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 #ifdef CONFIG_PREEMPT_COUNT
603 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
605 #define PREEMPT_DISABLED PREEMPT_ENABLED
609 * Disable preemption until the scheduler is running.
610 * Reset by start_kernel()->sched_init()->init_idle().
612 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
613 * before the scheduler is active -- see should_resched().
615 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
618 * struct thread_group_cputimer - thread group interval timer counts
619 * @cputime_atomic: atomic thread group interval timers.
620 * @running: non-zero when there are timers running and
621 * @cputime receives updates.
623 * This structure contains the version of task_cputime, above, that is
624 * used for thread group CPU timer calculations.
626 struct thread_group_cputimer {
627 struct task_cputime_atomic cputime_atomic;
631 #include <linux/rwsem.h>
635 * NOTE! "signal_struct" does not have its own
636 * locking, because a shared signal_struct always
637 * implies a shared sighand_struct, so locking
638 * sighand_struct is always a proper superset of
639 * the locking of signal_struct.
641 struct signal_struct {
645 struct list_head thread_head;
647 wait_queue_head_t wait_chldexit; /* for wait4() */
649 /* current thread group signal load-balancing target: */
650 struct task_struct *curr_target;
652 /* shared signal handling: */
653 struct sigpending shared_pending;
655 /* thread group exit support */
658 * - notify group_exit_task when ->count is equal to notify_count
659 * - everyone except group_exit_task is stopped during signal delivery
660 * of fatal signals, group_exit_task processes the signal.
663 struct task_struct *group_exit_task;
665 /* thread group stop support, overloads group_exit_code too */
666 int group_stop_count;
667 unsigned int flags; /* see SIGNAL_* flags below */
670 * PR_SET_CHILD_SUBREAPER marks a process, like a service
671 * manager, to re-parent orphan (double-forking) child processes
672 * to this process instead of 'init'. The service manager is
673 * able to receive SIGCHLD signals and is able to investigate
674 * the process until it calls wait(). All children of this
675 * process will inherit a flag if they should look for a
676 * child_subreaper process at exit.
678 unsigned int is_child_subreaper:1;
679 unsigned int has_child_subreaper:1;
681 /* POSIX.1b Interval Timers */
683 struct list_head posix_timers;
685 /* ITIMER_REAL timer for the process */
686 struct hrtimer real_timer;
687 struct pid *leader_pid;
688 ktime_t it_real_incr;
691 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
692 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
693 * values are defined to 0 and 1 respectively
695 struct cpu_itimer it[2];
698 * Thread group totals for process CPU timers.
699 * See thread_group_cputimer(), et al, for details.
701 struct thread_group_cputimer cputimer;
703 /* Earliest-expiration cache. */
704 struct task_cputime cputime_expires;
706 struct list_head cpu_timers[3];
708 struct pid *tty_old_pgrp;
710 /* boolean value for session group leader */
713 struct tty_struct *tty; /* NULL if no tty */
715 #ifdef CONFIG_SCHED_AUTOGROUP
716 struct autogroup *autogroup;
719 * Cumulative resource counters for dead threads in the group,
720 * and for reaped dead child processes forked by this group.
721 * Live threads maintain their own counters and add to these
722 * in __exit_signal, except for the group leader.
724 seqlock_t stats_lock;
725 cputime_t utime, stime, cutime, cstime;
728 struct prev_cputime prev_cputime;
729 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
730 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
731 unsigned long inblock, oublock, cinblock, coublock;
732 unsigned long maxrss, cmaxrss;
733 struct task_io_accounting ioac;
736 * Cumulative ns of schedule CPU time fo dead threads in the
737 * group, not including a zombie group leader, (This only differs
738 * from jiffies_to_ns(utime + stime) if sched_clock uses something
739 * other than jiffies.)
741 unsigned long long sum_sched_runtime;
744 * We don't bother to synchronize most readers of this at all,
745 * because there is no reader checking a limit that actually needs
746 * to get both rlim_cur and rlim_max atomically, and either one
747 * alone is a single word that can safely be read normally.
748 * getrlimit/setrlimit use task_lock(current->group_leader) to
749 * protect this instead of the siglock, because they really
750 * have no need to disable irqs.
752 struct rlimit rlim[RLIM_NLIMITS];
754 #ifdef CONFIG_BSD_PROCESS_ACCT
755 struct pacct_struct pacct; /* per-process accounting information */
757 #ifdef CONFIG_TASKSTATS
758 struct taskstats *stats;
762 unsigned audit_tty_log_passwd;
763 struct tty_audit_buf *tty_audit_buf;
765 #ifdef CONFIG_CGROUPS
767 * group_rwsem prevents new tasks from entering the threadgroup and
768 * member tasks from exiting,a more specifically, setting of
769 * PF_EXITING. fork and exit paths are protected with this rwsem
770 * using threadgroup_change_begin/end(). Users which require
771 * threadgroup to remain stable should use threadgroup_[un]lock()
772 * which also takes care of exec path. Currently, cgroup is the
775 struct rw_semaphore group_rwsem;
778 oom_flags_t oom_flags;
779 short oom_score_adj; /* OOM kill score adjustment */
780 short oom_score_adj_min; /* OOM kill score adjustment min value.
781 * Only settable by CAP_SYS_RESOURCE. */
783 struct mutex cred_guard_mutex; /* guard against foreign influences on
784 * credential calculations
785 * (notably. ptrace) */
789 * Bits in flags field of signal_struct.
791 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
792 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
793 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
794 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
796 * Pending notifications to parent.
798 #define SIGNAL_CLD_STOPPED 0x00000010
799 #define SIGNAL_CLD_CONTINUED 0x00000020
800 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
802 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
804 /* If true, all threads except ->group_exit_task have pending SIGKILL */
805 static inline int signal_group_exit(const struct signal_struct *sig)
807 return (sig->flags & SIGNAL_GROUP_EXIT) ||
808 (sig->group_exit_task != NULL);
812 * Some day this will be a full-fledged user tracking system..
815 atomic_t __count; /* reference count */
816 atomic_t processes; /* How many processes does this user have? */
817 atomic_t sigpending; /* How many pending signals does this user have? */
818 #ifdef CONFIG_INOTIFY_USER
819 atomic_t inotify_watches; /* How many inotify watches does this user have? */
820 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
822 #ifdef CONFIG_FANOTIFY
823 atomic_t fanotify_listeners;
826 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
828 #ifdef CONFIG_POSIX_MQUEUE
829 /* protected by mq_lock */
830 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
832 unsigned long locked_shm; /* How many pages of mlocked shm ? */
835 struct key *uid_keyring; /* UID specific keyring */
836 struct key *session_keyring; /* UID's default session keyring */
839 /* Hash table maintenance information */
840 struct hlist_node uidhash_node;
843 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
844 atomic_long_t locked_vm;
848 extern int uids_sysfs_init(void);
850 extern struct user_struct *find_user(kuid_t);
852 extern struct user_struct root_user;
853 #define INIT_USER (&root_user)
856 struct backing_dev_info;
857 struct reclaim_state;
859 #ifdef CONFIG_SCHED_INFO
861 /* cumulative counters */
862 unsigned long pcount; /* # of times run on this cpu */
863 unsigned long long run_delay; /* time spent waiting on a runqueue */
866 unsigned long long last_arrival,/* when we last ran on a cpu */
867 last_queued; /* when we were last queued to run */
869 #endif /* CONFIG_SCHED_INFO */
871 #ifdef CONFIG_TASK_DELAY_ACCT
872 struct task_delay_info {
874 unsigned int flags; /* Private per-task flags */
876 /* For each stat XXX, add following, aligned appropriately
878 * struct timespec XXX_start, XXX_end;
882 * Atomicity of updates to XXX_delay, XXX_count protected by
883 * single lock above (split into XXX_lock if contention is an issue).
887 * XXX_count is incremented on every XXX operation, the delay
888 * associated with the operation is added to XXX_delay.
889 * XXX_delay contains the accumulated delay time in nanoseconds.
891 u64 blkio_start; /* Shared by blkio, swapin */
892 u64 blkio_delay; /* wait for sync block io completion */
893 u64 swapin_delay; /* wait for swapin block io completion */
894 u32 blkio_count; /* total count of the number of sync block */
895 /* io operations performed */
896 u32 swapin_count; /* total count of the number of swapin block */
897 /* io operations performed */
900 u64 freepages_delay; /* wait for memory reclaim */
901 u32 freepages_count; /* total count of memory reclaim */
903 #endif /* CONFIG_TASK_DELAY_ACCT */
905 static inline int sched_info_on(void)
907 #ifdef CONFIG_SCHEDSTATS
909 #elif defined(CONFIG_TASK_DELAY_ACCT)
910 extern int delayacct_on;
925 * Increase resolution of cpu_capacity calculations
927 #define SCHED_CAPACITY_SHIFT 10
928 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
931 * Wake-queues are lists of tasks with a pending wakeup, whose
932 * callers have already marked the task as woken internally,
933 * and can thus carry on. A common use case is being able to
934 * do the wakeups once the corresponding user lock as been
937 * We hold reference to each task in the list across the wakeup,
938 * thus guaranteeing that the memory is still valid by the time
939 * the actual wakeups are performed in wake_up_q().
941 * One per task suffices, because there's never a need for a task to be
942 * in two wake queues simultaneously; it is forbidden to abandon a task
943 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
944 * already in a wake queue, the wakeup will happen soon and the second
945 * waker can just skip it.
947 * The WAKE_Q macro declares and initializes the list head.
948 * wake_up_q() does NOT reinitialize the list; it's expected to be
949 * called near the end of a function, where the fact that the queue is
950 * not used again will be easy to see by inspection.
952 * Note that this can cause spurious wakeups. schedule() callers
953 * must ensure the call is done inside a loop, confirming that the
954 * wakeup condition has in fact occurred.
957 struct wake_q_node *next;
961 struct wake_q_node *first;
962 struct wake_q_node **lastp;
965 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
967 #define WAKE_Q(name) \
968 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
970 extern void wake_q_add(struct wake_q_head *head,
971 struct task_struct *task);
972 extern void wake_up_q(struct wake_q_head *head);
975 * sched-domains (multiprocessor balancing) declarations:
978 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
979 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
980 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
981 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
982 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
983 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
984 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
985 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
986 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
987 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
988 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
989 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
990 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
991 #define SD_NUMA 0x4000 /* cross-node balancing */
993 #ifdef CONFIG_SCHED_SMT
994 static inline int cpu_smt_flags(void)
996 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1000 #ifdef CONFIG_SCHED_MC
1001 static inline int cpu_core_flags(void)
1003 return SD_SHARE_PKG_RESOURCES;
1008 static inline int cpu_numa_flags(void)
1014 struct sched_domain_attr {
1015 int relax_domain_level;
1018 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1019 .relax_domain_level = -1, \
1022 extern int sched_domain_level_max;
1026 struct sched_domain {
1027 /* These fields must be setup */
1028 struct sched_domain *parent; /* top domain must be null terminated */
1029 struct sched_domain *child; /* bottom domain must be null terminated */
1030 struct sched_group *groups; /* the balancing groups of the domain */
1031 unsigned long min_interval; /* Minimum balance interval ms */
1032 unsigned long max_interval; /* Maximum balance interval ms */
1033 unsigned int busy_factor; /* less balancing by factor if busy */
1034 unsigned int imbalance_pct; /* No balance until over watermark */
1035 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1036 unsigned int busy_idx;
1037 unsigned int idle_idx;
1038 unsigned int newidle_idx;
1039 unsigned int wake_idx;
1040 unsigned int forkexec_idx;
1041 unsigned int smt_gain;
1043 int nohz_idle; /* NOHZ IDLE status */
1044 int flags; /* See SD_* */
1047 /* Runtime fields. */
1048 unsigned long last_balance; /* init to jiffies. units in jiffies */
1049 unsigned int balance_interval; /* initialise to 1. units in ms. */
1050 unsigned int nr_balance_failed; /* initialise to 0 */
1052 /* idle_balance() stats */
1053 u64 max_newidle_lb_cost;
1054 unsigned long next_decay_max_lb_cost;
1056 #ifdef CONFIG_SCHEDSTATS
1057 /* load_balance() stats */
1058 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1059 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1060 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1061 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1062 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1063 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1064 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1065 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1067 /* Active load balancing */
1068 unsigned int alb_count;
1069 unsigned int alb_failed;
1070 unsigned int alb_pushed;
1072 /* SD_BALANCE_EXEC stats */
1073 unsigned int sbe_count;
1074 unsigned int sbe_balanced;
1075 unsigned int sbe_pushed;
1077 /* SD_BALANCE_FORK stats */
1078 unsigned int sbf_count;
1079 unsigned int sbf_balanced;
1080 unsigned int sbf_pushed;
1082 /* try_to_wake_up() stats */
1083 unsigned int ttwu_wake_remote;
1084 unsigned int ttwu_move_affine;
1085 unsigned int ttwu_move_balance;
1087 #ifdef CONFIG_SCHED_DEBUG
1091 void *private; /* used during construction */
1092 struct rcu_head rcu; /* used during destruction */
1095 unsigned int span_weight;
1097 * Span of all CPUs in this domain.
1099 * NOTE: this field is variable length. (Allocated dynamically
1100 * by attaching extra space to the end of the structure,
1101 * depending on how many CPUs the kernel has booted up with)
1103 unsigned long span[0];
1106 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1108 return to_cpumask(sd->span);
1111 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1112 struct sched_domain_attr *dattr_new);
1114 /* Allocate an array of sched domains, for partition_sched_domains(). */
1115 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1116 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1118 bool cpus_share_cache(int this_cpu, int that_cpu);
1120 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1121 typedef int (*sched_domain_flags_f)(void);
1123 #define SDTL_OVERLAP 0x01
1126 struct sched_domain **__percpu sd;
1127 struct sched_group **__percpu sg;
1128 struct sched_group_capacity **__percpu sgc;
1131 struct sched_domain_topology_level {
1132 sched_domain_mask_f mask;
1133 sched_domain_flags_f sd_flags;
1136 struct sd_data data;
1137 #ifdef CONFIG_SCHED_DEBUG
1142 extern struct sched_domain_topology_level *sched_domain_topology;
1144 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1145 extern void wake_up_if_idle(int cpu);
1147 #ifdef CONFIG_SCHED_DEBUG
1148 # define SD_INIT_NAME(type) .name = #type
1150 # define SD_INIT_NAME(type)
1153 #else /* CONFIG_SMP */
1155 struct sched_domain_attr;
1158 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1159 struct sched_domain_attr *dattr_new)
1163 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1168 #endif /* !CONFIG_SMP */
1171 struct io_context; /* See blkdev.h */
1174 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1175 extern void prefetch_stack(struct task_struct *t);
1177 static inline void prefetch_stack(struct task_struct *t) { }
1180 struct audit_context; /* See audit.c */
1182 struct pipe_inode_info;
1183 struct uts_namespace;
1185 struct load_weight {
1186 unsigned long weight;
1191 * The load_avg/util_avg accumulates an infinite geometric series.
1192 * 1) load_avg factors the amount of time that a sched_entity is
1193 * runnable on a rq into its weight. For cfs_rq, it is the aggregated
1194 * such weights of all runnable and blocked sched_entities.
1195 * 2) util_avg factors frequency scaling into the amount of time
1196 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1197 * For cfs_rq, it is the aggregated such times of all runnable and
1198 * blocked sched_entities.
1199 * The 64 bit load_sum can:
1200 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1201 * the highest weight (=88761) always runnable, we should not overflow
1202 * 2) for entity, support any load.weight always runnable
1205 u64 last_update_time, load_sum;
1206 u32 util_sum, period_contrib;
1207 unsigned long load_avg, util_avg;
1210 #ifdef CONFIG_SCHEDSTATS
1211 struct sched_statistics {
1221 s64 sum_sleep_runtime;
1228 u64 nr_migrations_cold;
1229 u64 nr_failed_migrations_affine;
1230 u64 nr_failed_migrations_running;
1231 u64 nr_failed_migrations_hot;
1232 u64 nr_forced_migrations;
1235 u64 nr_wakeups_sync;
1236 u64 nr_wakeups_migrate;
1237 u64 nr_wakeups_local;
1238 u64 nr_wakeups_remote;
1239 u64 nr_wakeups_affine;
1240 u64 nr_wakeups_affine_attempts;
1241 u64 nr_wakeups_passive;
1242 u64 nr_wakeups_idle;
1246 struct sched_entity {
1247 struct load_weight load; /* for load-balancing */
1248 struct rb_node run_node;
1249 struct list_head group_node;
1253 u64 sum_exec_runtime;
1255 u64 prev_sum_exec_runtime;
1259 #ifdef CONFIG_SCHEDSTATS
1260 struct sched_statistics statistics;
1263 #ifdef CONFIG_FAIR_GROUP_SCHED
1265 struct sched_entity *parent;
1266 /* rq on which this entity is (to be) queued: */
1267 struct cfs_rq *cfs_rq;
1268 /* rq "owned" by this entity/group: */
1269 struct cfs_rq *my_q;
1273 /* Per entity load average tracking */
1274 struct sched_avg avg;
1278 struct sched_rt_entity {
1279 struct list_head run_list;
1280 unsigned long timeout;
1281 unsigned long watchdog_stamp;
1282 unsigned int time_slice;
1284 struct sched_rt_entity *back;
1285 #ifdef CONFIG_RT_GROUP_SCHED
1286 struct sched_rt_entity *parent;
1287 /* rq on which this entity is (to be) queued: */
1288 struct rt_rq *rt_rq;
1289 /* rq "owned" by this entity/group: */
1294 struct sched_dl_entity {
1295 struct rb_node rb_node;
1298 * Original scheduling parameters. Copied here from sched_attr
1299 * during sched_setattr(), they will remain the same until
1300 * the next sched_setattr().
1302 u64 dl_runtime; /* maximum runtime for each instance */
1303 u64 dl_deadline; /* relative deadline of each instance */
1304 u64 dl_period; /* separation of two instances (period) */
1305 u64 dl_bw; /* dl_runtime / dl_deadline */
1308 * Actual scheduling parameters. Initialized with the values above,
1309 * they are continously updated during task execution. Note that
1310 * the remaining runtime could be < 0 in case we are in overrun.
1312 s64 runtime; /* remaining runtime for this instance */
1313 u64 deadline; /* absolute deadline for this instance */
1314 unsigned int flags; /* specifying the scheduler behaviour */
1319 * @dl_throttled tells if we exhausted the runtime. If so, the
1320 * task has to wait for a replenishment to be performed at the
1321 * next firing of dl_timer.
1323 * @dl_new tells if a new instance arrived. If so we must
1324 * start executing it with full runtime and reset its absolute
1327 * @dl_boosted tells if we are boosted due to DI. If so we are
1328 * outside bandwidth enforcement mechanism (but only until we
1329 * exit the critical section);
1331 * @dl_yielded tells if task gave up the cpu before consuming
1332 * all its available runtime during the last job.
1334 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1337 * Bandwidth enforcement timer. Each -deadline task has its
1338 * own bandwidth to be enforced, thus we need one timer per task.
1340 struct hrtimer dl_timer;
1352 enum perf_event_task_context {
1353 perf_invalid_context = -1,
1354 perf_hw_context = 0,
1356 perf_nr_task_contexts,
1359 /* Track pages that require TLB flushes */
1360 struct tlbflush_unmap_batch {
1362 * Each bit set is a CPU that potentially has a TLB entry for one of
1363 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1365 struct cpumask cpumask;
1367 /* True if any bit in cpumask is set */
1368 bool flush_required;
1371 * If true then the PTE was dirty when unmapped. The entry must be
1372 * flushed before IO is initiated or a stale TLB entry potentially
1373 * allows an update without redirtying the page.
1378 struct task_struct {
1379 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1382 unsigned int flags; /* per process flags, defined below */
1383 unsigned int ptrace;
1386 struct llist_node wake_entry;
1388 unsigned int wakee_flips;
1389 unsigned long wakee_flip_decay_ts;
1390 struct task_struct *last_wakee;
1396 int prio, static_prio, normal_prio;
1397 unsigned int rt_priority;
1398 const struct sched_class *sched_class;
1399 struct sched_entity se;
1400 struct sched_rt_entity rt;
1401 #ifdef CONFIG_CGROUP_SCHED
1402 struct task_group *sched_task_group;
1404 struct sched_dl_entity dl;
1406 #ifdef CONFIG_PREEMPT_NOTIFIERS
1407 /* list of struct preempt_notifier: */
1408 struct hlist_head preempt_notifiers;
1411 #ifdef CONFIG_BLK_DEV_IO_TRACE
1412 unsigned int btrace_seq;
1415 unsigned int policy;
1416 int nr_cpus_allowed;
1417 cpumask_t cpus_allowed;
1419 #ifdef CONFIG_PREEMPT_RCU
1420 int rcu_read_lock_nesting;
1421 union rcu_special rcu_read_unlock_special;
1422 struct list_head rcu_node_entry;
1423 struct rcu_node *rcu_blocked_node;
1424 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1425 #ifdef CONFIG_TASKS_RCU
1426 unsigned long rcu_tasks_nvcsw;
1427 bool rcu_tasks_holdout;
1428 struct list_head rcu_tasks_holdout_list;
1429 int rcu_tasks_idle_cpu;
1430 #endif /* #ifdef CONFIG_TASKS_RCU */
1432 #ifdef CONFIG_SCHED_INFO
1433 struct sched_info sched_info;
1436 struct list_head tasks;
1438 struct plist_node pushable_tasks;
1439 struct rb_node pushable_dl_tasks;
1442 struct mm_struct *mm, *active_mm;
1443 /* per-thread vma caching */
1444 u32 vmacache_seqnum;
1445 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1446 #if defined(SPLIT_RSS_COUNTING)
1447 struct task_rss_stat rss_stat;
1451 int exit_code, exit_signal;
1452 int pdeath_signal; /* The signal sent when the parent dies */
1453 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1455 /* Used for emulating ABI behavior of previous Linux versions */
1456 unsigned int personality;
1458 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1460 unsigned in_iowait:1;
1462 /* Revert to default priority/policy when forking */
1463 unsigned sched_reset_on_fork:1;
1464 unsigned sched_contributes_to_load:1;
1465 unsigned sched_migrated:1;
1467 #ifdef CONFIG_MEMCG_KMEM
1468 unsigned memcg_kmem_skip_account:1;
1470 #ifdef CONFIG_COMPAT_BRK
1471 unsigned brk_randomized:1;
1474 unsigned long atomic_flags; /* Flags needing atomic access. */
1476 struct restart_block restart_block;
1481 #ifdef CONFIG_CC_STACKPROTECTOR
1482 /* Canary value for the -fstack-protector gcc feature */
1483 unsigned long stack_canary;
1486 * pointers to (original) parent process, youngest child, younger sibling,
1487 * older sibling, respectively. (p->father can be replaced with
1488 * p->real_parent->pid)
1490 struct task_struct __rcu *real_parent; /* real parent process */
1491 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1493 * children/sibling forms the list of my natural children
1495 struct list_head children; /* list of my children */
1496 struct list_head sibling; /* linkage in my parent's children list */
1497 struct task_struct *group_leader; /* threadgroup leader */
1500 * ptraced is the list of tasks this task is using ptrace on.
1501 * This includes both natural children and PTRACE_ATTACH targets.
1502 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1504 struct list_head ptraced;
1505 struct list_head ptrace_entry;
1507 /* PID/PID hash table linkage. */
1508 struct pid_link pids[PIDTYPE_MAX];
1509 struct list_head thread_group;
1510 struct list_head thread_node;
1512 struct completion *vfork_done; /* for vfork() */
1513 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1514 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1516 cputime_t utime, stime, utimescaled, stimescaled;
1518 struct prev_cputime prev_cputime;
1519 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1520 seqlock_t vtime_seqlock;
1521 unsigned long long vtime_snap;
1526 } vtime_snap_whence;
1528 unsigned long nvcsw, nivcsw; /* context switch counts */
1529 u64 start_time; /* monotonic time in nsec */
1530 u64 real_start_time; /* boot based time in nsec */
1531 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1532 unsigned long min_flt, maj_flt;
1534 struct task_cputime cputime_expires;
1535 struct list_head cpu_timers[3];
1537 /* process credentials */
1538 const struct cred __rcu *real_cred; /* objective and real subjective task
1539 * credentials (COW) */
1540 const struct cred __rcu *cred; /* effective (overridable) subjective task
1541 * credentials (COW) */
1542 char comm[TASK_COMM_LEN]; /* executable name excluding path
1543 - access with [gs]et_task_comm (which lock
1544 it with task_lock())
1545 - initialized normally by setup_new_exec */
1546 /* file system info */
1547 struct nameidata *nameidata;
1548 #ifdef CONFIG_SYSVIPC
1550 struct sysv_sem sysvsem;
1551 struct sysv_shm sysvshm;
1553 #ifdef CONFIG_DETECT_HUNG_TASK
1554 /* hung task detection */
1555 unsigned long last_switch_count;
1557 /* filesystem information */
1558 struct fs_struct *fs;
1559 /* open file information */
1560 struct files_struct *files;
1562 struct nsproxy *nsproxy;
1563 /* signal handlers */
1564 struct signal_struct *signal;
1565 struct sighand_struct *sighand;
1567 sigset_t blocked, real_blocked;
1568 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1569 struct sigpending pending;
1571 unsigned long sas_ss_sp;
1573 int (*notifier)(void *priv);
1574 void *notifier_data;
1575 sigset_t *notifier_mask;
1576 struct callback_head *task_works;
1578 struct audit_context *audit_context;
1579 #ifdef CONFIG_AUDITSYSCALL
1581 unsigned int sessionid;
1583 struct seccomp seccomp;
1585 /* Thread group tracking */
1588 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1590 spinlock_t alloc_lock;
1592 /* Protection of the PI data structures: */
1593 raw_spinlock_t pi_lock;
1595 struct wake_q_node wake_q;
1597 #ifdef CONFIG_RT_MUTEXES
1598 /* PI waiters blocked on a rt_mutex held by this task */
1599 struct rb_root pi_waiters;
1600 struct rb_node *pi_waiters_leftmost;
1601 /* Deadlock detection and priority inheritance handling */
1602 struct rt_mutex_waiter *pi_blocked_on;
1605 #ifdef CONFIG_DEBUG_MUTEXES
1606 /* mutex deadlock detection */
1607 struct mutex_waiter *blocked_on;
1609 #ifdef CONFIG_TRACE_IRQFLAGS
1610 unsigned int irq_events;
1611 unsigned long hardirq_enable_ip;
1612 unsigned long hardirq_disable_ip;
1613 unsigned int hardirq_enable_event;
1614 unsigned int hardirq_disable_event;
1615 int hardirqs_enabled;
1616 int hardirq_context;
1617 unsigned long softirq_disable_ip;
1618 unsigned long softirq_enable_ip;
1619 unsigned int softirq_disable_event;
1620 unsigned int softirq_enable_event;
1621 int softirqs_enabled;
1622 int softirq_context;
1624 #ifdef CONFIG_LOCKDEP
1625 # define MAX_LOCK_DEPTH 48UL
1628 unsigned int lockdep_recursion;
1629 struct held_lock held_locks[MAX_LOCK_DEPTH];
1630 gfp_t lockdep_reclaim_gfp;
1633 /* journalling filesystem info */
1636 /* stacked block device info */
1637 struct bio_list *bio_list;
1640 /* stack plugging */
1641 struct blk_plug *plug;
1645 struct reclaim_state *reclaim_state;
1647 struct backing_dev_info *backing_dev_info;
1649 struct io_context *io_context;
1651 unsigned long ptrace_message;
1652 siginfo_t *last_siginfo; /* For ptrace use. */
1653 struct task_io_accounting ioac;
1654 #if defined(CONFIG_TASK_XACCT)
1655 u64 acct_rss_mem1; /* accumulated rss usage */
1656 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1657 cputime_t acct_timexpd; /* stime + utime since last update */
1659 #ifdef CONFIG_CPUSETS
1660 nodemask_t mems_allowed; /* Protected by alloc_lock */
1661 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1662 int cpuset_mem_spread_rotor;
1663 int cpuset_slab_spread_rotor;
1665 #ifdef CONFIG_CGROUPS
1666 /* Control Group info protected by css_set_lock */
1667 struct css_set __rcu *cgroups;
1668 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1669 struct list_head cg_list;
1672 struct robust_list_head __user *robust_list;
1673 #ifdef CONFIG_COMPAT
1674 struct compat_robust_list_head __user *compat_robust_list;
1676 struct list_head pi_state_list;
1677 struct futex_pi_state *pi_state_cache;
1679 #ifdef CONFIG_PERF_EVENTS
1680 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1681 struct mutex perf_event_mutex;
1682 struct list_head perf_event_list;
1684 #ifdef CONFIG_DEBUG_PREEMPT
1685 unsigned long preempt_disable_ip;
1688 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1690 short pref_node_fork;
1692 #ifdef CONFIG_NUMA_BALANCING
1694 unsigned int numa_scan_period;
1695 unsigned int numa_scan_period_max;
1696 int numa_preferred_nid;
1697 unsigned long numa_migrate_retry;
1698 u64 node_stamp; /* migration stamp */
1699 u64 last_task_numa_placement;
1700 u64 last_sum_exec_runtime;
1701 struct callback_head numa_work;
1703 struct list_head numa_entry;
1704 struct numa_group *numa_group;
1707 * numa_faults is an array split into four regions:
1708 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1709 * in this precise order.
1711 * faults_memory: Exponential decaying average of faults on a per-node
1712 * basis. Scheduling placement decisions are made based on these
1713 * counts. The values remain static for the duration of a PTE scan.
1714 * faults_cpu: Track the nodes the process was running on when a NUMA
1715 * hinting fault was incurred.
1716 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1717 * during the current scan window. When the scan completes, the counts
1718 * in faults_memory and faults_cpu decay and these values are copied.
1720 unsigned long *numa_faults;
1721 unsigned long total_numa_faults;
1724 * numa_faults_locality tracks if faults recorded during the last
1725 * scan window were remote/local or failed to migrate. The task scan
1726 * period is adapted based on the locality of the faults with different
1727 * weights depending on whether they were shared or private faults
1729 unsigned long numa_faults_locality[3];
1731 unsigned long numa_pages_migrated;
1732 #endif /* CONFIG_NUMA_BALANCING */
1734 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1735 struct tlbflush_unmap_batch tlb_ubc;
1738 struct rcu_head rcu;
1741 * cache last used pipe for splice
1743 struct pipe_inode_info *splice_pipe;
1745 struct page_frag task_frag;
1747 #ifdef CONFIG_TASK_DELAY_ACCT
1748 struct task_delay_info *delays;
1750 #ifdef CONFIG_FAULT_INJECTION
1754 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1755 * balance_dirty_pages() for some dirty throttling pause
1758 int nr_dirtied_pause;
1759 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1761 #ifdef CONFIG_LATENCYTOP
1762 int latency_record_count;
1763 struct latency_record latency_record[LT_SAVECOUNT];
1766 * time slack values; these are used to round up poll() and
1767 * select() etc timeout values. These are in nanoseconds.
1769 unsigned long timer_slack_ns;
1770 unsigned long default_timer_slack_ns;
1773 unsigned int kasan_depth;
1775 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1776 /* Index of current stored address in ret_stack */
1778 /* Stack of return addresses for return function tracing */
1779 struct ftrace_ret_stack *ret_stack;
1780 /* time stamp for last schedule */
1781 unsigned long long ftrace_timestamp;
1783 * Number of functions that haven't been traced
1784 * because of depth overrun.
1786 atomic_t trace_overrun;
1787 /* Pause for the tracing */
1788 atomic_t tracing_graph_pause;
1790 #ifdef CONFIG_TRACING
1791 /* state flags for use by tracers */
1792 unsigned long trace;
1793 /* bitmask and counter of trace recursion */
1794 unsigned long trace_recursion;
1795 #endif /* CONFIG_TRACING */
1797 struct memcg_oom_info {
1798 struct mem_cgroup *memcg;
1801 unsigned int may_oom:1;
1804 #ifdef CONFIG_UPROBES
1805 struct uprobe_task *utask;
1807 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1808 unsigned int sequential_io;
1809 unsigned int sequential_io_avg;
1811 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1812 unsigned long task_state_change;
1814 int pagefault_disabled;
1815 /* CPU-specific state of this task */
1816 struct thread_struct thread;
1818 * WARNING: on x86, 'thread_struct' contains a variable-sized
1819 * structure. It *MUST* be at the end of 'task_struct'.
1821 * Do not put anything below here!
1825 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1826 extern int arch_task_struct_size __read_mostly;
1828 # define arch_task_struct_size (sizeof(struct task_struct))
1831 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1832 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1834 #define TNF_MIGRATED 0x01
1835 #define TNF_NO_GROUP 0x02
1836 #define TNF_SHARED 0x04
1837 #define TNF_FAULT_LOCAL 0x08
1838 #define TNF_MIGRATE_FAIL 0x10
1840 #ifdef CONFIG_NUMA_BALANCING
1841 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1842 extern pid_t task_numa_group_id(struct task_struct *p);
1843 extern void set_numabalancing_state(bool enabled);
1844 extern void task_numa_free(struct task_struct *p);
1845 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1846 int src_nid, int dst_cpu);
1848 static inline void task_numa_fault(int last_node, int node, int pages,
1852 static inline pid_t task_numa_group_id(struct task_struct *p)
1856 static inline void set_numabalancing_state(bool enabled)
1859 static inline void task_numa_free(struct task_struct *p)
1862 static inline bool should_numa_migrate_memory(struct task_struct *p,
1863 struct page *page, int src_nid, int dst_cpu)
1869 static inline struct pid *task_pid(struct task_struct *task)
1871 return task->pids[PIDTYPE_PID].pid;
1874 static inline struct pid *task_tgid(struct task_struct *task)
1876 return task->group_leader->pids[PIDTYPE_PID].pid;
1880 * Without tasklist or rcu lock it is not safe to dereference
1881 * the result of task_pgrp/task_session even if task == current,
1882 * we can race with another thread doing sys_setsid/sys_setpgid.
1884 static inline struct pid *task_pgrp(struct task_struct *task)
1886 return task->group_leader->pids[PIDTYPE_PGID].pid;
1889 static inline struct pid *task_session(struct task_struct *task)
1891 return task->group_leader->pids[PIDTYPE_SID].pid;
1894 struct pid_namespace;
1897 * the helpers to get the task's different pids as they are seen
1898 * from various namespaces
1900 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1901 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1903 * task_xid_nr_ns() : id seen from the ns specified;
1905 * set_task_vxid() : assigns a virtual id to a task;
1907 * see also pid_nr() etc in include/linux/pid.h
1909 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1910 struct pid_namespace *ns);
1912 static inline pid_t task_pid_nr(struct task_struct *tsk)
1917 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1918 struct pid_namespace *ns)
1920 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1923 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1925 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1929 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1934 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1936 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1938 return pid_vnr(task_tgid(tsk));
1942 static inline int pid_alive(const struct task_struct *p);
1943 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1949 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1955 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1957 return task_ppid_nr_ns(tsk, &init_pid_ns);
1960 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1961 struct pid_namespace *ns)
1963 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1966 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1968 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1972 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1973 struct pid_namespace *ns)
1975 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1978 static inline pid_t task_session_vnr(struct task_struct *tsk)
1980 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1983 /* obsolete, do not use */
1984 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1986 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1990 * pid_alive - check that a task structure is not stale
1991 * @p: Task structure to be checked.
1993 * Test if a process is not yet dead (at most zombie state)
1994 * If pid_alive fails, then pointers within the task structure
1995 * can be stale and must not be dereferenced.
1997 * Return: 1 if the process is alive. 0 otherwise.
1999 static inline int pid_alive(const struct task_struct *p)
2001 return p->pids[PIDTYPE_PID].pid != NULL;
2005 * is_global_init - check if a task structure is init
2006 * @tsk: Task structure to be checked.
2008 * Check if a task structure is the first user space task the kernel created.
2010 * Return: 1 if the task structure is init. 0 otherwise.
2012 static inline int is_global_init(struct task_struct *tsk)
2014 return tsk->pid == 1;
2017 extern struct pid *cad_pid;
2019 extern void free_task(struct task_struct *tsk);
2020 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2022 extern void __put_task_struct(struct task_struct *t);
2024 static inline void put_task_struct(struct task_struct *t)
2026 if (atomic_dec_and_test(&t->usage))
2027 __put_task_struct(t);
2030 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2031 extern void task_cputime(struct task_struct *t,
2032 cputime_t *utime, cputime_t *stime);
2033 extern void task_cputime_scaled(struct task_struct *t,
2034 cputime_t *utimescaled, cputime_t *stimescaled);
2035 extern cputime_t task_gtime(struct task_struct *t);
2037 static inline void task_cputime(struct task_struct *t,
2038 cputime_t *utime, cputime_t *stime)
2046 static inline void task_cputime_scaled(struct task_struct *t,
2047 cputime_t *utimescaled,
2048 cputime_t *stimescaled)
2051 *utimescaled = t->utimescaled;
2053 *stimescaled = t->stimescaled;
2056 static inline cputime_t task_gtime(struct task_struct *t)
2061 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2062 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2067 #define PF_EXITING 0x00000004 /* getting shut down */
2068 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2069 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2070 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2071 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2072 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2073 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2074 #define PF_DUMPCORE 0x00000200 /* dumped core */
2075 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2076 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2077 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2078 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2079 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2080 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2081 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2082 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2083 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2084 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2085 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2086 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2087 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2088 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2089 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2090 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2091 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2092 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2093 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2096 * Only the _current_ task can read/write to tsk->flags, but other
2097 * tasks can access tsk->flags in readonly mode for example
2098 * with tsk_used_math (like during threaded core dumping).
2099 * There is however an exception to this rule during ptrace
2100 * or during fork: the ptracer task is allowed to write to the
2101 * child->flags of its traced child (same goes for fork, the parent
2102 * can write to the child->flags), because we're guaranteed the
2103 * child is not running and in turn not changing child->flags
2104 * at the same time the parent does it.
2106 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2107 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2108 #define clear_used_math() clear_stopped_child_used_math(current)
2109 #define set_used_math() set_stopped_child_used_math(current)
2110 #define conditional_stopped_child_used_math(condition, child) \
2111 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2112 #define conditional_used_math(condition) \
2113 conditional_stopped_child_used_math(condition, current)
2114 #define copy_to_stopped_child_used_math(child) \
2115 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2116 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2117 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2118 #define used_math() tsk_used_math(current)
2120 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2121 * __GFP_FS is also cleared as it implies __GFP_IO.
2123 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2125 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2126 flags &= ~(__GFP_IO | __GFP_FS);
2130 static inline unsigned int memalloc_noio_save(void)
2132 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2133 current->flags |= PF_MEMALLOC_NOIO;
2137 static inline void memalloc_noio_restore(unsigned int flags)
2139 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2142 /* Per-process atomic flags. */
2143 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2144 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2145 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2148 #define TASK_PFA_TEST(name, func) \
2149 static inline bool task_##func(struct task_struct *p) \
2150 { return test_bit(PFA_##name, &p->atomic_flags); }
2151 #define TASK_PFA_SET(name, func) \
2152 static inline void task_set_##func(struct task_struct *p) \
2153 { set_bit(PFA_##name, &p->atomic_flags); }
2154 #define TASK_PFA_CLEAR(name, func) \
2155 static inline void task_clear_##func(struct task_struct *p) \
2156 { clear_bit(PFA_##name, &p->atomic_flags); }
2158 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2159 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2161 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2162 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2163 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2165 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2166 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2167 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2170 * task->jobctl flags
2172 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2174 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2175 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2176 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2177 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2178 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2179 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2180 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2182 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2183 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2184 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2185 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2186 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2187 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2188 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2190 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2191 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2193 extern bool task_set_jobctl_pending(struct task_struct *task,
2194 unsigned long mask);
2195 extern void task_clear_jobctl_trapping(struct task_struct *task);
2196 extern void task_clear_jobctl_pending(struct task_struct *task,
2197 unsigned long mask);
2199 static inline void rcu_copy_process(struct task_struct *p)
2201 #ifdef CONFIG_PREEMPT_RCU
2202 p->rcu_read_lock_nesting = 0;
2203 p->rcu_read_unlock_special.s = 0;
2204 p->rcu_blocked_node = NULL;
2205 INIT_LIST_HEAD(&p->rcu_node_entry);
2206 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2207 #ifdef CONFIG_TASKS_RCU
2208 p->rcu_tasks_holdout = false;
2209 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2210 p->rcu_tasks_idle_cpu = -1;
2211 #endif /* #ifdef CONFIG_TASKS_RCU */
2214 static inline void tsk_restore_flags(struct task_struct *task,
2215 unsigned long orig_flags, unsigned long flags)
2217 task->flags &= ~flags;
2218 task->flags |= orig_flags & flags;
2221 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2222 const struct cpumask *trial);
2223 extern int task_can_attach(struct task_struct *p,
2224 const struct cpumask *cs_cpus_allowed);
2226 extern void do_set_cpus_allowed(struct task_struct *p,
2227 const struct cpumask *new_mask);
2229 extern int set_cpus_allowed_ptr(struct task_struct *p,
2230 const struct cpumask *new_mask);
2232 static inline void do_set_cpus_allowed(struct task_struct *p,
2233 const struct cpumask *new_mask)
2236 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2237 const struct cpumask *new_mask)
2239 if (!cpumask_test_cpu(0, new_mask))
2245 #ifdef CONFIG_NO_HZ_COMMON
2246 void calc_load_enter_idle(void);
2247 void calc_load_exit_idle(void);
2249 static inline void calc_load_enter_idle(void) { }
2250 static inline void calc_load_exit_idle(void) { }
2251 #endif /* CONFIG_NO_HZ_COMMON */
2254 * Do not use outside of architecture code which knows its limitations.
2256 * sched_clock() has no promise of monotonicity or bounded drift between
2257 * CPUs, use (which you should not) requires disabling IRQs.
2259 * Please use one of the three interfaces below.
2261 extern unsigned long long notrace sched_clock(void);
2263 * See the comment in kernel/sched/clock.c
2265 extern u64 cpu_clock(int cpu);
2266 extern u64 local_clock(void);
2267 extern u64 running_clock(void);
2268 extern u64 sched_clock_cpu(int cpu);
2271 extern void sched_clock_init(void);
2273 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2274 static inline void sched_clock_tick(void)
2278 static inline void sched_clock_idle_sleep_event(void)
2282 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2287 * Architectures can set this to 1 if they have specified
2288 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2289 * but then during bootup it turns out that sched_clock()
2290 * is reliable after all:
2292 extern int sched_clock_stable(void);
2293 extern void set_sched_clock_stable(void);
2294 extern void clear_sched_clock_stable(void);
2296 extern void sched_clock_tick(void);
2297 extern void sched_clock_idle_sleep_event(void);
2298 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2301 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2303 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2304 * The reason for this explicit opt-in is not to have perf penalty with
2305 * slow sched_clocks.
2307 extern void enable_sched_clock_irqtime(void);
2308 extern void disable_sched_clock_irqtime(void);
2310 static inline void enable_sched_clock_irqtime(void) {}
2311 static inline void disable_sched_clock_irqtime(void) {}
2314 extern unsigned long long
2315 task_sched_runtime(struct task_struct *task);
2317 /* sched_exec is called by processes performing an exec */
2319 extern void sched_exec(void);
2321 #define sched_exec() {}
2324 extern void sched_clock_idle_sleep_event(void);
2325 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2327 #ifdef CONFIG_HOTPLUG_CPU
2328 extern void idle_task_exit(void);
2330 static inline void idle_task_exit(void) {}
2333 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2334 extern void wake_up_nohz_cpu(int cpu);
2336 static inline void wake_up_nohz_cpu(int cpu) { }
2339 #ifdef CONFIG_NO_HZ_FULL
2340 extern bool sched_can_stop_tick(void);
2341 extern u64 scheduler_tick_max_deferment(void);
2343 static inline bool sched_can_stop_tick(void) { return false; }
2346 #ifdef CONFIG_SCHED_AUTOGROUP
2347 extern void sched_autogroup_create_attach(struct task_struct *p);
2348 extern void sched_autogroup_detach(struct task_struct *p);
2349 extern void sched_autogroup_fork(struct signal_struct *sig);
2350 extern void sched_autogroup_exit(struct signal_struct *sig);
2351 #ifdef CONFIG_PROC_FS
2352 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2353 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2356 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2357 static inline void sched_autogroup_detach(struct task_struct *p) { }
2358 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2359 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2362 extern int yield_to(struct task_struct *p, bool preempt);
2363 extern void set_user_nice(struct task_struct *p, long nice);
2364 extern int task_prio(const struct task_struct *p);
2366 * task_nice - return the nice value of a given task.
2367 * @p: the task in question.
2369 * Return: The nice value [ -20 ... 0 ... 19 ].
2371 static inline int task_nice(const struct task_struct *p)
2373 return PRIO_TO_NICE((p)->static_prio);
2375 extern int can_nice(const struct task_struct *p, const int nice);
2376 extern int task_curr(const struct task_struct *p);
2377 extern int idle_cpu(int cpu);
2378 extern int sched_setscheduler(struct task_struct *, int,
2379 const struct sched_param *);
2380 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2381 const struct sched_param *);
2382 extern int sched_setattr(struct task_struct *,
2383 const struct sched_attr *);
2384 extern struct task_struct *idle_task(int cpu);
2386 * is_idle_task - is the specified task an idle task?
2387 * @p: the task in question.
2389 * Return: 1 if @p is an idle task. 0 otherwise.
2391 static inline bool is_idle_task(const struct task_struct *p)
2395 extern struct task_struct *curr_task(int cpu);
2396 extern void set_curr_task(int cpu, struct task_struct *p);
2400 union thread_union {
2401 struct thread_info thread_info;
2402 unsigned long stack[THREAD_SIZE/sizeof(long)];
2405 #ifndef __HAVE_ARCH_KSTACK_END
2406 static inline int kstack_end(void *addr)
2408 /* Reliable end of stack detection:
2409 * Some APM bios versions misalign the stack
2411 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2415 extern union thread_union init_thread_union;
2416 extern struct task_struct init_task;
2418 extern struct mm_struct init_mm;
2420 extern struct pid_namespace init_pid_ns;
2423 * find a task by one of its numerical ids
2425 * find_task_by_pid_ns():
2426 * finds a task by its pid in the specified namespace
2427 * find_task_by_vpid():
2428 * finds a task by its virtual pid
2430 * see also find_vpid() etc in include/linux/pid.h
2433 extern struct task_struct *find_task_by_vpid(pid_t nr);
2434 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2435 struct pid_namespace *ns);
2437 /* per-UID process charging. */
2438 extern struct user_struct * alloc_uid(kuid_t);
2439 static inline struct user_struct *get_uid(struct user_struct *u)
2441 atomic_inc(&u->__count);
2444 extern void free_uid(struct user_struct *);
2446 #include <asm/current.h>
2448 extern void xtime_update(unsigned long ticks);
2450 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2451 extern int wake_up_process(struct task_struct *tsk);
2452 extern void wake_up_new_task(struct task_struct *tsk);
2454 extern void kick_process(struct task_struct *tsk);
2456 static inline void kick_process(struct task_struct *tsk) { }
2458 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2459 extern void sched_dead(struct task_struct *p);
2461 extern void proc_caches_init(void);
2462 extern void flush_signals(struct task_struct *);
2463 extern void ignore_signals(struct task_struct *);
2464 extern void flush_signal_handlers(struct task_struct *, int force_default);
2465 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2467 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2469 unsigned long flags;
2472 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2473 ret = dequeue_signal(tsk, mask, info);
2474 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2479 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2481 extern void unblock_all_signals(void);
2482 extern void release_task(struct task_struct * p);
2483 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2484 extern int force_sigsegv(int, struct task_struct *);
2485 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2486 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2487 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2488 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2489 const struct cred *, u32);
2490 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2491 extern int kill_pid(struct pid *pid, int sig, int priv);
2492 extern int kill_proc_info(int, struct siginfo *, pid_t);
2493 extern __must_check bool do_notify_parent(struct task_struct *, int);
2494 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2495 extern void force_sig(int, struct task_struct *);
2496 extern int send_sig(int, struct task_struct *, int);
2497 extern int zap_other_threads(struct task_struct *p);
2498 extern struct sigqueue *sigqueue_alloc(void);
2499 extern void sigqueue_free(struct sigqueue *);
2500 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2501 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2503 static inline void restore_saved_sigmask(void)
2505 if (test_and_clear_restore_sigmask())
2506 __set_current_blocked(¤t->saved_sigmask);
2509 static inline sigset_t *sigmask_to_save(void)
2511 sigset_t *res = ¤t->blocked;
2512 if (unlikely(test_restore_sigmask()))
2513 res = ¤t->saved_sigmask;
2517 static inline int kill_cad_pid(int sig, int priv)
2519 return kill_pid(cad_pid, sig, priv);
2522 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2523 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2524 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2525 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2528 * True if we are on the alternate signal stack.
2530 static inline int on_sig_stack(unsigned long sp)
2532 #ifdef CONFIG_STACK_GROWSUP
2533 return sp >= current->sas_ss_sp &&
2534 sp - current->sas_ss_sp < current->sas_ss_size;
2536 return sp > current->sas_ss_sp &&
2537 sp - current->sas_ss_sp <= current->sas_ss_size;
2541 static inline int sas_ss_flags(unsigned long sp)
2543 if (!current->sas_ss_size)
2546 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2549 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2551 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2552 #ifdef CONFIG_STACK_GROWSUP
2553 return current->sas_ss_sp;
2555 return current->sas_ss_sp + current->sas_ss_size;
2561 * Routines for handling mm_structs
2563 extern struct mm_struct * mm_alloc(void);
2565 /* mmdrop drops the mm and the page tables */
2566 extern void __mmdrop(struct mm_struct *);
2567 static inline void mmdrop(struct mm_struct * mm)
2569 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2573 /* mmput gets rid of the mappings and all user-space */
2574 extern void mmput(struct mm_struct *);
2575 /* Grab a reference to a task's mm, if it is not already going away */
2576 extern struct mm_struct *get_task_mm(struct task_struct *task);
2578 * Grab a reference to a task's mm, if it is not already going away
2579 * and ptrace_may_access with the mode parameter passed to it
2582 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2583 /* Remove the current tasks stale references to the old mm_struct */
2584 extern void mm_release(struct task_struct *, struct mm_struct *);
2586 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2587 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2588 struct task_struct *, unsigned long);
2590 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2591 struct task_struct *);
2593 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2594 * via pt_regs, so ignore the tls argument passed via C. */
2595 static inline int copy_thread_tls(
2596 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2597 struct task_struct *p, unsigned long tls)
2599 return copy_thread(clone_flags, sp, arg, p);
2602 extern void flush_thread(void);
2603 extern void exit_thread(void);
2605 extern void exit_files(struct task_struct *);
2606 extern void __cleanup_sighand(struct sighand_struct *);
2608 extern void exit_itimers(struct signal_struct *);
2609 extern void flush_itimer_signals(void);
2611 extern void do_group_exit(int);
2613 extern int do_execve(struct filename *,
2614 const char __user * const __user *,
2615 const char __user * const __user *);
2616 extern int do_execveat(int, struct filename *,
2617 const char __user * const __user *,
2618 const char __user * const __user *,
2620 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2621 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2622 struct task_struct *fork_idle(int);
2623 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2625 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2626 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2628 __set_task_comm(tsk, from, false);
2630 extern char *get_task_comm(char *to, struct task_struct *tsk);
2633 void scheduler_ipi(void);
2634 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2636 static inline void scheduler_ipi(void) { }
2637 static inline unsigned long wait_task_inactive(struct task_struct *p,
2644 #define tasklist_empty() \
2645 list_empty(&init_task.tasks)
2647 #define next_task(p) \
2648 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2650 #define for_each_process(p) \
2651 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2653 extern bool current_is_single_threaded(void);
2656 * Careful: do_each_thread/while_each_thread is a double loop so
2657 * 'break' will not work as expected - use goto instead.
2659 #define do_each_thread(g, t) \
2660 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2662 #define while_each_thread(g, t) \
2663 while ((t = next_thread(t)) != g)
2665 #define __for_each_thread(signal, t) \
2666 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2668 #define for_each_thread(p, t) \
2669 __for_each_thread((p)->signal, t)
2671 /* Careful: this is a double loop, 'break' won't work as expected. */
2672 #define for_each_process_thread(p, t) \
2673 for_each_process(p) for_each_thread(p, t)
2675 static inline int get_nr_threads(struct task_struct *tsk)
2677 return tsk->signal->nr_threads;
2680 static inline bool thread_group_leader(struct task_struct *p)
2682 return p->exit_signal >= 0;
2685 /* Do to the insanities of de_thread it is possible for a process
2686 * to have the pid of the thread group leader without actually being
2687 * the thread group leader. For iteration through the pids in proc
2688 * all we care about is that we have a task with the appropriate
2689 * pid, we don't actually care if we have the right task.
2691 static inline bool has_group_leader_pid(struct task_struct *p)
2693 return task_pid(p) == p->signal->leader_pid;
2697 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2699 return p1->signal == p2->signal;
2702 static inline struct task_struct *next_thread(const struct task_struct *p)
2704 return list_entry_rcu(p->thread_group.next,
2705 struct task_struct, thread_group);
2708 static inline int thread_group_empty(struct task_struct *p)
2710 return list_empty(&p->thread_group);
2713 #define delay_group_leader(p) \
2714 (thread_group_leader(p) && !thread_group_empty(p))
2717 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2718 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2719 * pins the final release of task.io_context. Also protects ->cpuset and
2720 * ->cgroup.subsys[]. And ->vfork_done.
2722 * Nests both inside and outside of read_lock(&tasklist_lock).
2723 * It must not be nested with write_lock_irq(&tasklist_lock),
2724 * neither inside nor outside.
2726 static inline void task_lock(struct task_struct *p)
2728 spin_lock(&p->alloc_lock);
2731 static inline void task_unlock(struct task_struct *p)
2733 spin_unlock(&p->alloc_lock);
2736 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2737 unsigned long *flags);
2739 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2740 unsigned long *flags)
2742 struct sighand_struct *ret;
2744 ret = __lock_task_sighand(tsk, flags);
2745 (void)__cond_lock(&tsk->sighand->siglock, ret);
2749 static inline void unlock_task_sighand(struct task_struct *tsk,
2750 unsigned long *flags)
2752 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2756 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2757 * @tsk: task causing the changes
2759 * All operations which modify a threadgroup - a new thread joining the
2760 * group, death of a member thread (the assertion of PF_EXITING) and
2761 * exec(2) dethreading the process and replacing the leader - are wrapped
2762 * by threadgroup_change_{begin|end}(). This is to provide a place which
2763 * subsystems needing threadgroup stability can hook into for
2766 static inline void threadgroup_change_begin(struct task_struct *tsk)
2769 cgroup_threadgroup_change_begin(tsk);
2773 * threadgroup_change_end - mark the end of changes to a threadgroup
2774 * @tsk: task causing the changes
2776 * See threadgroup_change_begin().
2778 static inline void threadgroup_change_end(struct task_struct *tsk)
2780 cgroup_threadgroup_change_end(tsk);
2783 #ifndef __HAVE_THREAD_FUNCTIONS
2785 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2786 #define task_stack_page(task) ((task)->stack)
2788 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2790 *task_thread_info(p) = *task_thread_info(org);
2791 task_thread_info(p)->task = p;
2795 * Return the address of the last usable long on the stack.
2797 * When the stack grows down, this is just above the thread
2798 * info struct. Going any lower will corrupt the threadinfo.
2800 * When the stack grows up, this is the highest address.
2801 * Beyond that position, we corrupt data on the next page.
2803 static inline unsigned long *end_of_stack(struct task_struct *p)
2805 #ifdef CONFIG_STACK_GROWSUP
2806 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2808 return (unsigned long *)(task_thread_info(p) + 1);
2813 #define task_stack_end_corrupted(task) \
2814 (*(end_of_stack(task)) != STACK_END_MAGIC)
2816 static inline int object_is_on_stack(void *obj)
2818 void *stack = task_stack_page(current);
2820 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2823 extern void thread_info_cache_init(void);
2825 #ifdef CONFIG_DEBUG_STACK_USAGE
2826 static inline unsigned long stack_not_used(struct task_struct *p)
2828 unsigned long *n = end_of_stack(p);
2830 do { /* Skip over canary */
2834 return (unsigned long)n - (unsigned long)end_of_stack(p);
2837 extern void set_task_stack_end_magic(struct task_struct *tsk);
2839 /* set thread flags in other task's structures
2840 * - see asm/thread_info.h for TIF_xxxx flags available
2842 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2844 set_ti_thread_flag(task_thread_info(tsk), flag);
2847 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2849 clear_ti_thread_flag(task_thread_info(tsk), flag);
2852 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2854 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2857 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2859 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2862 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2864 return test_ti_thread_flag(task_thread_info(tsk), flag);
2867 static inline void set_tsk_need_resched(struct task_struct *tsk)
2869 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2872 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2874 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2877 static inline int test_tsk_need_resched(struct task_struct *tsk)
2879 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2882 static inline int restart_syscall(void)
2884 set_tsk_thread_flag(current, TIF_SIGPENDING);
2885 return -ERESTARTNOINTR;
2888 static inline int signal_pending(struct task_struct *p)
2890 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2893 static inline int __fatal_signal_pending(struct task_struct *p)
2895 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2898 static inline int fatal_signal_pending(struct task_struct *p)
2900 return signal_pending(p) && __fatal_signal_pending(p);
2903 static inline int signal_pending_state(long state, struct task_struct *p)
2905 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2907 if (!signal_pending(p))
2910 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2914 * cond_resched() and cond_resched_lock(): latency reduction via
2915 * explicit rescheduling in places that are safe. The return
2916 * value indicates whether a reschedule was done in fact.
2917 * cond_resched_lock() will drop the spinlock before scheduling,
2918 * cond_resched_softirq() will enable bhs before scheduling.
2920 extern int _cond_resched(void);
2922 #define cond_resched() ({ \
2923 ___might_sleep(__FILE__, __LINE__, 0); \
2927 extern int __cond_resched_lock(spinlock_t *lock);
2929 #define cond_resched_lock(lock) ({ \
2930 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2931 __cond_resched_lock(lock); \
2934 extern int __cond_resched_softirq(void);
2936 #define cond_resched_softirq() ({ \
2937 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2938 __cond_resched_softirq(); \
2941 static inline void cond_resched_rcu(void)
2943 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2951 * Does a critical section need to be broken due to another
2952 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2953 * but a general need for low latency)
2955 static inline int spin_needbreak(spinlock_t *lock)
2957 #ifdef CONFIG_PREEMPT
2958 return spin_is_contended(lock);
2965 * Idle thread specific functions to determine the need_resched
2968 #ifdef TIF_POLLING_NRFLAG
2969 static inline int tsk_is_polling(struct task_struct *p)
2971 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2974 static inline void __current_set_polling(void)
2976 set_thread_flag(TIF_POLLING_NRFLAG);
2979 static inline bool __must_check current_set_polling_and_test(void)
2981 __current_set_polling();
2984 * Polling state must be visible before we test NEED_RESCHED,
2985 * paired by resched_curr()
2987 smp_mb__after_atomic();
2989 return unlikely(tif_need_resched());
2992 static inline void __current_clr_polling(void)
2994 clear_thread_flag(TIF_POLLING_NRFLAG);
2997 static inline bool __must_check current_clr_polling_and_test(void)
2999 __current_clr_polling();
3002 * Polling state must be visible before we test NEED_RESCHED,
3003 * paired by resched_curr()
3005 smp_mb__after_atomic();
3007 return unlikely(tif_need_resched());
3011 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3012 static inline void __current_set_polling(void) { }
3013 static inline void __current_clr_polling(void) { }
3015 static inline bool __must_check current_set_polling_and_test(void)
3017 return unlikely(tif_need_resched());
3019 static inline bool __must_check current_clr_polling_and_test(void)
3021 return unlikely(tif_need_resched());
3025 static inline void current_clr_polling(void)
3027 __current_clr_polling();
3030 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3031 * Once the bit is cleared, we'll get IPIs with every new
3032 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3035 smp_mb(); /* paired with resched_curr() */
3037 preempt_fold_need_resched();
3040 static __always_inline bool need_resched(void)
3042 return unlikely(tif_need_resched());
3046 * Thread group CPU time accounting.
3048 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3049 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3052 * Reevaluate whether the task has signals pending delivery.
3053 * Wake the task if so.
3054 * This is required every time the blocked sigset_t changes.
3055 * callers must hold sighand->siglock.
3057 extern void recalc_sigpending_and_wake(struct task_struct *t);
3058 extern void recalc_sigpending(void);
3060 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3062 static inline void signal_wake_up(struct task_struct *t, bool resume)
3064 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3066 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3068 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3072 * Wrappers for p->thread_info->cpu access. No-op on UP.
3076 static inline unsigned int task_cpu(const struct task_struct *p)
3078 return task_thread_info(p)->cpu;
3081 static inline int task_node(const struct task_struct *p)
3083 return cpu_to_node(task_cpu(p));
3086 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3090 static inline unsigned int task_cpu(const struct task_struct *p)
3095 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3099 #endif /* CONFIG_SMP */
3101 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3102 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3104 #ifdef CONFIG_CGROUP_SCHED
3105 extern struct task_group root_task_group;
3106 #endif /* CONFIG_CGROUP_SCHED */
3108 extern int task_can_switch_user(struct user_struct *up,
3109 struct task_struct *tsk);
3111 #ifdef CONFIG_TASK_XACCT
3112 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3114 tsk->ioac.rchar += amt;
3117 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3119 tsk->ioac.wchar += amt;
3122 static inline void inc_syscr(struct task_struct *tsk)
3127 static inline void inc_syscw(struct task_struct *tsk)
3132 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3136 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3140 static inline void inc_syscr(struct task_struct *tsk)
3144 static inline void inc_syscw(struct task_struct *tsk)
3149 #ifndef TASK_SIZE_OF
3150 #define TASK_SIZE_OF(tsk) TASK_SIZE
3154 extern void mm_update_next_owner(struct mm_struct *mm);
3156 static inline void mm_update_next_owner(struct mm_struct *mm)
3159 #endif /* CONFIG_MEMCG */
3161 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3164 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3167 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3170 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3173 static inline unsigned long rlimit(unsigned int limit)
3175 return task_rlimit(current, limit);
3178 static inline unsigned long rlimit_max(unsigned int limit)
3180 return task_rlimit_max(current, limit);