREEMPT
1682 unsigned long preempt_disable_ip;
1685 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1687 short pref_node_fork;
1689 #ifdef CONFIG_NUMA_BALANCING
1691 unsigned int numa_scan_period;
1692 unsigned int numa_scan_period_max;
1693 int numa_preferred_nid;
1694 unsigned long numa_migrate_retry;
1695 u64 node_stamp; /* migration stamp */
1696 u64 last_task_numa_placement;
1697 u64 last_sum_exec_runtime;
1698 struct callback_head numa_work;
1700 struct list_head numa_entry;
1701 struct numa_group *numa_group;
1704 * numa_faults is an array split into four regions:
1705 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1706 * in this precise order.
1708 * faults_memory: Exponential decaying average of faults on a per-node
1709 * basis. Scheduling placement decisions are made based on these
1710 * counts. The values remain static for the duration of a PTE scan.
1711 * faults_cpu: Track the nodes the process was running on when a NUMA
1712 * hinting fault was incurred.
1713 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1714 * during the current scan window. When the scan completes, the counts
1715 * in faults_memory and faults_cpu decay and these values are copied.
1717 unsigned long *numa_faults;
1718 unsigned long total_numa_faults;
1721 * numa_faults_locality tracks if faults recorded during the last
1722 * scan window were remote/local or failed to migrate. The task scan
1723 * period is adapted based on the locality of the faults with different
1724 * weights depending on whether they were shared or private faults
1726 unsigned long numa_faults_locality[3];
1728 unsigned long numa_pages_migrated;
1729 #endif /* CONFIG_NUMA_BALANCING */
1731 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1732 struct tlbflush_unmap_batch tlb_ubc;
1735 struct rcu_head rcu;
1738 * cache last used pipe for splice
1740 struct pipe_inode_info *splice_pipe;
1742 struct page_frag task_frag;
1744 #ifdef CONFIG_TASK_DELAY_ACCT
1745 struct task_delay_info *delays;
1747 #ifdef CONFIG_FAULT_INJECTION
1751 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1752 * balance_dirty_pages() for some dirty throttling pause
1755 int nr_dirtied_pause;
1756 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1758 #ifdef CONFIG_LATENCYTOP
1759 int latency_record_count;
1760 struct latency_record latency_record[LT_SAVECOUNT];
1763 * time slack values; these are used to round up poll() and
1764 * select() etc timeout values. These are in nanoseconds.
1766 unsigned long timer_slack_ns;
1767 unsigned long default_timer_slack_ns;
1770 unsigned int kasan_depth;
1772 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1773 /* Index of current stored address in ret_stack */
1775 /* Stack of return addresses for return function tracing */
1776 struct ftrace_ret_stack *ret_stack;
1777 /* time stamp for last schedule */
1778 unsigned long long ftrace_timestamp;
1780 * Number of functions that haven't been traced
1781 * because of depth overrun.
1783 atomic_t trace_overrun;
1784 /* Pause for the tracing */
1785 atomic_t tracing_graph_pause;
1787 #ifdef CONFIG_TRACING
1788 /* state flags for use by tracers */
1789 unsigned long trace;
1790 /* bitmask and counter of trace recursion */
1791 unsigned long trace_recursion;
1792 #endif /* CONFIG_TRACING */
1794 struct memcg_oom_info {
1795 struct mem_cgroup *memcg;
1798 unsigned int may_oom:1;
1801 #ifdef CONFIG_UPROBES
1802 struct uprobe_task *utask;
1804 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1805 unsigned int sequential_io;
1806 unsigned int sequential_io_avg;
1808 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1809 unsigned long task_state_change;
1811 int pagefault_disabled;
1812 /* CPU-specific state of this task */
1813 struct thread_struct thread;
1815 * WARNING: on x86, 'thread_struct' contains a variable-sized
1816 * structure. It *MUST* be at the end of 'task_struct'.
1818 * Do not put anything below here!
1822 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1823 extern int arch_task_struct_size __read_mostly;
1825 # define arch_task_struct_size (sizeof(struct task_struct))
1828 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1829 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1831 #define TNF_MIGRATED 0x01
1832 #define TNF_NO_GROUP 0x02
1833 #define TNF_SHARED 0x04
1834 #define TNF_FAULT_LOCAL 0x08
1835 #define TNF_MIGRATE_FAIL 0x10
1837 #ifdef CONFIG_NUMA_BALANCING
1838 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1839 extern pid_t task_numa_group_id(struct task_struct *p);
1840 extern void set_numabalancing_state(bool enabled);
1841 extern void task_numa_free(struct task_struct *p);
1842 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1843 int src_nid, int dst_cpu);
1845 static inline void task_numa_fault(int last_node, int node, int pages,
1849 static inline pid_t task_numa_group_id(struct task_struct *p)
1853 static inline void set_numabalancing_state(bool enabled)
1856 static inline void task_numa_free(struct task_struct *p)
1859 static inline bool should_numa_migrate_memory(struct task_struct *p,
1860 struct page *page, int src_nid, int dst_cpu)
1866 static inline struct pid *task_pid(struct task_struct *task)
1868 return task->pids[PIDTYPE_PID].pid;
1871 static inline struct pid *task_tgid(struct task_struct *task)
1873 return task->group_leader->pids[PIDTYPE_PID].pid;
1877 * Without tasklist or rcu lock it is not safe to dereference
1878 * the result of task_pgrp/task_session even if task == current,
1879 * we can race with another thread doing sys_setsid/sys_setpgid.
1881 static inline struct pid *task_pgrp(struct task_struct *task)
1883 return task->group_leader->pids[PIDTYPE_PGID].pid;
1886 static inline struct pid *task_session(struct task_struct *task)
1888 return task->group_leader->pids[PIDTYPE_SID].pid;
1891 struct pid_namespace;
1894 * the helpers to get the task's different pids as they are seen
1895 * from various namespaces
1897 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1898 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1900 * task_xid_nr_ns() : id seen from the ns specified;
1902 * set_task_vxid() : assigns a virtual id to a task;
1904 * see also pid_nr() etc in include/linux/pid.h
1906 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1907 struct pid_namespace *ns);
1909 static inline pid_t task_pid_nr(struct task_struct *tsk)
1914 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1915 struct pid_namespace *ns)
1917 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1920 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1922 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1926 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1931 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1933 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1935 return pid_vnr(task_tgid(tsk));
1939 static inline int pid_alive(const struct task_struct *p);
1940 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1946 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1952 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1954 return task_ppid_nr_ns(tsk, &init_pid_ns);
1957 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1958 struct pid_namespace *ns)
1960 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1963 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1965 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1969 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1970 struct pid_namespace *ns)
1972 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1975 static inline pid_t task_session_vnr(struct task_struct *tsk)
1977 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1980 /* obsolete, do not use */
1981 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1983 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1987 * pid_alive - check that a task structure is not stale
1988 * @p: Task structure to be checked.
1990 * Test if a process is not yet dead (at most zombie state)
1991 * If pid_alive fails, then pointers within the task structure
1992 * can be stale and must not be dereferenced.
1994 * Return: 1 if the process is alive. 0 otherwise.
1996 static inline int pid_alive(const struct task_struct *p)
1998 return p->pids[PIDTYPE_PID].pid != NULL;
2002 * is_global_init - check if a task structure is init
2003 * @tsk: Task structure to be checked.
2005 * Check if a task structure is the first user space task the kernel created.
2007 * Return: 1 if the task structure is init. 0 otherwise.
2009 static inline int is_global_init(struct task_struct *tsk)
2011 return tsk->pid == 1;
2014 extern struct pid *cad_pid;
2016 extern void free_task(struct task_struct *tsk);
2017 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2019 extern void __put_task_struct(struct task_struct *t);
2021 static inline void put_task_struct(struct task_struct *t)
2023 if (atomic_dec_and_test(&t->usage))
2024 __put_task_struct(t);
2027 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2028 extern void task_cputime(struct task_struct *t,
2029 cputime_t *utime, cputime_t *stime);
2030 extern void task_cputime_scaled(struct task_struct *t,
2031 cputime_t *utimescaled, cputime_t *stimescaled);
2032 extern cputime_t task_gtime(struct task_struct *t);
2034 static inline void task_cputime(struct task_struct *t,
2035 cputime_t *utime, cputime_t *stime)
2043 static inline void task_cputime_scaled(struct task_struct *t,
2044 cputime_t *utimescaled,
2045 cputime_t *stimescaled)
2048 *utimescaled = t->utimescaled;
2050 *stimescaled = t->stimescaled;
2053 static inline cputime_t task_gtime(struct task_struct *t)
2058 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2059 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2064 #define PF_EXITING 0x00000004 /* getting shut down */
2065 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2066 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2067 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2068 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2069 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2070 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2071 #define PF_DUMPCORE 0x00000200 /* dumped core */
2072 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2073 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2074 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2075 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2076 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2077 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2078 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2079 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2080 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2081 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2082 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2083 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2084 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2085 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2086 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2087 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2088 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2089 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2090 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2093 * Only the _current_ task can read/write to tsk->flags, but other
2094 * tasks can access tsk->flags in readonly mode for example
2095 * with tsk_used_math (like during threaded core dumping).
2096 * There is however an exception to this rule during ptrace
2097 * or during fork: the ptracer task is allowed to write to the
2098 * child->flags of its traced child (same goes for fork, the parent
2099 * can write to the child->flags), because we're guaranteed the
2100 * child is not running and in turn not changing child->flags
2101 * at the same time the parent does it.
2103 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2104 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2105 #define clear_used_math() clear_stopped_child_used_math(current)
2106 #define set_used_math() set_stopped_child_used_math(current)
2107 #define conditional_stopped_child_used_math(condition, child) \
2108 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2109 #define conditional_used_math(condition) \
2110 conditional_stopped_child_used_math(condition, current)
2111 #define copy_to_stopped_child_used_math(child) \
2112 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2113 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2114 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2115 #define used_math() tsk_used_math(current)
2117 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2118 * __GFP_FS is also cleared as it implies __GFP_IO.
2120 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2122 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2123 flags &= ~(__GFP_IO | __GFP_FS);
2127 static inline unsigned int memalloc_noio_save(void)
2129 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2130 current->flags |= PF_MEMALLOC_NOIO;
2134 static inline void memalloc_noio_restore(unsigned int flags)
2136 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2139 /* Per-process atomic flags. */
2140 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2141 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2142 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2145 #define TASK_PFA_TEST(name, func) \
2146 static inline bool task_##func(struct task_struct *p) \
2147 { return test_bit(PFA_##name, &p->atomic_flags); }
2148 #define TASK_PFA_SET(name, func) \
2149 static inline void task_set_##func(struct task_struct *p) \
2150 { set_bit(PFA_##name, &p->atomic_flags); }
2151 #define TASK_PFA_CLEAR(name, func) \
2152 static inline void task_clear_##func(struct task_struct *p) \
2153 { clear_bit(PFA_##name, &p->atomic_flags); }
2155 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2156 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2158 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2159 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2160 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2162 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2163 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2164 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2167 * task->jobctl flags
2169 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2171 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2172 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2173 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2174 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2175 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2176 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2177 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2179 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2180 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2181 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2182 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2183 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2184 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2185 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2187 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2188 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2190 extern bool task_set_jobctl_pending(struct task_struct *task,
2191 unsigned long mask);
2192 extern void task_clear_jobctl_trapping(struct task_struct *task);
2193 extern void task_clear_jobctl_pending(struct task_struct *task,
2194 unsigned long mask);
2196 static inline void rcu_copy_process(struct task_struct *p)
2198 #ifdef CONFIG_PREEMPT_RCU
2199 p->rcu_read_lock_nesting = 0;
2200 p->rcu_read_unlock_special.s = 0;
2201 p->rcu_blocked_node = NULL;
2202 INIT_LIST_HEAD(&p->rcu_node_entry);
2203 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2204 #ifdef CONFIG_TASKS_RCU
2205 p->rcu_tasks_holdout = false;
2206 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2207 p->rcu_tasks_idle_cpu = -1;
2208 #endif /* #ifdef CONFIG_TASKS_RCU */
2211 static inline void tsk_restore_flags(struct task_struct *task,
2212 unsigned long orig_flags, unsigned long flags)
2214 task->flags &= ~flags;
2215 task->flags |= orig_flags & flags;
2218 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2219 const struct cpumask *trial);
2220 extern int task_can_attach(struct task_struct *p,
2221 const struct cpumask *cs_cpus_allowed);
2223 extern void do_set_cpus_allowed(struct task_struct *p,
2224 const struct cpumask *new_mask);
2226 extern int set_cpus_allowed_ptr(struct task_struct *p,
2227 const struct cpumask *new_mask);
2229 static inline void do_set_cpus_allowed(struct task_struct *p,
2230 const struct cpumask *new_mask)
2233 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2234 const struct cpumask *new_mask)
2236 if (!cpumask_test_cpu(0, new_mask))
2242 #ifdef CONFIG_NO_HZ_COMMON
2243 void calc_load_enter_idle(void);
2244 void calc_load_exit_idle(void);
2246 static inline void calc_load_enter_idle(void) { }
2247 static inline void calc_load_exit_idle(void) { }
2248 #endif /* CONFIG_NO_HZ_COMMON */
2251 * Do not use outside of architecture code which knows its limitations.
2253 * sched_clock() has no promise of monotonicity or bounded drift between
2254 * CPUs, use (which you should not) requires disabling IRQs.
2256 * Please use one of the three interfaces below.
2258 extern unsigned long long notrace sched_clock(void);
2260 * See the comment in kernel/sched/clock.c
2262 extern u64 cpu_clock(int cpu);
2263 extern u64 local_clock(void);
2264 extern u64 running_clock(void);
2265 extern u64 sched_clock_cpu(int cpu);
2268 extern void sched_clock_init(void);
2270 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2271 static inline void sched_clock_tick(void)
2275 static inline void sched_clock_idle_sleep_event(void)
2279 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2284 * Architectures can set this to 1 if they have specified
2285 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2286 * but then during bootup it turns out that sched_clock()
2287 * is reliable after all:
2289 extern int sched_clock_stable(void);
2290 extern void set_sched_clock_stable(void);
2291 extern void clear_sched_clock_stable(void);
2293 extern void sched_clock_tick(void);
2294 extern void sched_clock_idle_sleep_event(void);
2295 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2298 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2300 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2301 * The reason for this explicit opt-in is not to have perf penalty with
2302 * slow sched_clocks.
2304 extern void enable_sched_clock_irqtime(void);
2305 extern void disable_sched_clock_irqtime(void);
2307 static inline void enable_sched_clock_irqtime(void) {}
2308 static inline void disable_sched_clock_irqtime(void) {}
2311 extern unsigned long long
2312 task_sched_runtime(struct task_struct *task);
2314 /* sched_exec is called by processes performing an exec */
2316 extern void sched_exec(void);
2318 #define sched_exec() {}
2321 extern void sched_clock_idle_sleep_event(void);
2322 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2324 #ifdef CONFIG_HOTPLUG_CPU
2325 extern void idle_task_exit(void);
2327 static inline void idle_task_exit(void) {}
2330 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2331 extern void wake_up_nohz_cpu(int cpu);
2333 static inline void wake_up_nohz_cpu(int cpu) { }
2336 #ifdef CONFIG_NO_HZ_FULL
2337 extern bool sched_can_stop_tick(void);
2338 extern u64 scheduler_tick_max_deferment(void);
2340 static inline bool sched_can_stop_tick(void) { return false; }
2343 #ifdef CONFIG_SCHED_AUTOGROUP
2344 extern void sched_autogroup_create_attach(struct task_struct *p);
2345 extern void sched_autogroup_detach(struct task_struct *p);
2346 extern void sched_autogroup_fork(struct signal_struct *sig);
2347 extern void sched_autogroup_exit(struct signal_struct *sig);
2348 #ifdef CONFIG_PROC_FS
2349 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2350 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2353 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2354 static inline void sched_autogroup_detach(struct task_struct *p) { }
2355 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2356 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2359 extern int yield_to(struct task_struct *p, bool preempt);
2360 extern void set_user_nice(struct task_struct *p, long nice);
2361 extern int task_prio(const struct task_struct *p);
2363 * task_nice - return the nice value of a given task.
2364 * @p: the task in question.
2366 * Return: The nice value [ -20 ... 0 ... 19 ].
2368 static inline int task_nice(const struct task_struct *p)
2370 return PRIO_TO_NICE((p)->static_prio);
2372 extern int can_nice(const struct task_struct *p, const int nice);
2373 extern int task_curr(const struct task_struct *p);
2374 extern int idle_cpu(int cpu);
2375 extern int sched_setscheduler(struct task_struct *, int,
2376 const struct sched_param *);
2377 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2378 const struct sched_param *);
2379 extern int sched_setattr(struct task_struct *,
2380 const struct sched_attr *);
2381 extern struct task_struct *idle_task(int cpu);
2383 * is_idle_task - is the specified task an idle task?
2384 * @p: the task in question.
2386 * Return: 1 if @p is an idle task. 0 otherwise.
2388 static inline bool is_idle_task(const struct task_struct *p)
2392 extern struct task_struct *curr_task(int cpu);
2393 extern void set_curr_task(int cpu, struct task_struct *p);
2397 union thread_union {
2398 struct thread_info thread_info;
2399 unsigned long stack[THREAD_SIZE/sizeof(long)];
2402 #ifndef __HAVE_ARCH_KSTACK_END
2403 static inline int kstack_end(void *addr)
2405 /* Reliable end of stack detection:
2406 * Some APM bios versions misalign the stack
2408 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2412 extern union thread_union init_thread_union;
2413 extern struct task_struct init_task;
2415 extern struct mm_struct init_mm;
2417 extern struct pid_namespace init_pid_ns;
2420 * find a task by one of its numerical ids
2422 * find_task_by_pid_ns():
2423 * finds a task by its pid in the specified namespace
2424 * find_task_by_vpid():
2425 * finds a task by its virtual pid
2427 * see also find_vpid() etc in include/linux/pid.h
2430 extern struct task_struct *find_task_by_vpid(pid_t nr);
2431 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2432 struct pid_namespace *ns);
2434 /* per-UID process charging. */
2435 extern struct user_struct * alloc_uid(kuid_t);
2436 static inline struct user_struct *get_uid(struct user_struct *u)
2438 atomic_inc(&u->__count);
2441 extern void free_uid(struct user_struct *);
2443 #include <asm/current.h>
2445 extern void xtime_update(unsigned long ticks);
2447 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2448 extern int wake_up_process(struct task_struct *tsk);
2449 extern void wake_up_new_task(struct task_struct *tsk);
2451 extern void kick_process(struct task_struct *tsk);
2453 static inline void kick_process(struct task_struct *tsk) { }
2455 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2456 extern void sched_dead(struct task_struct *p);
2458 extern void proc_caches_init(void);
2459 extern void flush_signals(struct task_struct *);
2460 extern void ignore_signals(struct task_struct *);
2461 extern void flush_signal_handlers(struct task_struct *, int force_default);
2462 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2464 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2466 unsigned long flags;
2469 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2470 ret = dequeue_signal(tsk, mask, info);
2471 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2476 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2478 extern void unblock_all_signals(void);
2479 extern void release_task(struct task_struct * p);
2480 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2481 extern int force_sigsegv(int, struct task_struct *);
2482 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2483 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2484 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2485 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2486 const struct cred *, u32);
2487 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2488 extern int kill_pid(struct pid *pid, int sig, int priv);
2489 extern int kill_proc_info(int, struct siginfo *, pid_t);
2490 extern __must_check bool do_notify_parent(struct task_struct *, int);
2491 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2492 extern void force_sig(int, struct task_struct *);
2493 extern int send_sig(int, struct task_struct *, int);
2494 extern int zap_other_threads(struct task_struct *p);
2495 extern struct sigqueue *sigqueue_alloc(void);
2496 extern void sigqueue_free(struct sigqueue *);
2497 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2498 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2500 static inline void restore_saved_sigmask(void)
2502 if (test_and_clear_restore_sigmask())
2503 __set_current_blocked(¤t->saved_sigmask);
2506 static inline sigset_t *sigmask_to_save(void)
2508 sigset_t *res = ¤t->blocked;
2509 if (unlikely(test_restore_sigmask()))
2510 res = ¤t->saved_sigmask;
2514 static inline int kill_cad_pid(int sig, int priv)
2516 return kill_pid(cad_pid, sig, priv);
2519 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2520 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2521 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2522 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2525 * True if we are on the alternate signal stack.
2527 static inline int on_sig_stack(unsigned long sp)
2529 #ifdef CONFIG_STACK_GROWSUP
2530 return sp >= current->sas_ss_sp &&
2531 sp - current->sas_ss_sp < current->sas_ss_size;
2533 return sp > current->sas_ss_sp &&
2534 sp - current->sas_ss_sp <= current->sas_ss_size;
2538 static inline int sas_ss_flags(unsigned long sp)
2540 if (!current->sas_ss_size)
2543 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2546 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2548 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2549 #ifdef CONFIG_STACK_GROWSUP
2550 return current->sas_ss_sp;
2552 return current->sas_ss_sp + current->sas_ss_size;
2558 * Routines for handling mm_structs
2560 extern struct mm_struct * mm_alloc(void);
2562 /* mmdrop drops the mm and the page tables */
2563 extern void __mmdrop(struct mm_struct *);
2564 static inline void mmdrop(struct mm_struct * mm)
2566 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2570 /* mmput gets rid of the mappings and all user-space */
2571 extern void mmput(struct mm_struct *);
2572 /* Grab a reference to a task's mm, if it is not already going away */
2573 extern struct mm_struct *get_task_mm(struct task_struct *task);
2575 * Grab a reference to a task's mm, if it is not already going away
2576 * and ptrace_may_access with the mode parameter passed to it
2579 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2580 /* Remove the current tasks stale references to the old mm_struct */
2581 extern void mm_release(struct task_struct *, struct mm_struct *);
2583 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2584 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2585 struct task_struct *, unsigned long);
2587 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2588 struct task_struct *);
2590 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2591 * via pt_regs, so ignore the tls argument passed via C. */
2592 static inline int copy_thread_tls(
2593 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2594 struct task_struct *p, unsigned long tls)
2596 return copy_thread(clone_flags, sp, arg, p);
2599 extern void flush_thread(void);
2600 extern void exit_thread(void);
2602 extern void exit_files(struct task_struct *);
2603 extern void __cleanup_sighand(struct sighand_struct *);
2605 extern void exit_itimers(struct signal_struct *);
2606 extern void flush_itimer_signals(void);
2608 extern void do_group_exit(int);
2610 extern int do_execve(struct filename *,
2611 const char __user * const __user *,
2612 const char __user * const __user *);
2613 extern int do_execveat(int, struct filename *,
2614 const char __user * const __user *,
2615 const char __user * const __user *,
2617 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2618 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2619 struct task_struct *fork_idle(int);
2620 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2622 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2623 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2625 __set_task_comm(tsk, from, false);
2627 extern char *get_task_comm(char *to, struct task_struct *tsk);
2630 void scheduler_ipi(void);
2631 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2633 static inline void scheduler_ipi(void) { }
2634 static inline unsigned long wait_task_inactive(struct task_struct *p,
2641 #define tasklist_empty() \
2642 list_empty(&init_task.tasks)
2644 #define next_task(p) \
2645 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2647 #define for_each_process(p) \
2648 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2650 extern bool current_is_single_threaded(void);
2653 * Careful: do_each_thread/while_each_thread is a double loop so
2654 * 'break' will not work as expected - use goto instead.
2656 #define do_each_thread(g, t) \
2657 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2659 #define while_each_thread(g, t) \
2660 while ((t = next_thread(t)) != g)
2662 #define __for_each_thread(signal, t) \
2663 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2665 #define for_each_thread(p, t) \
2666 __for_each_thread((p)->signal, t)
2668 /* Careful: this is a double loop, 'break' won't work as expected. */
2669 #define for_each_process_thread(p, t) \
2670 for_each_process(p) for_each_thread(p, t)
2672 static inline int get_nr_threads(struct task_struct *tsk)
2674 return tsk->signal->nr_threads;
2677 static inline bool thread_group_leader(struct task_struct *p)
2679 return p->exit_signal >= 0;
2682 /* Do to the insanities of de_thread it is possible for a process
2683 * to have the pid of the thread group leader without actually being
2684 * the thread group leader. For iteration through the pids in proc
2685 * all we care about is that we have a task with the appropriate
2686 * pid, we don't actually care if we have the right task.
2688 static inline bool has_group_leader_pid(struct task_struct *p)
2690 return task_pid(p) == p->signal->leader_pid;
2694 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2696 return p1->signal == p2->signal;
2699 static inline struct task_struct *next_thread(const struct task_struct *p)
2701 return list_entry_rcu(p->thread_group.next,
2702 struct task_struct, thread_group);
2705 static inline int thread_group_empty(struct task_struct *p)
2707 return list_empty(&p->thread_group);
2710 #define delay_group_leader(p) \
2711 (thread_group_leader(p) && !thread_group_empty(p))
2714 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2715 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2716 * pins the final release of task.io_context. Also protects ->cpuset and
2717 * ->cgroup.subsys[]. And ->vfork_done.
2719 * Nests both inside and outside of read_lock(&tasklist_lock).
2720 * It must not be nested with write_lock_irq(&tasklist_lock),
2721 * neither inside nor outside.
2723 static inline void task_lock(struct task_struct *p)
2725 spin_lock(&p->alloc_lock);
2728 static inline void task_unlock(struct task_struct *p)
2730 spin_unlock(&p->alloc_lock);
2733 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2734 unsigned long *flags);
2736 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2737 unsigned long *flags)
2739 struct sighand_struct *ret;
2741 ret = __lock_task_sighand(tsk, flags);
2742 (void)__cond_lock(&tsk->sighand->siglock, ret);
2746 static inline void unlock_task_sighand(struct task_struct *tsk,
2747 unsigned long *flags)
2749 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2753 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2754 * @tsk: task causing the changes
2756 * All operations which modify a threadgroup - a new thread joining the
2757 * group, death of a member thread (the assertion of PF_EXITING) and
2758 * exec(2) dethreading the process and replacing the leader - are wrapped
2759 * by threadgroup_change_{begin|end}(). This is to provide a place which
2760 * subsystems needing threadgroup stability can hook into for
2763 static inline void threadgroup_change_begin(struct task_struct *tsk)
2766 cgroup_threadgroup_change_begin(tsk);
2770 * threadgroup_change_end - mark the end of changes to a threadgroup
2771 * @tsk: task causing the changes
2773 * See threadgroup_change_begin().
2775 static inline void threadgroup_change_end(struct task_struct *tsk)
2777 cgroup_threadgroup_change_end(tsk);
2780 #ifndef __HAVE_THREAD_FUNCTIONS
2782 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2783 #define task_stack_page(task) ((task)->stack)
2785 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2787 *task_thread_info(p) = *task_thread_info(org);
2788 task_thread_info(p)->task = p;
2792 * Return the address of the last usable long on the stack.
2794 * When the stack grows down, this is just above the thread
2795 * info struct. Going any lower will corrupt the threadinfo.
2797 * When the stack grows up, this is the highest address.
2798 * Beyond that position, we corrupt data on the next page.
2800 static inline unsigned long *end_of_stack(struct task_struct *p)
2802 #ifdef CONFIG_STACK_GROWSUP
2803 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2805 return (unsigned long *)(task_thread_info(p) + 1);
2810 #define task_stack_end_corrupted(task) \
2811 (*(end_of_stack(task)) != STACK_END_MAGIC)
2813 static inline int object_is_on_stack(void *obj)
2815 void *stack = task_stack_page(current);
2817 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2820 extern void thread_info_cache_init(void);
2822 #ifdef CONFIG_DEBUG_STACK_USAGE
2823 static inline unsigned long stack_not_used(struct task_struct *p)
2825 unsigned long *n = end_of_stack(p);
2827 do { /* Skip over canary */
2831 return (unsigned long)n - (unsigned long)end_of_stack(p);
2834 extern void set_task_stack_end_magic(struct task_struct *tsk);
2836 /* set thread flags in other task's structures
2837 * - see asm/thread_info.h for TIF_xxxx flags available
2839 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2841 set_ti_thread_flag(task_thread_info(tsk), flag);
2844 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2846 clear_ti_thread_flag(task_thread_info(tsk), flag);
2849 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2851 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2854 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2856 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2859 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2861 return test_ti_thread_flag(task_thread_info(tsk), flag);
2864 static inline void set_tsk_need_resched(struct task_struct *tsk)
2866 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2869 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2871 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2874 static inline int test_tsk_need_resched(struct task_struct *tsk)
2876 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2879 static inline int restart_syscall(void)
2881 set_tsk_thread_flag(current, TIF_SIGPENDING);
2882 return -ERESTARTNOINTR;
2885 static inline int signal_pending(struct task_struct *p)
2887 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2890 static inline int __fatal_signal_pending(struct task_struct *p)
2892 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2895 static inline int fatal_signal_pending(struct task_struct *p)
2897 return signal_pending(p) && __fatal_signal_pending(p);
2900 static inline int signal_pending_state(long state, struct task_struct *p)
2902 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2904 if (!signal_pending(p))
2907 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2911 * cond_resched() and cond_resched_lock(): latency reduction via
2912 * explicit rescheduling in places that are safe. The return
2913 * value indicates whether a reschedule was done in fact.
2914 * cond_resched_lock() will drop the spinlock before scheduling,
2915 * cond_resched_softirq() will enable bhs before scheduling.
2917 extern int _cond_resched(void);
2919 #define cond_resched() ({ \
2920 ___might_sleep(__FILE__, __LINE__, 0); \
2924 extern int __cond_resched_lock(spinlock_t *lock);
2926 #define cond_resched_lock(lock) ({ \
2927 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2928 __cond_resched_lock(lock); \
2931 extern int __cond_resched_softirq(void);
2933 #define cond_resched_softirq() ({ \
2934 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2935 __cond_resched_softirq(); \
2938 static inline void cond_resched_rcu(void)
2940 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2948 * Does a critical section need to be broken due to another
2949 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2950 * but a general need for low latency)
2952 static inline int spin_needbreak(spinlock_t *lock)
2954 #ifdef CONFIG_PREEMPT
2955 return spin_is_contended(lock);
2962 * Idle thread specific functions to determine the need_resched
2965 #ifdef TIF_POLLING_NRFLAG
2966 static inline int tsk_is_polling(struct task_struct *p)
2968 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2971 static inline void __current_set_polling(void)
2973 set_thread_flag(TIF_POLLING_NRFLAG);
2976 static inline bool __must_check current_set_polling_and_test(void)
2978 __current_set_polling();
2981 * Polling state must be visible before we test NEED_RESCHED,
2982 * paired by resched_curr()
2984 smp_mb__after_atomic();
2986 return unlikely(tif_need_resched());
2989 static inline void __current_clr_polling(void)
2991 clear_thread_flag(TIF_POLLING_NRFLAG);
2994 static inline bool __must_check current_clr_polling_and_test(void)
2996 __current_clr_polling();
2999 * Polling state must be visible before we test NEED_RESCHED,
3000 * paired by resched_curr()
3002 smp_mb__after_atomic();
3004 return unlikely(tif_need_resched());
3008 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3009 static inline void __current_set_polling(void) { }
3010 static inline void __current_clr_polling(void) { }
3012 static inline bool __must_check current_set_polling_and_test(void)
3014 return unlikely(tif_need_resched());
3016 static inline bool __must_check current_clr_polling_and_test(void)
3018 return unlikely(tif_need_resched());
3022 static inline void current_clr_polling(void)
3024 __current_clr_polling();
3027 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3028 * Once the bit is cleared, we'll get IPIs with every new
3029 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3032 smp_mb(); /* paired with resched_curr() */
3034 preempt_fold_need_resched();
3037 static __always_inline bool need_resched(void)
3039 return unlikely(tif_need_resched());
3043 * Thread group CPU time accounting.
3045 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3046 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3049 * Reevaluate whether the task has signals pending delivery.
3050 * Wake the task if so.
3051 * This is required every time the blocked sigset_t changes.
3052 * callers must hold sighand->siglock.
3054 extern void recalc_sigpending_and_wake(struct task_struct *t);
3055 extern void recalc_sigpending(void);
3057 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3059 static inline void signal_wake_up(struct task_struct *t, bool resume)
3061 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3063 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3065 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3069 * Wrappers for p->thread_info->cpu access. No-op on UP.
3073 static inline unsigned int task_cpu(const struct task_struct *p)
3075 return task_thread_info(p)->cpu;
3078 static inline int task_node(const struct task_struct *p)
3080 return cpu_to_node(task_cpu(p));
3083 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3087 static inline unsigned int task_cpu(const struct task_struct *p)
3092 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3096 #endif /* CONFIG_SMP */
3098 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3099 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3101 #ifdef CONFIG_CGROUP_SCHED
3102 extern struct task_group root_task_group;
3103 #endif /* CONFIG_CGROUP_SCHED */
3105 extern int task_can_switch_user(struct user_struct *up,
3106 struct task_struct *tsk);
3108 #ifdef CONFIG_TASK_XACCT
3109 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3111 tsk->ioac.rchar += amt;
3114 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3116 tsk->ioac.wchar += amt;
3119 static inline void inc_syscr(struct task_struct *tsk)
3124 static inline void inc_syscw(struct task_struct *tsk)
3129 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3133 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3137 static inline void inc_syscr(struct task_struct *tsk)
3141 static inline void inc_syscw(struct task_struct *tsk)
3146 #ifndef TASK_SIZE_OF
3147 #define TASK_SIZE_OF(tsk) TASK_SIZE
3151 extern void mm_update_next_owner(struct mm_struct *mm);
3153 static inline void mm_update_next_owner(struct mm_struct *mm)
3156 #endif /* CONFIG_MEMCG */
3158 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3161 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3164 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3167 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3170 static inline unsigned long rlimit(unsigned int limit)
3172 return task_rlimit(current, limit);
3175 static inline unsigned long rlimit_max(unsigned int limit)
3177 return task_rlimit_max(current, limit);