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>
62 #include <asm/processor.h>
64 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
67 * Extended scheduling parameters data structure.
69 * This is needed because the original struct sched_param can not be
70 * altered without introducing ABI issues with legacy applications
71 * (e.g., in sched_getparam()).
73 * However, the possibility of specifying more than just a priority for
74 * the tasks may be useful for a wide variety of application fields, e.g.,
75 * multimedia, streaming, automation and control, and many others.
77 * This variant (sched_attr) is meant at describing a so-called
78 * sporadic time-constrained task. In such model a task is specified by:
79 * - the activation period or minimum instance inter-arrival time;
80 * - the maximum (or average, depending on the actual scheduling
81 * discipline) computation time of all instances, a.k.a. runtime;
82 * - the deadline (relative to the actual activation time) of each
84 * Very briefly, a periodic (sporadic) task asks for the execution of
85 * some specific computation --which is typically called an instance--
86 * (at most) every period. Moreover, each instance typically lasts no more
87 * than the runtime and must be completed by time instant t equal to
88 * the instance activation time + the deadline.
90 * This is reflected by the actual fields of the sched_attr structure:
92 * @size size of the structure, for fwd/bwd compat.
94 * @sched_policy task's scheduling policy
95 * @sched_flags for customizing the scheduler behaviour
96 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
97 * @sched_priority task's static priority (SCHED_FIFO/RR)
98 * @sched_deadline representative of the task's deadline
99 * @sched_runtime representative of the task's runtime
100 * @sched_period representative of the task's period
102 * Given this task model, there are a multiplicity of scheduling algorithms
103 * and policies, that can be used to ensure all the tasks will make their
104 * timing constraints.
106 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
107 * only user of this new interface. More information about the algorithm
108 * available in the scheduling class file or in Documentation/.
116 /* SCHED_NORMAL, SCHED_BATCH */
119 /* SCHED_FIFO, SCHED_RR */
128 struct futex_pi_state;
129 struct robust_list_head;
132 struct perf_event_context;
137 #define VMACACHE_BITS 2
138 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
139 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
142 * These are the constant used to fake the fixed-point load-average
143 * counting. Some notes:
144 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
145 * a load-average precision of 10 bits integer + 11 bits fractional
146 * - if you want to count load-averages more often, you need more
147 * precision, or rounding will get you. With 2-second counting freq,
148 * the EXP_n values would be 1981, 2034 and 2043 if still using only
151 extern unsigned long avenrun[]; /* Load averages */
152 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
154 #define FSHIFT 11 /* nr of bits of precision */
155 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
156 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
157 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
158 #define EXP_5 2014 /* 1/exp(5sec/5min) */
159 #define EXP_15 2037 /* 1/exp(5sec/15min) */
161 #define CALC_LOAD(load,exp,n) \
163 load += n*(FIXED_1-exp); \
166 extern unsigned long total_forks;
167 extern int nr_threads;
168 DECLARE_PER_CPU(unsigned long, process_counts);
169 extern int nr_processes(void);
170 extern unsigned long nr_running(void);
171 extern bool single_task_running(void);
172 extern unsigned long nr_iowait(void);
173 extern unsigned long nr_iowait_cpu(int cpu);
174 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
176 extern void calc_global_load(unsigned long ticks);
178 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
179 extern void update_cpu_load_nohz(void);
181 static inline void update_cpu_load_nohz(void) { }
184 extern unsigned long get_parent_ip(unsigned long addr);
186 extern void dump_cpu_task(int cpu);
191 #ifdef CONFIG_SCHED_DEBUG
192 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
193 extern void proc_sched_set_task(struct task_struct *p);
195 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
199 * Task state bitmask. NOTE! These bits are also
200 * encoded in fs/proc/array.c: get_task_state().
202 * We have two separate sets of flags: task->state
203 * is about runnability, while task->exit_state are
204 * about the task exiting. Confusing, but this way
205 * modifying one set can't modify the other one by
208 #define TASK_RUNNING 0
209 #define TASK_INTERRUPTIBLE 1
210 #define TASK_UNINTERRUPTIBLE 2
211 #define __TASK_STOPPED 4
212 #define __TASK_TRACED 8
213 /* in tsk->exit_state */
215 #define EXIT_ZOMBIE 32
216 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
217 /* in tsk->state again */
219 #define TASK_WAKEKILL 128
220 #define TASK_WAKING 256
221 #define TASK_PARKED 512
222 #define TASK_NOLOAD 1024
223 #define TASK_STATE_MAX 2048
225 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
227 extern char ___assert_task_state[1 - 2*!!(
228 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
230 /* Convenience macros for the sake of set_task_state */
231 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
232 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
233 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
235 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
237 /* Convenience macros for the sake of wake_up */
238 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
239 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
241 /* get_task_state() */
242 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
243 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
244 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
246 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
247 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
248 #define task_is_stopped_or_traced(task) \
249 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
250 #define task_contributes_to_load(task) \
251 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
252 (task->flags & PF_FROZEN) == 0 && \
253 (task->state & TASK_NOLOAD) == 0)
255 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
257 #define __set_task_state(tsk, state_value) \
259 (tsk)->task_state_change = _THIS_IP_; \
260 (tsk)->state = (state_value); \
262 #define set_task_state(tsk, state_value) \
264 (tsk)->task_state_change = _THIS_IP_; \
265 smp_store_mb((tsk)->state, (state_value)); \
269 * set_current_state() includes a barrier so that the write of current->state
270 * is correctly serialised wrt the caller's subsequent test of whether to
273 * set_current_state(TASK_UNINTERRUPTIBLE);
274 * if (do_i_need_to_sleep())
277 * If the caller does not need such serialisation then use __set_current_state()
279 #define __set_current_state(state_value) \
281 current->task_state_change = _THIS_IP_; \
282 current->state = (state_value); \
284 #define set_current_state(state_value) \
286 current->task_state_change = _THIS_IP_; \
287 smp_store_mb(current->state, (state_value)); \
292 #define __set_task_state(tsk, state_value) \
293 do { (tsk)->state = (state_value); } while (0)
294 #define set_task_state(tsk, state_value) \
295 smp_store_mb((tsk)->state, (state_value))
298 * set_current_state() includes a barrier so that the write of current->state
299 * is correctly serialised wrt the caller's subsequent test of whether to
302 * set_current_state(TASK_UNINTERRUPTIBLE);
303 * if (do_i_need_to_sleep())
306 * If the caller does not need such serialisation then use __set_current_state()
308 #define __set_current_state(state_value) \
309 do { current->state = (state_value); } while (0)
310 #define set_current_state(state_value) \
311 smp_store_mb(current->state, (state_value))
315 /* Task command name length */
316 #define TASK_COMM_LEN 16
318 #include <linux/spinlock.h>
321 * This serializes "schedule()" and also protects
322 * the run-queue from deletions/modifications (but
323 * _adding_ to the beginning of the run-queue has
326 extern rwlock_t tasklist_lock;
327 extern spinlock_t mmlist_lock;
331 #ifdef CONFIG_PROVE_RCU
332 extern int lockdep_tasklist_lock_is_held(void);
333 #endif /* #ifdef CONFIG_PROVE_RCU */
335 extern void sched_init(void);
336 extern void sched_init_smp(void);
337 extern asmlinkage void schedule_tail(struct task_struct *prev);
338 extern void init_idle(struct task_struct *idle, int cpu);
339 extern void init_idle_bootup_task(struct task_struct *idle);
341 extern cpumask_var_t cpu_isolated_map;
343 extern int runqueue_is_locked(int cpu);
345 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
346 extern void nohz_balance_enter_idle(int cpu);
347 extern void set_cpu_sd_state_idle(void);
348 extern int get_nohz_timer_target(int pinned);
350 static inline void nohz_balance_enter_idle(int cpu) { }
351 static inline void set_cpu_sd_state_idle(void) { }
352 static inline int get_nohz_timer_target(int pinned)
354 return smp_processor_id();
359 * Only dump TASK_* tasks. (0 for all tasks)
361 extern void show_state_filter(unsigned long state_filter);
363 static inline void show_state(void)
365 show_state_filter(0);
368 extern void show_regs(struct pt_regs *);
371 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
372 * task), SP is the stack pointer of the first frame that should be shown in the back
373 * trace (or NULL if the entire call-chain of the task should be shown).
375 extern void show_stack(struct task_struct *task, unsigned long *sp);
377 extern void cpu_init (void);
378 extern void trap_init(void);
379 extern void update_process_times(int user);
380 extern void scheduler_tick(void);
382 extern void sched_show_task(struct task_struct *p);
384 #ifdef CONFIG_LOCKUP_DETECTOR
385 extern void touch_softlockup_watchdog(void);
386 extern void touch_softlockup_watchdog_sync(void);
387 extern void touch_all_softlockup_watchdogs(void);
388 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
390 size_t *lenp, loff_t *ppos);
391 extern unsigned int softlockup_panic;
392 void lockup_detector_init(void);
394 static inline void touch_softlockup_watchdog(void)
397 static inline void touch_softlockup_watchdog_sync(void)
400 static inline void touch_all_softlockup_watchdogs(void)
403 static inline void lockup_detector_init(void)
408 #ifdef CONFIG_DETECT_HUNG_TASK
409 void reset_hung_task_detector(void);
411 static inline void reset_hung_task_detector(void)
416 /* Attach to any functions which should be ignored in wchan output. */
417 #define __sched __attribute__((__section__(".sched.text")))
419 /* Linker adds these: start and end of __sched functions */
420 extern char __sched_text_start[], __sched_text_end[];
422 /* Is this address in the __sched functions? */
423 extern int in_sched_functions(unsigned long addr);
425 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
426 extern signed long schedule_timeout(signed long timeout);
427 extern signed long schedule_timeout_interruptible(signed long timeout);
428 extern signed long schedule_timeout_killable(signed long timeout);
429 extern signed long schedule_timeout_uninterruptible(signed long timeout);
430 asmlinkage void schedule(void);
431 extern void schedule_preempt_disabled(void);
433 extern long io_schedule_timeout(long timeout);
435 static inline void io_schedule(void)
437 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
441 struct user_namespace;
444 extern void arch_pick_mmap_layout(struct mm_struct *mm);
446 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
447 unsigned long, unsigned long);
449 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
450 unsigned long len, unsigned long pgoff,
451 unsigned long flags);
453 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
456 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
457 #define SUID_DUMP_USER 1 /* Dump as user of process */
458 #define SUID_DUMP_ROOT 2 /* Dump as root */
462 /* for SUID_DUMP_* above */
463 #define MMF_DUMPABLE_BITS 2
464 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
466 extern void set_dumpable(struct mm_struct *mm, int value);
468 * This returns the actual value of the suid_dumpable flag. For things
469 * that are using this for checking for privilege transitions, it must
470 * test against SUID_DUMP_USER rather than treating it as a boolean
473 static inline int __get_dumpable(unsigned long mm_flags)
475 return mm_flags & MMF_DUMPABLE_MASK;
478 static inline int get_dumpable(struct mm_struct *mm)
480 return __get_dumpable(mm->flags);
483 /* coredump filter bits */
484 #define MMF_DUMP_ANON_PRIVATE 2
485 #define MMF_DUMP_ANON_SHARED 3
486 #define MMF_DUMP_MAPPED_PRIVATE 4
487 #define MMF_DUMP_MAPPED_SHARED 5
488 #define MMF_DUMP_ELF_HEADERS 6
489 #define MMF_DUMP_HUGETLB_PRIVATE 7
490 #define MMF_DUMP_HUGETLB_SHARED 8
492 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
493 #define MMF_DUMP_FILTER_BITS 7
494 #define MMF_DUMP_FILTER_MASK \
495 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
496 #define MMF_DUMP_FILTER_DEFAULT \
497 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
498 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
500 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
501 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
503 # define MMF_DUMP_MASK_DEFAULT_ELF 0
505 /* leave room for more dump flags */
506 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
507 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
508 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
510 #define MMF_HAS_UPROBES 19 /* has uprobes */
511 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
513 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
515 struct sighand_struct {
517 struct k_sigaction action[_NSIG];
519 wait_queue_head_t signalfd_wqh;
522 struct pacct_struct {
525 unsigned long ac_mem;
526 cputime_t ac_utime, ac_stime;
527 unsigned long ac_minflt, ac_majflt;
538 * struct cputime - snaphsot of system and user cputime
539 * @utime: time spent in user mode
540 * @stime: time spent in system mode
542 * Gathers a generic snapshot of user and system time.
550 * struct task_cputime - collected CPU time counts
551 * @utime: time spent in user mode, in &cputime_t units
552 * @stime: time spent in kernel mode, in &cputime_t units
553 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
555 * This is an extension of struct cputime that includes the total runtime
556 * spent by the task from the scheduler point of view.
558 * As a result, this structure groups together three kinds of CPU time
559 * that are tracked for threads and thread groups. Most things considering
560 * CPU time want to group these counts together and treat all three
561 * of them in parallel.
563 struct task_cputime {
566 unsigned long long sum_exec_runtime;
568 /* Alternate field names when used to cache expirations. */
569 #define prof_exp stime
570 #define virt_exp utime
571 #define sched_exp sum_exec_runtime
573 #define INIT_CPUTIME \
574 (struct task_cputime) { \
577 .sum_exec_runtime = 0, \
581 * This is the atomic variant of task_cputime, which can be used for
582 * storing and updating task_cputime statistics without locking.
584 struct task_cputime_atomic {
587 atomic64_t sum_exec_runtime;
590 #define INIT_CPUTIME_ATOMIC \
591 (struct task_cputime_atomic) { \
592 .utime = ATOMIC64_INIT(0), \
593 .stime = ATOMIC64_INIT(0), \
594 .sum_exec_runtime = ATOMIC64_INIT(0), \
597 #ifdef CONFIG_PREEMPT_COUNT
598 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
600 #define PREEMPT_DISABLED PREEMPT_ENABLED
604 * Disable preemption until the scheduler is running.
605 * Reset by start_kernel()->sched_init()->init_idle().
607 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
608 * before the scheduler is active -- see should_resched().
610 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
613 * struct thread_group_cputimer - thread group interval timer counts
614 * @cputime_atomic: atomic thread group interval timers.
615 * @running: non-zero when there are timers running and
616 * @cputime receives updates.
618 * This structure contains the version of task_cputime, above, that is
619 * used for thread group CPU timer calculations.
621 struct thread_group_cputimer {
622 struct task_cputime_atomic cputime_atomic;
626 #include <linux/rwsem.h>
630 * NOTE! "signal_struct" does not have its own
631 * locking, because a shared signal_struct always
632 * implies a shared sighand_struct, so locking
633 * sighand_struct is always a proper superset of
634 * the locking of signal_struct.
636 struct signal_struct {
640 struct list_head thread_head;
642 wait_queue_head_t wait_chldexit; /* for wait4() */
644 /* current thread group signal load-balancing target: */
645 struct task_struct *curr_target;
647 /* shared signal handling: */
648 struct sigpending shared_pending;
650 /* thread group exit support */
653 * - notify group_exit_task when ->count is equal to notify_count
654 * - everyone except group_exit_task is stopped during signal delivery
655 * of fatal signals, group_exit_task processes the signal.
658 struct task_struct *group_exit_task;
660 /* thread group stop support, overloads group_exit_code too */
661 int group_stop_count;
662 unsigned int flags; /* see SIGNAL_* flags below */
665 * PR_SET_CHILD_SUBREAPER marks a process, like a service
666 * manager, to re-parent orphan (double-forking) child processes
667 * to this process instead of 'init'. The service manager is
668 * able to receive SIGCHLD signals and is able to investigate
669 * the process until it calls wait(). All children of this
670 * process will inherit a flag if they should look for a
671 * child_subreaper process at exit.
673 unsigned int is_child_subreaper:1;
674 unsigned int has_child_subreaper:1;
676 /* POSIX.1b Interval Timers */
678 struct list_head posix_timers;
680 /* ITIMER_REAL timer for the process */
681 struct hrtimer real_timer;
682 struct pid *leader_pid;
683 ktime_t it_real_incr;
686 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
687 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
688 * values are defined to 0 and 1 respectively
690 struct cpu_itimer it[2];
693 * Thread group totals for process CPU timers.
694 * See thread_group_cputimer(), et al, for details.
696 struct thread_group_cputimer cputimer;
698 /* Earliest-expiration cache. */
699 struct task_cputime cputime_expires;
701 struct list_head cpu_timers[3];
703 struct pid *tty_old_pgrp;
705 /* boolean value for session group leader */
708 struct tty_struct *tty; /* NULL if no tty */
710 #ifdef CONFIG_SCHED_AUTOGROUP
711 struct autogroup *autogroup;
714 * Cumulative resource counters for dead threads in the group,
715 * and for reaped dead child processes forked by this group.
716 * Live threads maintain their own counters and add to these
717 * in __exit_signal, except for the group leader.
719 seqlock_t stats_lock;
720 cputime_t utime, stime, cutime, cstime;
723 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
724 struct cputime prev_cputime;
726 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
727 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
728 unsigned long inblock, oublock, cinblock, coublock;
729 unsigned long maxrss, cmaxrss;
730 struct task_io_accounting ioac;
733 * Cumulative ns of schedule CPU time fo dead threads in the
734 * group, not including a zombie group leader, (This only differs
735 * from jiffies_to_ns(utime + stime) if sched_clock uses something
736 * other than jiffies.)
738 unsigned long long sum_sched_runtime;
741 * We don't bother to synchronize most readers of this at all,
742 * because there is no reader checking a limit that actually needs
743 * to get both rlim_cur and rlim_max atomically, and either one
744 * alone is a single word that can safely be read normally.
745 * getrlimit/setrlimit use task_lock(current->group_leader) to
746 * protect this instead of the siglock, because they really
747 * have no need to disable irqs.
749 struct rlimit rlim[RLIM_NLIMITS];
751 #ifdef CONFIG_BSD_PROCESS_ACCT
752 struct pacct_struct pacct; /* per-process accounting information */
754 #ifdef CONFIG_TASKSTATS
755 struct taskstats *stats;
759 unsigned audit_tty_log_passwd;
760 struct tty_audit_buf *tty_audit_buf;
762 #ifdef CONFIG_CGROUPS
764 * group_rwsem prevents new tasks from entering the threadgroup and
765 * member tasks from exiting,a more specifically, setting of
766 * PF_EXITING. fork and exit paths are protected with this rwsem
767 * using threadgroup_change_begin/end(). Users which require
768 * threadgroup to remain stable should use threadgroup_[un]lock()
769 * which also takes care of exec path. Currently, cgroup is the
772 struct rw_semaphore group_rwsem;
775 oom_flags_t oom_flags;
776 short oom_score_adj; /* OOM kill score adjustment */
777 short oom_score_adj_min; /* OOM kill score adjustment min value.
778 * Only settable by CAP_SYS_RESOURCE. */
780 struct mutex cred_guard_mutex; /* guard against foreign influences on
781 * credential calculations
782 * (notably. ptrace) */
786 * Bits in flags field of signal_struct.
788 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
789 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
790 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
791 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
793 * Pending notifications to parent.
795 #define SIGNAL_CLD_STOPPED 0x00000010
796 #define SIGNAL_CLD_CONTINUED 0x00000020
797 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
799 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
801 /* If true, all threads except ->group_exit_task have pending SIGKILL */
802 static inline int signal_group_exit(const struct signal_struct *sig)
804 return (sig->flags & SIGNAL_GROUP_EXIT) ||
805 (sig->group_exit_task != NULL);
809 * Some day this will be a full-fledged user tracking system..
812 atomic_t __count; /* reference count */
813 atomic_t processes; /* How many processes does this user have? */
814 atomic_t sigpending; /* How many pending signals does this user have? */
815 #ifdef CONFIG_INOTIFY_USER
816 atomic_t inotify_watches; /* How many inotify watches does this user have? */
817 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
819 #ifdef CONFIG_FANOTIFY
820 atomic_t fanotify_listeners;
823 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
825 #ifdef CONFIG_POSIX_MQUEUE
826 /* protected by mq_lock */
827 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
829 unsigned long locked_shm; /* How many pages of mlocked shm ? */
832 struct key *uid_keyring; /* UID specific keyring */
833 struct key *session_keyring; /* UID's default session keyring */
836 /* Hash table maintenance information */
837 struct hlist_node uidhash_node;
840 #ifdef CONFIG_PERF_EVENTS
841 atomic_long_t locked_vm;
845 extern int uids_sysfs_init(void);
847 extern struct user_struct *find_user(kuid_t);
849 extern struct user_struct root_user;
850 #define INIT_USER (&root_user)
853 struct backing_dev_info;
854 struct reclaim_state;
856 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
858 /* cumulative counters */
859 unsigned long pcount; /* # of times run on this cpu */
860 unsigned long long run_delay; /* time spent waiting on a runqueue */
863 unsigned long long last_arrival,/* when we last ran on a cpu */
864 last_queued; /* when we were last queued to run */
866 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
868 #ifdef CONFIG_TASK_DELAY_ACCT
869 struct task_delay_info {
871 unsigned int flags; /* Private per-task flags */
873 /* For each stat XXX, add following, aligned appropriately
875 * struct timespec XXX_start, XXX_end;
879 * Atomicity of updates to XXX_delay, XXX_count protected by
880 * single lock above (split into XXX_lock if contention is an issue).
884 * XXX_count is incremented on every XXX operation, the delay
885 * associated with the operation is added to XXX_delay.
886 * XXX_delay contains the accumulated delay time in nanoseconds.
888 u64 blkio_start; /* Shared by blkio, swapin */
889 u64 blkio_delay; /* wait for sync block io completion */
890 u64 swapin_delay; /* wait for swapin block io completion */
891 u32 blkio_count; /* total count of the number of sync block */
892 /* io operations performed */
893 u32 swapin_count; /* total count of the number of swapin block */
894 /* io operations performed */
897 u64 freepages_delay; /* wait for memory reclaim */
898 u32 freepages_count; /* total count of memory reclaim */
900 #endif /* CONFIG_TASK_DELAY_ACCT */
902 static inline int sched_info_on(void)
904 #ifdef CONFIG_SCHEDSTATS
906 #elif defined(CONFIG_TASK_DELAY_ACCT)
907 extern int delayacct_on;
922 * Increase resolution of cpu_capacity calculations
924 #define SCHED_CAPACITY_SHIFT 10
925 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
928 * Wake-queues are lists of tasks with a pending wakeup, whose
929 * callers have already marked the task as woken internally,
930 * and can thus carry on. A common use case is being able to
931 * do the wakeups once the corresponding user lock as been
934 * We hold reference to each task in the list across the wakeup,
935 * thus guaranteeing that the memory is still valid by the time
936 * the actual wakeups are performed in wake_up_q().
938 * One per task suffices, because there's never a need for a task to be
939 * in two wake queues simultaneously; it is forbidden to abandon a task
940 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
941 * already in a wake queue, the wakeup will happen soon and the second
942 * waker can just skip it.
944 * The WAKE_Q macro declares and initializes the list head.
945 * wake_up_q() does NOT reinitialize the list; it's expected to be
946 * called near the end of a function, where the fact that the queue is
947 * not used again will be easy to see by inspection.
949 * Note that this can cause spurious wakeups. schedule() callers
950 * must ensure the call is done inside a loop, confirming that the
951 * wakeup condition has in fact occurred.
954 struct wake_q_node *next;
958 struct wake_q_node *first;
959 struct wake_q_node **lastp;
962 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
964 #define WAKE_Q(name) \
965 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
967 extern void wake_q_add(struct wake_q_head *head,
968 struct task_struct *task);
969 extern void wake_up_q(struct wake_q_head *head);
972 * sched-domains (multiprocessor balancing) declarations:
975 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
976 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
977 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
978 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
979 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
980 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
981 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
982 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
983 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
984 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
985 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
986 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
987 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
988 #define SD_NUMA 0x4000 /* cross-node balancing */
990 #ifdef CONFIG_SCHED_SMT
991 static inline int cpu_smt_flags(void)
993 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
997 #ifdef CONFIG_SCHED_MC
998 static inline int cpu_core_flags(void)
1000 return SD_SHARE_PKG_RESOURCES;
1005 static inline int cpu_numa_flags(void)
1011 struct sched_domain_attr {
1012 int relax_domain_level;
1015 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1016 .relax_domain_level = -1, \
1019 extern int sched_domain_level_max;
1023 struct sched_domain {
1024 /* These fields must be setup */
1025 struct sched_domain *parent; /* top domain must be null terminated */
1026 struct sched_domain *child; /* bottom domain must be null terminated */
1027 struct sched_group *groups; /* the balancing groups of the domain */
1028 unsigned long min_interval; /* Minimum balance interval ms */
1029 unsigned long max_interval; /* Maximum balance interval ms */
1030 unsigned int busy_factor; /* less balancing by factor if busy */
1031 unsigned int imbalance_pct; /* No balance until over watermark */
1032 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1033 unsigned int busy_idx;
1034 unsigned int idle_idx;
1035 unsigned int newidle_idx;
1036 unsigned int wake_idx;
1037 unsigned int forkexec_idx;
1038 unsigned int smt_gain;
1040 int nohz_idle; /* NOHZ IDLE status */
1041 int flags; /* See SD_* */
1044 /* Runtime fields. */
1045 unsigned long last_balance; /* init to jiffies. units in jiffies */
1046 unsigned int balance_interval; /* initialise to 1. units in ms. */
1047 unsigned int nr_balance_failed; /* initialise to 0 */
1049 /* idle_balance() stats */
1050 u64 max_newidle_lb_cost;
1051 unsigned long next_decay_max_lb_cost;
1053 #ifdef CONFIG_SCHEDSTATS
1054 /* load_balance() stats */
1055 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1056 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1057 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1058 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1059 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1060 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1061 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1062 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1064 /* Active load balancing */
1065 unsigned int alb_count;
1066 unsigned int alb_failed;
1067 unsigned int alb_pushed;
1069 /* SD_BALANCE_EXEC stats */
1070 unsigned int sbe_count;
1071 unsigned int sbe_balanced;
1072 unsigned int sbe_pushed;
1074 /* SD_BALANCE_FORK stats */
1075 unsigned int sbf_count;
1076 unsigned int sbf_balanced;
1077 unsigned int sbf_pushed;
1079 /* try_to_wake_up() stats */
1080 unsigned int ttwu_wake_remote;
1081 unsigned int ttwu_move_affine;
1082 unsigned int ttwu_move_balance;
1084 #ifdef CONFIG_SCHED_DEBUG
1088 void *private; /* used during construction */
1089 struct rcu_head rcu; /* used during destruction */
1092 unsigned int span_weight;
1094 * Span of all CPUs in this domain.
1096 * NOTE: this field is variable length. (Allocated dynamically
1097 * by attaching extra space to the end of the structure,
1098 * depending on how many CPUs the kernel has booted up with)
1100 unsigned long span[0];
1103 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1105 return to_cpumask(sd->span);
1108 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1109 struct sched_domain_attr *dattr_new);
1111 /* Allocate an array of sched domains, for partition_sched_domains(). */
1112 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1113 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1115 bool cpus_share_cache(int this_cpu, int that_cpu);
1117 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1118 typedef int (*sched_domain_flags_f)(void);
1120 #define SDTL_OVERLAP 0x01
1123 struct sched_domain **__percpu sd;
1124 struct sched_group **__percpu sg;
1125 struct sched_group_capacity **__percpu sgc;
1128 struct sched_domain_topology_level {
1129 sched_domain_mask_f mask;
1130 sched_domain_flags_f sd_flags;
1133 struct sd_data data;
1134 #ifdef CONFIG_SCHED_DEBUG
1139 extern struct sched_domain_topology_level *sched_domain_topology;
1141 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1142 extern void wake_up_if_idle(int cpu);
1144 #ifdef CONFIG_SCHED_DEBUG
1145 # define SD_INIT_NAME(type) .name = #type
1147 # define SD_INIT_NAME(type)
1150 #else /* CONFIG_SMP */
1152 struct sched_domain_attr;
1155 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1156 struct sched_domain_attr *dattr_new)
1160 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1165 #endif /* !CONFIG_SMP */
1168 struct io_context; /* See blkdev.h */
1171 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1172 extern void prefetch_stack(struct task_struct *t);
1174 static inline void prefetch_stack(struct task_struct *t) { }
1177 struct audit_context; /* See audit.c */
1179 struct pipe_inode_info;
1180 struct uts_namespace;
1182 struct load_weight {
1183 unsigned long weight;
1188 u64 last_runnable_update;
1191 * utilization_avg_contrib describes the amount of time that a
1192 * sched_entity is running on a CPU. It is based on running_avg_sum
1193 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1194 * load_avg_contrib described the amount of time that a sched_entity
1195 * is runnable on a rq. It is based on both runnable_avg_sum and the
1196 * weight of the task.
1198 unsigned long load_avg_contrib, utilization_avg_contrib;
1200 * These sums represent an infinite geometric series and so are bound
1201 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1202 * choices of y < 1-2^(-32)*1024.
1203 * running_avg_sum reflects the time that the sched_entity is
1204 * effectively running on the CPU.
1205 * runnable_avg_sum represents the amount of time a sched_entity is on
1206 * a runqueue which includes the running time that is monitored by
1209 u32 runnable_avg_sum, avg_period, running_avg_sum;
1212 #ifdef CONFIG_SCHEDSTATS
1213 struct sched_statistics {
1223 s64 sum_sleep_runtime;
1230 u64 nr_migrations_cold;
1231 u64 nr_failed_migrations_affine;
1232 u64 nr_failed_migrations_running;
1233 u64 nr_failed_migrations_hot;
1234 u64 nr_forced_migrations;
1237 u64 nr_wakeups_sync;
1238 u64 nr_wakeups_migrate;
1239 u64 nr_wakeups_local;
1240 u64 nr_wakeups_remote;
1241 u64 nr_wakeups_affine;
1242 u64 nr_wakeups_affine_attempts;
1243 u64 nr_wakeups_passive;
1244 u64 nr_wakeups_idle;
1248 struct sched_entity {
1249 struct load_weight load; /* for load-balancing */
1250 struct rb_node run_node;
1251 struct list_head group_node;
1255 u64 sum_exec_runtime;
1257 u64 prev_sum_exec_runtime;
1261 #ifdef CONFIG_SCHEDSTATS
1262 struct sched_statistics statistics;
1265 #ifdef CONFIG_FAIR_GROUP_SCHED
1267 struct sched_entity *parent;
1268 /* rq on which this entity is (to be) queued: */
1269 struct cfs_rq *cfs_rq;
1270 /* rq "owned" by this entity/group: */
1271 struct cfs_rq *my_q;
1275 /* Per-entity load-tracking */
1276 struct sched_avg avg;
1280 struct sched_rt_entity {
1281 struct list_head run_list;
1282 unsigned long timeout;
1283 unsigned long watchdog_stamp;
1284 unsigned int time_slice;
1286 struct sched_rt_entity *back;
1287 #ifdef CONFIG_RT_GROUP_SCHED
1288 struct sched_rt_entity *parent;
1289 /* rq on which this entity is (to be) queued: */
1290 struct rt_rq *rt_rq;
1291 /* rq "owned" by this entity/group: */
1296 struct sched_dl_entity {
1297 struct rb_node rb_node;
1300 * Original scheduling parameters. Copied here from sched_attr
1301 * during sched_setattr(), they will remain the same until
1302 * the next sched_setattr().
1304 u64 dl_runtime; /* maximum runtime for each instance */
1305 u64 dl_deadline; /* relative deadline of each instance */
1306 u64 dl_period; /* separation of two instances (period) */
1307 u64 dl_bw; /* dl_runtime / dl_deadline */
1310 * Actual scheduling parameters. Initialized with the values above,
1311 * they are continously updated during task execution. Note that
1312 * the remaining runtime could be < 0 in case we are in overrun.
1314 s64 runtime; /* remaining runtime for this instance */
1315 u64 deadline; /* absolute deadline for this instance */
1316 unsigned int flags; /* specifying the scheduler behaviour */
1321 * @dl_throttled tells if we exhausted the runtime. If so, the
1322 * task has to wait for a replenishment to be performed at the
1323 * next firing of dl_timer.
1325 * @dl_new tells if a new instance arrived. If so we must
1326 * start executing it with full runtime and reset its absolute
1329 * @dl_boosted tells if we are boosted due to DI. If so we are
1330 * outside bandwidth enforcement mechanism (but only until we
1331 * exit the critical section);
1333 * @dl_yielded tells if task gave up the cpu before consuming
1334 * all its available runtime during the last job.
1336 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1339 * Bandwidth enforcement timer. Each -deadline task has its
1340 * own bandwidth to be enforced, thus we need one timer per task.
1342 struct hrtimer dl_timer;
1354 enum perf_event_task_context {
1355 perf_invalid_context = -1,
1356 perf_hw_context = 0,
1358 perf_nr_task_contexts,
1361 struct task_struct {
1362 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1365 unsigned int flags; /* per process flags, defined below */
1366 unsigned int ptrace;
1369 struct llist_node wake_entry;
1371 struct task_struct *last_wakee;
1372 unsigned long wakee_flips;
1373 unsigned long wakee_flip_decay_ts;
1379 int prio, static_prio, normal_prio;
1380 unsigned int rt_priority;
1381 const struct sched_class *sched_class;
1382 struct sched_entity se;
1383 struct sched_rt_entity rt;
1384 #ifdef CONFIG_CGROUP_SCHED
1385 struct task_group *sched_task_group;
1387 struct sched_dl_entity dl;
1389 #ifdef CONFIG_PREEMPT_NOTIFIERS
1390 /* list of struct preempt_notifier: */
1391 struct hlist_head preempt_notifiers;
1394 #ifdef CONFIG_BLK_DEV_IO_TRACE
1395 unsigned int btrace_seq;
1398 unsigned int policy;
1399 int nr_cpus_allowed;
1400 cpumask_t cpus_allowed;
1402 #ifdef CONFIG_PREEMPT_RCU
1403 int rcu_read_lock_nesting;
1404 union rcu_special rcu_read_unlock_special;
1405 struct list_head rcu_node_entry;
1406 struct rcu_node *rcu_blocked_node;
1407 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1408 #ifdef CONFIG_TASKS_RCU
1409 unsigned long rcu_tasks_nvcsw;
1410 bool rcu_tasks_holdout;
1411 struct list_head rcu_tasks_holdout_list;
1412 int rcu_tasks_idle_cpu;
1413 #endif /* #ifdef CONFIG_TASKS_RCU */
1415 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1416 struct sched_info sched_info;
1419 struct list_head tasks;
1421 struct plist_node pushable_tasks;
1422 struct rb_node pushable_dl_tasks;
1425 struct mm_struct *mm, *active_mm;
1426 /* per-thread vma caching */
1427 u32 vmacache_seqnum;
1428 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1429 #if defined(SPLIT_RSS_COUNTING)
1430 struct task_rss_stat rss_stat;
1434 int exit_code, exit_signal;
1435 int pdeath_signal; /* The signal sent when the parent dies */
1436 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1438 /* Used for emulating ABI behavior of previous Linux versions */
1439 unsigned int personality;
1441 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1443 unsigned in_iowait:1;
1445 /* Revert to default priority/policy when forking */
1446 unsigned sched_reset_on_fork:1;
1447 unsigned sched_contributes_to_load:1;
1448 unsigned sched_migrated:1;
1450 #ifdef CONFIG_MEMCG_KMEM
1451 unsigned memcg_kmem_skip_account:1;
1453 #ifdef CONFIG_COMPAT_BRK
1454 unsigned brk_randomized:1;
1457 unsigned long atomic_flags; /* Flags needing atomic access. */
1459 struct restart_block restart_block;
1464 #ifdef CONFIG_CC_STACKPROTECTOR
1465 /* Canary value for the -fstack-protector gcc feature */
1466 unsigned long stack_canary;
1469 * pointers to (original) parent process, youngest child, younger sibling,
1470 * older sibling, respectively. (p->father can be replaced with
1471 * p->real_parent->pid)
1473 struct task_struct __rcu *real_parent; /* real parent process */
1474 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1476 * children/sibling forms the list of my natural children
1478 struct list_head children; /* list of my children */
1479 struct list_head sibling; /* linkage in my parent's children list */
1480 struct task_struct *group_leader; /* threadgroup leader */
1483 * ptraced is the list of tasks this task is using ptrace on.
1484 * This includes both natural children and PTRACE_ATTACH targets.
1485 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1487 struct list_head ptraced;
1488 struct list_head ptrace_entry;
1490 /* PID/PID hash table linkage. */
1491 struct pid_link pids[PIDTYPE_MAX];
1492 struct list_head thread_group;
1493 struct list_head thread_node;
1495 struct completion *vfork_done; /* for vfork() */
1496 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1497 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1499 cputime_t utime, stime, utimescaled, stimescaled;
1501 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1502 struct cputime prev_cputime;
1504 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1505 seqlock_t vtime_seqlock;
1506 unsigned long long vtime_snap;
1511 } vtime_snap_whence;
1513 unsigned long nvcsw, nivcsw; /* context switch counts */
1514 u64 start_time; /* monotonic time in nsec */
1515 u64 real_start_time; /* boot based time in nsec */
1516 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1517 unsigned long min_flt, maj_flt;
1519 struct task_cputime cputime_expires;
1520 struct list_head cpu_timers[3];
1522 /* process credentials */
1523 const struct cred __rcu *real_cred; /* objective and real subjective task
1524 * credentials (COW) */
1525 const struct cred __rcu *cred; /* effective (overridable) subjective task
1526 * credentials (COW) */
1527 char comm[TASK_COMM_LEN]; /* executable name excluding path
1528 - access with [gs]et_task_comm (which lock
1529 it with task_lock())
1530 - initialized normally by setup_new_exec */
1531 /* file system info */
1532 struct nameidata *nameidata;
1533 #ifdef CONFIG_SYSVIPC
1535 struct sysv_sem sysvsem;
1536 struct sysv_shm sysvshm;
1538 #ifdef CONFIG_DETECT_HUNG_TASK
1539 /* hung task detection */
1540 unsigned long last_switch_count;
1542 /* CPU-specific state of this task */
1543 struct thread_struct thread;
1544 /* filesystem information */
1545 struct fs_struct *fs;
1546 /* open file information */
1547 struct files_struct *files;
1549 struct nsproxy *nsproxy;
1550 /* signal handlers */
1551 struct signal_struct *signal;
1552 struct sighand_struct *sighand;
1554 sigset_t blocked, real_blocked;
1555 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1556 struct sigpending pending;
1558 unsigned long sas_ss_sp;
1560 int (*notifier)(void *priv);
1561 void *notifier_data;
1562 sigset_t *notifier_mask;
1563 struct callback_head *task_works;
1565 struct audit_context *audit_context;
1566 #ifdef CONFIG_AUDITSYSCALL
1568 unsigned int sessionid;
1570 struct seccomp seccomp;
1572 /* Thread group tracking */
1575 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1577 spinlock_t alloc_lock;
1579 /* Protection of the PI data structures: */
1580 raw_spinlock_t pi_lock;
1582 struct wake_q_node wake_q;
1584 #ifdef CONFIG_RT_MUTEXES
1585 /* PI waiters blocked on a rt_mutex held by this task */
1586 struct rb_root pi_waiters;
1587 struct rb_node *pi_waiters_leftmost;
1588 /* Deadlock detection and priority inheritance handling */
1589 struct rt_mutex_waiter *pi_blocked_on;
1592 #ifdef CONFIG_DEBUG_MUTEXES
1593 /* mutex deadlock detection */
1594 struct mutex_waiter *blocked_on;
1596 #ifdef CONFIG_TRACE_IRQFLAGS
1597 unsigned int irq_events;
1598 unsigned long hardirq_enable_ip;
1599 unsigned long hardirq_disable_ip;
1600 unsigned int hardirq_enable_event;
1601 unsigned int hardirq_disable_event;
1602 int hardirqs_enabled;
1603 int hardirq_context;
1604 unsigned long softirq_disable_ip;
1605 unsigned long softirq_enable_ip;
1606 unsigned int softirq_disable_event;
1607 unsigned int softirq_enable_event;
1608 int softirqs_enabled;
1609 int softirq_context;
1611 #ifdef CONFIG_LOCKDEP
1612 # define MAX_LOCK_DEPTH 48UL
1615 unsigned int lockdep_recursion;
1616 struct held_lock held_locks[MAX_LOCK_DEPTH];
1617 gfp_t lockdep_reclaim_gfp;
1620 /* journalling filesystem info */
1623 /* stacked block device info */
1624 struct bio_list *bio_list;
1627 /* stack plugging */
1628 struct blk_plug *plug;
1632 struct reclaim_state *reclaim_state;
1634 struct backing_dev_info *backing_dev_info;
1636 struct io_context *io_context;
1638 unsigned long ptrace_message;
1639 siginfo_t *last_siginfo; /* For ptrace use. */
1640 struct task_io_accounting ioac;
1641 #if defined(CONFIG_TASK_XACCT)
1642 u64 acct_rss_mem1; /* accumulated rss usage */
1643 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1644 cputime_t acct_timexpd; /* stime + utime since last update */
1646 #ifdef CONFIG_CPUSETS
1647 nodemask_t mems_allowed; /* Protected by alloc_lock */
1648 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1649 int cpuset_mem_spread_rotor;
1650 int cpuset_slab_spread_rotor;
1652 #ifdef CONFIG_CGROUPS
1653 /* Control Group info protected by css_set_lock */
1654 struct css_set __rcu *cgroups;
1655 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1656 struct list_head cg_list;
1659 struct robust_list_head __user *robust_list;
1660 #ifdef CONFIG_COMPAT
1661 struct compat_robust_list_head __user *compat_robust_list;
1663 struct list_head pi_state_list;
1664 struct futex_pi_state *pi_state_cache;
1666 #ifdef CONFIG_PERF_EVENTS
1667 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1668 struct mutex perf_event_mutex;
1669 struct list_head perf_event_list;
1671 #ifdef CONFIG_DEBUG_PREEMPT
1672 unsigned long preempt_disable_ip;
1675 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1677 short pref_node_fork;
1679 #ifdef CONFIG_NUMA_BALANCING
1681 unsigned int numa_scan_period;
1682 unsigned int numa_scan_period_max;
1683 int numa_preferred_nid;
1684 unsigned long numa_migrate_retry;
1685 u64 node_stamp; /* migration stamp */
1686 u64 last_task_numa_placement;
1687 u64 last_sum_exec_runtime;
1688 struct callback_head numa_work;
1690 struct list_head numa_entry;
1691 struct numa_group *numa_group;
1694 * numa_faults is an array split into four regions:
1695 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1696 * in this precise order.
1698 * faults_memory: Exponential decaying average of faults on a per-node
1699 * basis. Scheduling placement decisions are made based on these
1700 * counts. The values remain static for the duration of a PTE scan.
1701 * faults_cpu: Track the nodes the process was running on when a NUMA
1702 * hinting fault was incurred.
1703 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1704 * during the current scan window. When the scan completes, the counts
1705 * in faults_memory and faults_cpu decay and these values are copied.
1707 unsigned long *numa_faults;
1708 unsigned long total_numa_faults;
1711 * numa_faults_locality tracks if faults recorded during the last
1712 * scan window were remote/local or failed to migrate. The task scan
1713 * period is adapted based on the locality of the faults with different
1714 * weights depending on whether they were shared or private faults
1716 unsigned long numa_faults_locality[3];
1718 unsigned long numa_pages_migrated;
1719 #endif /* CONFIG_NUMA_BALANCING */
1721 struct rcu_head rcu;
1724 * cache last used pipe for splice
1726 struct pipe_inode_info *splice_pipe;
1728 struct page_frag task_frag;
1730 #ifdef CONFIG_TASK_DELAY_ACCT
1731 struct task_delay_info *delays;
1733 #ifdef CONFIG_FAULT_INJECTION
1737 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1738 * balance_dirty_pages() for some dirty throttling pause
1741 int nr_dirtied_pause;
1742 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1744 #ifdef CONFIG_LATENCYTOP
1745 int latency_record_count;
1746 struct latency_record latency_record[LT_SAVECOUNT];
1749 * time slack values; these are used to round up poll() and
1750 * select() etc timeout values. These are in nanoseconds.
1752 unsigned long timer_slack_ns;
1753 unsigned long default_timer_slack_ns;
1756 unsigned int kasan_depth;
1758 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1759 /* Index of current stored address in ret_stack */
1761 /* Stack of return addresses for return function tracing */
1762 struct ftrace_ret_stack *ret_stack;
1763 /* time stamp for last schedule */
1764 unsigned long long ftrace_timestamp;
1766 * Number of functions that haven't been traced
1767 * because of depth overrun.
1769 atomic_t trace_overrun;
1770 /* Pause for the tracing */
1771 atomic_t tracing_graph_pause;
1773 #ifdef CONFIG_TRACING
1774 /* state flags for use by tracers */
1775 unsigned long trace;
1776 /* bitmask and counter of trace recursion */
1777 unsigned long trace_recursion;
1778 #endif /* CONFIG_TRACING */
1780 struct memcg_oom_info {
1781 struct mem_cgroup *memcg;
1784 unsigned int may_oom:1;
1787 #ifdef CONFIG_UPROBES
1788 struct uprobe_task *utask;
1790 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1791 unsigned int sequential_io;
1792 unsigned int sequential_io_avg;
1794 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1795 unsigned long task_state_change;
1797 int pagefault_disabled;
1800 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1801 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1803 #define TNF_MIGRATED 0x01
1804 #define TNF_NO_GROUP 0x02
1805 #define TNF_SHARED 0x04
1806 #define TNF_FAULT_LOCAL 0x08
1807 #define TNF_MIGRATE_FAIL 0x10
1809 #ifdef CONFIG_NUMA_BALANCING
1810 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1811 extern pid_t task_numa_group_id(struct task_struct *p);
1812 extern void set_numabalancing_state(bool enabled);
1813 extern void task_numa_free(struct task_struct *p);
1814 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1815 int src_nid, int dst_cpu);
1817 static inline void task_numa_fault(int last_node, int node, int pages,
1821 static inline pid_t task_numa_group_id(struct task_struct *p)
1825 static inline void set_numabalancing_state(bool enabled)
1828 static inline void task_numa_free(struct task_struct *p)
1831 static inline bool should_numa_migrate_memory(struct task_struct *p,
1832 struct page *page, int src_nid, int dst_cpu)
1838 static inline struct pid *task_pid(struct task_struct *task)
1840 return task->pids[PIDTYPE_PID].pid;
1843 static inline struct pid *task_tgid(struct task_struct *task)
1845 return task->group_leader->pids[PIDTYPE_PID].pid;
1849 * Without tasklist or rcu lock it is not safe to dereference
1850 * the result of task_pgrp/task_session even if task == current,
1851 * we can race with another thread doing sys_setsid/sys_setpgid.
1853 static inline struct pid *task_pgrp(struct task_struct *task)
1855 return task->group_leader->pids[PIDTYPE_PGID].pid;
1858 static inline struct pid *task_session(struct task_struct *task)
1860 return task->group_leader->pids[PIDTYPE_SID].pid;
1863 struct pid_namespace;
1866 * the helpers to get the task's different pids as they are seen
1867 * from various namespaces
1869 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1870 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1872 * task_xid_nr_ns() : id seen from the ns specified;
1874 * set_task_vxid() : assigns a virtual id to a task;
1876 * see also pid_nr() etc in include/linux/pid.h
1878 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1879 struct pid_namespace *ns);
1881 static inline pid_t task_pid_nr(struct task_struct *tsk)
1886 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1887 struct pid_namespace *ns)
1889 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1892 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1894 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1898 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1903 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1905 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1907 return pid_vnr(task_tgid(tsk));
1911 static inline int pid_alive(const struct task_struct *p);
1912 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1918 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1924 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1926 return task_ppid_nr_ns(tsk, &init_pid_ns);
1929 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1930 struct pid_namespace *ns)
1932 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1935 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1937 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1941 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1942 struct pid_namespace *ns)
1944 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1947 static inline pid_t task_session_vnr(struct task_struct *tsk)
1949 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1952 /* obsolete, do not use */
1953 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1955 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1959 * pid_alive - check that a task structure is not stale
1960 * @p: Task structure to be checked.
1962 * Test if a process is not yet dead (at most zombie state)
1963 * If pid_alive fails, then pointers within the task structure
1964 * can be stale and must not be dereferenced.
1966 * Return: 1 if the process is alive. 0 otherwise.
1968 static inline int pid_alive(const struct task_struct *p)
1970 return p->pids[PIDTYPE_PID].pid != NULL;
1974 * is_global_init - check if a task structure is init
1975 * @tsk: Task structure to be checked.
1977 * Check if a task structure is the first user space task the kernel created.
1979 * Return: 1 if the task structure is init. 0 otherwise.
1981 static inline int is_global_init(struct task_struct *tsk)
1983 return tsk->pid == 1;
1986 extern struct pid *cad_pid;
1988 extern void free_task(struct task_struct *tsk);
1989 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1991 extern void __put_task_struct(struct task_struct *t);
1993 static inline void put_task_struct(struct task_struct *t)
1995 if (atomic_dec_and_test(&t->usage))
1996 __put_task_struct(t);
1999 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2000 extern void task_cputime(struct task_struct *t,
2001 cputime_t *utime, cputime_t *stime);
2002 extern void task_cputime_scaled(struct task_struct *t,
2003 cputime_t *utimescaled, cputime_t *stimescaled);
2004 extern cputime_t task_gtime(struct task_struct *t);
2006 static inline void task_cputime(struct task_struct *t,
2007 cputime_t *utime, cputime_t *stime)
2015 static inline void task_cputime_scaled(struct task_struct *t,
2016 cputime_t *utimescaled,
2017 cputime_t *stimescaled)
2020 *utimescaled = t->utimescaled;
2022 *stimescaled = t->stimescaled;
2025 static inline cputime_t task_gtime(struct task_struct *t)
2030 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2031 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2036 #define PF_EXITING 0x00000004 /* getting shut down */
2037 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2038 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2039 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2040 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2041 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2042 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2043 #define PF_DUMPCORE 0x00000200 /* dumped core */
2044 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2045 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2046 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2047 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2048 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2049 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2050 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2051 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2052 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2053 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2054 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2055 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2056 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2057 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2058 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2059 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2060 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2061 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2062 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2065 * Only the _current_ task can read/write to tsk->flags, but other
2066 * tasks can access tsk->flags in readonly mode for example
2067 * with tsk_used_math (like during threaded core dumping).
2068 * There is however an exception to this rule during ptrace
2069 * or during fork: the ptracer task is allowed to write to the
2070 * child->flags of its traced child (same goes for fork, the parent
2071 * can write to the child->flags), because we're guaranteed the
2072 * child is not running and in turn not changing child->flags
2073 * at the same time the parent does it.
2075 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2076 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2077 #define clear_used_math() clear_stopped_child_used_math(current)
2078 #define set_used_math() set_stopped_child_used_math(current)
2079 #define conditional_stopped_child_used_math(condition, child) \
2080 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2081 #define conditional_used_math(condition) \
2082 conditional_stopped_child_used_math(condition, current)
2083 #define copy_to_stopped_child_used_math(child) \
2084 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2085 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2086 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2087 #define used_math() tsk_used_math(current)
2089 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2090 * __GFP_FS is also cleared as it implies __GFP_IO.
2092 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2094 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2095 flags &= ~(__GFP_IO | __GFP_FS);
2099 static inline unsigned int memalloc_noio_save(void)
2101 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2102 current->flags |= PF_MEMALLOC_NOIO;
2106 static inline void memalloc_noio_restore(unsigned int flags)
2108 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2111 /* Per-process atomic flags. */
2112 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2113 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2114 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2117 #define TASK_PFA_TEST(name, func) \
2118 static inline bool task_##func(struct task_struct *p) \
2119 { return test_bit(PFA_##name, &p->atomic_flags); }
2120 #define TASK_PFA_SET(name, func) \
2121 static inline void task_set_##func(struct task_struct *p) \
2122 { set_bit(PFA_##name, &p->atomic_flags); }
2123 #define TASK_PFA_CLEAR(name, func) \
2124 static inline void task_clear_##func(struct task_struct *p) \
2125 { clear_bit(PFA_##name, &p->atomic_flags); }
2127 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2128 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2130 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2131 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2132 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2134 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2135 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2136 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2139 * task->jobctl flags
2141 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2143 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2144 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2145 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2146 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2147 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2148 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2149 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2151 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2152 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2153 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2154 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2155 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2156 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2157 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2159 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2160 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2162 extern bool task_set_jobctl_pending(struct task_struct *task,
2163 unsigned long mask);
2164 extern void task_clear_jobctl_trapping(struct task_struct *task);
2165 extern void task_clear_jobctl_pending(struct task_struct *task,
2166 unsigned long mask);
2168 static inline void rcu_copy_process(struct task_struct *p)
2170 #ifdef CONFIG_PREEMPT_RCU
2171 p->rcu_read_lock_nesting = 0;
2172 p->rcu_read_unlock_special.s = 0;
2173 p->rcu_blocked_node = NULL;
2174 INIT_LIST_HEAD(&p->rcu_node_entry);
2175 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2176 #ifdef CONFIG_TASKS_RCU
2177 p->rcu_tasks_holdout = false;
2178 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2179 p->rcu_tasks_idle_cpu = -1;
2180 #endif /* #ifdef CONFIG_TASKS_RCU */
2183 static inline void tsk_restore_flags(struct task_struct *task,
2184 unsigned long orig_flags, unsigned long flags)
2186 task->flags &= ~flags;
2187 task->flags |= orig_flags & flags;
2190 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2191 const struct cpumask *trial);
2192 extern int task_can_attach(struct task_struct *p,
2193 const struct cpumask *cs_cpus_allowed);
2195 extern void do_set_cpus_allowed(struct task_struct *p,
2196 const struct cpumask *new_mask);
2198 extern int set_cpus_allowed_ptr(struct task_struct *p,
2199 const struct cpumask *new_mask);
2201 static inline void do_set_cpus_allowed(struct task_struct *p,
2202 const struct cpumask *new_mask)
2205 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2206 const struct cpumask *new_mask)
2208 if (!cpumask_test_cpu(0, new_mask))
2214 #ifdef CONFIG_NO_HZ_COMMON
2215 void calc_load_enter_idle(void);
2216 void calc_load_exit_idle(void);
2218 static inline void calc_load_enter_idle(void) { }
2219 static inline void calc_load_exit_idle(void) { }
2220 #endif /* CONFIG_NO_HZ_COMMON */
2222 #ifndef CONFIG_CPUMASK_OFFSTACK
2223 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2225 return set_cpus_allowed_ptr(p, &new_mask);
2230 * Do not use outside of architecture code which knows its limitations.
2232 * sched_clock() has no promise of monotonicity or bounded drift between
2233 * CPUs, use (which you should not) requires disabling IRQs.
2235 * Please use one of the three interfaces below.
2237 extern unsigned long long notrace sched_clock(void);
2239 * See the comment in kernel/sched/clock.c
2241 extern u64 cpu_clock(int cpu);
2242 extern u64 local_clock(void);
2243 extern u64 running_clock(void);
2244 extern u64 sched_clock_cpu(int cpu);
2247 extern void sched_clock_init(void);
2249 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2250 static inline void sched_clock_tick(void)
2254 static inline void sched_clock_idle_sleep_event(void)
2258 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2263 * Architectures can set this to 1 if they have specified
2264 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2265 * but then during bootup it turns out that sched_clock()
2266 * is reliable after all:
2268 extern int sched_clock_stable(void);
2269 extern void set_sched_clock_stable(void);
2270 extern void clear_sched_clock_stable(void);
2272 extern void sched_clock_tick(void);
2273 extern void sched_clock_idle_sleep_event(void);
2274 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2277 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2279 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2280 * The reason for this explicit opt-in is not to have perf penalty with
2281 * slow sched_clocks.
2283 extern void enable_sched_clock_irqtime(void);
2284 extern void disable_sched_clock_irqtime(void);
2286 static inline void enable_sched_clock_irqtime(void) {}
2287 static inline void disable_sched_clock_irqtime(void) {}
2290 extern unsigned long long
2291 task_sched_runtime(struct task_struct *task);
2293 /* sched_exec is called by processes performing an exec */
2295 extern void sched_exec(void);
2297 #define sched_exec() {}
2300 extern void sched_clock_idle_sleep_event(void);
2301 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2303 #ifdef CONFIG_HOTPLUG_CPU
2304 extern void idle_task_exit(void);
2306 static inline void idle_task_exit(void) {}
2309 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2310 extern void wake_up_nohz_cpu(int cpu);
2312 static inline void wake_up_nohz_cpu(int cpu) { }
2315 #ifdef CONFIG_NO_HZ_FULL
2316 extern bool sched_can_stop_tick(void);
2317 extern u64 scheduler_tick_max_deferment(void);
2319 static inline bool sched_can_stop_tick(void) { return false; }
2322 #ifdef CONFIG_SCHED_AUTOGROUP
2323 extern void sched_autogroup_create_attach(struct task_struct *p);
2324 extern void sched_autogroup_detach(struct task_struct *p);
2325 extern void sched_autogroup_fork(struct signal_struct *sig);
2326 extern void sched_autogroup_exit(struct signal_struct *sig);
2327 #ifdef CONFIG_PROC_FS
2328 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2329 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2332 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2333 static inline void sched_autogroup_detach(struct task_struct *p) { }
2334 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2335 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2338 extern int yield_to(struct task_struct *p, bool preempt);
2339 extern void set_user_nice(struct task_struct *p, long nice);
2340 extern int task_prio(const struct task_struct *p);
2342 * task_nice - return the nice value of a given task.
2343 * @p: the task in question.
2345 * Return: The nice value [ -20 ... 0 ... 19 ].
2347 static inline int task_nice(const struct task_struct *p)
2349 return PRIO_TO_NICE((p)->static_prio);
2351 extern int can_nice(const struct task_struct *p, const int nice);
2352 extern int task_curr(const struct task_struct *p);
2353 extern int idle_cpu(int cpu);
2354 extern int sched_setscheduler(struct task_struct *, int,
2355 const struct sched_param *);
2356 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2357 const struct sched_param *);
2358 extern int sched_setattr(struct task_struct *,
2359 const struct sched_attr *);
2360 extern struct task_struct *idle_task(int cpu);
2362 * is_idle_task - is the specified task an idle task?
2363 * @p: the task in question.
2365 * Return: 1 if @p is an idle task. 0 otherwise.
2367 static inline bool is_idle_task(const struct task_struct *p)
2371 extern struct task_struct *curr_task(int cpu);
2372 extern void set_curr_task(int cpu, struct task_struct *p);
2376 union thread_union {
2377 struct thread_info thread_info;
2378 unsigned long stack[THREAD_SIZE/sizeof(long)];
2381 #ifndef __HAVE_ARCH_KSTACK_END
2382 static inline int kstack_end(void *addr)
2384 /* Reliable end of stack detection:
2385 * Some APM bios versions misalign the stack
2387 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2391 extern union thread_union init_thread_union;
2392 extern struct task_struct init_task;
2394 extern struct mm_struct init_mm;
2396 extern struct pid_namespace init_pid_ns;
2399 * find a task by one of its numerical ids
2401 * find_task_by_pid_ns():
2402 * finds a task by its pid in the specified namespace
2403 * find_task_by_vpid():
2404 * finds a task by its virtual pid
2406 * see also find_vpid() etc in include/linux/pid.h
2409 extern struct task_struct *find_task_by_vpid(pid_t nr);
2410 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2411 struct pid_namespace *ns);
2413 /* per-UID process charging. */
2414 extern struct user_struct * alloc_uid(kuid_t);
2415 static inline struct user_struct *get_uid(struct user_struct *u)
2417 atomic_inc(&u->__count);
2420 extern void free_uid(struct user_struct *);
2422 #include <asm/current.h>
2424 extern void xtime_update(unsigned long ticks);
2426 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2427 extern int wake_up_process(struct task_struct *tsk);
2428 extern void wake_up_new_task(struct task_struct *tsk);
2430 extern void kick_process(struct task_struct *tsk);
2432 static inline void kick_process(struct task_struct *tsk) { }
2434 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2435 extern void sched_dead(struct task_struct *p);
2437 extern void proc_caches_init(void);
2438 extern void flush_signals(struct task_struct *);
2439 extern void __flush_signals(struct task_struct *);
2440 extern void ignore_signals(struct task_struct *);
2441 extern void flush_signal_handlers(struct task_struct *, int force_default);
2442 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2444 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2446 unsigned long flags;
2449 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2450 ret = dequeue_signal(tsk, mask, info);
2451 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2456 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2458 extern void unblock_all_signals(void);
2459 extern void release_task(struct task_struct * p);
2460 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2461 extern int force_sigsegv(int, struct task_struct *);
2462 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2463 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2464 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2465 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2466 const struct cred *, u32);
2467 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2468 extern int kill_pid(struct pid *pid, int sig, int priv);
2469 extern int kill_proc_info(int, struct siginfo *, pid_t);
2470 extern __must_check bool do_notify_parent(struct task_struct *, int);
2471 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2472 extern void force_sig(int, struct task_struct *);
2473 extern int send_sig(int, struct task_struct *, int);
2474 extern int zap_other_threads(struct task_struct *p);
2475 extern struct sigqueue *sigqueue_alloc(void);
2476 extern void sigqueue_free(struct sigqueue *);
2477 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2478 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2480 static inline void restore_saved_sigmask(void)
2482 if (test_and_clear_restore_sigmask())
2483 __set_current_blocked(¤t->saved_sigmask);
2486 static inline sigset_t *sigmask_to_save(void)
2488 sigset_t *res = ¤t->blocked;
2489 if (unlikely(test_restore_sigmask()))
2490 res = ¤t->saved_sigmask;
2494 static inline int kill_cad_pid(int sig, int priv)
2496 return kill_pid(cad_pid, sig, priv);
2499 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2500 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2501 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2502 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2505 * True if we are on the alternate signal stack.
2507 static inline int on_sig_stack(unsigned long sp)
2509 #ifdef CONFIG_STACK_GROWSUP
2510 return sp >= current->sas_ss_sp &&
2511 sp - current->sas_ss_sp < current->sas_ss_size;
2513 return sp > current->sas_ss_sp &&
2514 sp - current->sas_ss_sp <= current->sas_ss_size;
2518 static inline int sas_ss_flags(unsigned long sp)
2520 if (!current->sas_ss_size)
2523 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2526 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2528 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2529 #ifdef CONFIG_STACK_GROWSUP
2530 return current->sas_ss_sp;
2532 return current->sas_ss_sp + current->sas_ss_size;
2538 * Routines for handling mm_structs
2540 extern struct mm_struct * mm_alloc(void);
2542 /* mmdrop drops the mm and the page tables */
2543 extern void __mmdrop(struct mm_struct *);
2544 static inline void mmdrop(struct mm_struct * mm)
2546 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2550 /* mmput gets rid of the mappings and all user-space */
2551 extern void mmput(struct mm_struct *);
2552 /* Grab a reference to a task's mm, if it is not already going away */
2553 extern struct mm_struct *get_task_mm(struct task_struct *task);
2555 * Grab a reference to a task's mm, if it is not already going away
2556 * and ptrace_may_access with the mode parameter passed to it
2559 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2560 /* Remove the current tasks stale references to the old mm_struct */
2561 extern void mm_release(struct task_struct *, struct mm_struct *);
2563 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2564 struct task_struct *);
2565 extern void flush_thread(void);
2566 extern void exit_thread(void);
2568 extern void exit_files(struct task_struct *);
2569 extern void __cleanup_sighand(struct sighand_struct *);
2571 extern void exit_itimers(struct signal_struct *);
2572 extern void flush_itimer_signals(void);
2574 extern void do_group_exit(int);
2576 extern int do_execve(struct filename *,
2577 const char __user * const __user *,
2578 const char __user * const __user *);
2579 extern int do_execveat(int, struct filename *,
2580 const char __user * const __user *,
2581 const char __user * const __user *,
2583 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2584 struct task_struct *fork_idle(int);
2585 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2587 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2588 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2590 __set_task_comm(tsk, from, false);
2592 extern char *get_task_comm(char *to, struct task_struct *tsk);
2595 void scheduler_ipi(void);
2596 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2598 static inline void scheduler_ipi(void) { }
2599 static inline unsigned long wait_task_inactive(struct task_struct *p,
2606 #define next_task(p) \
2607 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2609 #define for_each_process(p) \
2610 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2612 extern bool current_is_single_threaded(void);
2615 * Careful: do_each_thread/while_each_thread is a double loop so
2616 * 'break' will not work as expected - use goto instead.
2618 #define do_each_thread(g, t) \
2619 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2621 #define while_each_thread(g, t) \
2622 while ((t = next_thread(t)) != g)
2624 #define __for_each_thread(signal, t) \
2625 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2627 #define for_each_thread(p, t) \
2628 __for_each_thread((p)->signal, t)
2630 /* Careful: this is a double loop, 'break' won't work as expected. */
2631 #define for_each_process_thread(p, t) \
2632 for_each_process(p) for_each_thread(p, t)
2634 static inline int get_nr_threads(struct task_struct *tsk)
2636 return tsk->signal->nr_threads;
2639 static inline bool thread_group_leader(struct task_struct *p)
2641 return p->exit_signal >= 0;
2644 /* Do to the insanities of de_thread it is possible for a process
2645 * to have the pid of the thread group leader without actually being
2646 * the thread group leader. For iteration through the pids in proc
2647 * all we care about is that we have a task with the appropriate
2648 * pid, we don't actually care if we have the right task.
2650 static inline bool has_group_leader_pid(struct task_struct *p)
2652 return task_pid(p) == p->signal->leader_pid;
2656 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2658 return p1->signal == p2->signal;
2661 static inline struct task_struct *next_thread(const struct task_struct *p)
2663 return list_entry_rcu(p->thread_group.next,
2664 struct task_struct, thread_group);
2667 static inline int thread_group_empty(struct task_struct *p)
2669 return list_empty(&p->thread_group);
2672 #define delay_group_leader(p) \
2673 (thread_group_leader(p) && !thread_group_empty(p))
2676 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2677 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2678 * pins the final release of task.io_context. Also protects ->cpuset and
2679 * ->cgroup.subsys[]. And ->vfork_done.
2681 * Nests both inside and outside of read_lock(&tasklist_lock).
2682 * It must not be nested with write_lock_irq(&tasklist_lock),
2683 * neither inside nor outside.
2685 static inline void task_lock(struct task_struct *p)
2687 spin_lock(&p->alloc_lock);
2690 static inline void task_unlock(struct task_struct *p)
2692 spin_unlock(&p->alloc_lock);
2695 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2696 unsigned long *flags);
2698 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2699 unsigned long *flags)
2701 struct sighand_struct *ret;
2703 ret = __lock_task_sighand(tsk, flags);
2704 (void)__cond_lock(&tsk->sighand->siglock, ret);
2708 static inline void unlock_task_sighand(struct task_struct *tsk,
2709 unsigned long *flags)
2711 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2714 #ifdef CONFIG_CGROUPS
2715 static inline void threadgroup_change_begin(struct task_struct *tsk)
2717 down_read(&tsk->signal->group_rwsem);
2719 static inline void threadgroup_change_end(struct task_struct *tsk)
2721 up_read(&tsk->signal->group_rwsem);
2725 * threadgroup_lock - lock threadgroup
2726 * @tsk: member task of the threadgroup to lock
2728 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2729 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2730 * change ->group_leader/pid. This is useful for cases where the threadgroup
2731 * needs to stay stable across blockable operations.
2733 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2734 * synchronization. While held, no new task will be added to threadgroup
2735 * and no existing live task will have its PF_EXITING set.
2737 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2738 * sub-thread becomes a new leader.
2740 static inline void threadgroup_lock(struct task_struct *tsk)
2742 down_write(&tsk->signal->group_rwsem);
2746 * threadgroup_unlock - unlock threadgroup
2747 * @tsk: member task of the threadgroup to unlock
2749 * Reverse threadgroup_lock().
2751 static inline void threadgroup_unlock(struct task_struct *tsk)
2753 up_write(&tsk->signal->group_rwsem);
2756 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2757 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2758 static inline void threadgroup_lock(struct task_struct *tsk) {}
2759 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2762 #ifndef __HAVE_THREAD_FUNCTIONS
2764 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2765 #define task_stack_page(task) ((task)->stack)
2767 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2769 *task_thread_info(p) = *task_thread_info(org);
2770 task_thread_info(p)->task = p;
2774 * Return the address of the last usable long on the stack.
2776 * When the stack grows down, this is just above the thread
2777 * info struct. Going any lower will corrupt the threadinfo.
2779 * When the stack grows up, this is the highest address.
2780 * Beyond that position, we corrupt data on the next page.
2782 static inline unsigned long *end_of_stack(struct task_struct *p)
2784 #ifdef CONFIG_STACK_GROWSUP
2785 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2787 return (unsigned long *)(task_thread_info(p) + 1);
2792 #define task_stack_end_corrupted(task) \
2793 (*(end_of_stack(task)) != STACK_END_MAGIC)
2795 static inline int object_is_on_stack(void *obj)
2797 void *stack = task_stack_page(current);
2799 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2802 extern void thread_info_cache_init(void);
2804 #ifdef CONFIG_DEBUG_STACK_USAGE
2805 static inline unsigned long stack_not_used(struct task_struct *p)
2807 unsigned long *n = end_of_stack(p);
2809 do { /* Skip over canary */
2813 return (unsigned long)n - (unsigned long)end_of_stack(p);
2816 extern void set_task_stack_end_magic(struct task_struct *tsk);
2818 /* set thread flags in other task's structures
2819 * - see asm/thread_info.h for TIF_xxxx flags available
2821 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2823 set_ti_thread_flag(task_thread_info(tsk), flag);
2826 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2828 clear_ti_thread_flag(task_thread_info(tsk), flag);
2831 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2833 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2836 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2838 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2841 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2843 return test_ti_thread_flag(task_thread_info(tsk), flag);
2846 static inline void set_tsk_need_resched(struct task_struct *tsk)
2848 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2851 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2853 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2856 static inline int test_tsk_need_resched(struct task_struct *tsk)
2858 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2861 static inline int restart_syscall(void)
2863 set_tsk_thread_flag(current, TIF_SIGPENDING);
2864 return -ERESTARTNOINTR;
2867 static inline int signal_pending(struct task_struct *p)
2869 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2872 static inline int __fatal_signal_pending(struct task_struct *p)
2874 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2877 static inline int fatal_signal_pending(struct task_struct *p)
2879 return signal_pending(p) && __fatal_signal_pending(p);
2882 static inline int signal_pending_state(long state, struct task_struct *p)
2884 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2886 if (!signal_pending(p))
2889 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2893 * cond_resched() and cond_resched_lock(): latency reduction via
2894 * explicit rescheduling in places that are safe. The return
2895 * value indicates whether a reschedule was done in fact.
2896 * cond_resched_lock() will drop the spinlock before scheduling,
2897 * cond_resched_softirq() will enable bhs before scheduling.
2899 extern int _cond_resched(void);
2901 #define cond_resched() ({ \
2902 ___might_sleep(__FILE__, __LINE__, 0); \
2906 extern int __cond_resched_lock(spinlock_t *lock);
2908 #ifdef CONFIG_PREEMPT_COUNT
2909 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2911 #define PREEMPT_LOCK_OFFSET 0
2914 #define cond_resched_lock(lock) ({ \
2915 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2916 __cond_resched_lock(lock); \
2919 extern int __cond_resched_softirq(void);
2921 #define cond_resched_softirq() ({ \
2922 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2923 __cond_resched_softirq(); \
2926 static inline void cond_resched_rcu(void)
2928 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2936 * Does a critical section need to be broken due to another
2937 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2938 * but a general need for low latency)
2940 static inline int spin_needbreak(spinlock_t *lock)
2942 #ifdef CONFIG_PREEMPT
2943 return spin_is_contended(lock);
2950 * Idle thread specific functions to determine the need_resched
2953 #ifdef TIF_POLLING_NRFLAG
2954 static inline int tsk_is_polling(struct task_struct *p)
2956 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2959 static inline void __current_set_polling(void)
2961 set_thread_flag(TIF_POLLING_NRFLAG);
2964 static inline bool __must_check current_set_polling_and_test(void)
2966 __current_set_polling();
2969 * Polling state must be visible before we test NEED_RESCHED,
2970 * paired by resched_curr()
2972 smp_mb__after_atomic();
2974 return unlikely(tif_need_resched());
2977 static inline void __current_clr_polling(void)
2979 clear_thread_flag(TIF_POLLING_NRFLAG);
2982 static inline bool __must_check current_clr_polling_and_test(void)
2984 __current_clr_polling();
2987 * Polling state must be visible before we test NEED_RESCHED,
2988 * paired by resched_curr()
2990 smp_mb__after_atomic();
2992 return unlikely(tif_need_resched());
2996 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2997 static inline void __current_set_polling(void) { }
2998 static inline void __current_clr_polling(void) { }
3000 static inline bool __must_check current_set_polling_and_test(void)
3002 return unlikely(tif_need_resched());
3004 static inline bool __must_check current_clr_polling_and_test(void)
3006 return unlikely(tif_need_resched());
3010 static inline void current_clr_polling(void)
3012 __current_clr_polling();
3015 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3016 * Once the bit is cleared, we'll get IPIs with every new
3017 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3020 smp_mb(); /* paired with resched_curr() */
3022 preempt_fold_need_resched();
3025 static __always_inline bool need_resched(void)
3027 return unlikely(tif_need_resched());
3031 * Thread group CPU time accounting.
3033 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3034 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3037 * Reevaluate whether the task has signals pending delivery.
3038 * Wake the task if so.
3039 * This is required every time the blocked sigset_t changes.
3040 * callers must hold sighand->siglock.
3042 extern void recalc_sigpending_and_wake(struct task_struct *t);
3043 extern void recalc_sigpending(void);
3045 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3047 static inline void signal_wake_up(struct task_struct *t, bool resume)
3049 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3051 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3053 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3057 * Wrappers for p->thread_info->cpu access. No-op on UP.
3061 static inline unsigned int task_cpu(const struct task_struct *p)
3063 return task_thread_info(p)->cpu;
3066 static inline int task_node(const struct task_struct *p)
3068 return cpu_to_node(task_cpu(p));
3071 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3075 static inline unsigned int task_cpu(const struct task_struct *p)
3080 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3084 #endif /* CONFIG_SMP */
3086 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3087 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3089 #ifdef CONFIG_CGROUP_SCHED
3090 extern struct task_group root_task_group;
3091 #endif /* CONFIG_CGROUP_SCHED */
3093 extern int task_can_switch_user(struct user_struct *up,
3094 struct task_struct *tsk);
3096 #ifdef CONFIG_TASK_XACCT
3097 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3099 tsk->ioac.rchar += amt;
3102 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3104 tsk->ioac.wchar += amt;
3107 static inline void inc_syscr(struct task_struct *tsk)
3112 static inline void inc_syscw(struct task_struct *tsk)
3117 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3121 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3125 static inline void inc_syscr(struct task_struct *tsk)
3129 static inline void inc_syscw(struct task_struct *tsk)
3134 #ifndef TASK_SIZE_OF
3135 #define TASK_SIZE_OF(tsk) TASK_SIZE
3139 extern void mm_update_next_owner(struct mm_struct *mm);
3141 static inline void mm_update_next_owner(struct mm_struct *mm)
3144 #endif /* CONFIG_MEMCG */
3146 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3149 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3152 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3155 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3158 static inline unsigned long rlimit(unsigned int limit)
3160 return task_rlimit(current, limit);
3163 static inline unsigned long rlimit_max(unsigned int limit)
3165 return task_rlimit_max(current, limit);