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;
136 #define VMACACHE_BITS 2
137 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
138 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
150 extern unsigned long avenrun[]; /* Load averages */
151 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
153 #define FSHIFT 11 /* nr of bits of precision */
154 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
155 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
156 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157 #define EXP_5 2014 /* 1/exp(5sec/5min) */
158 #define EXP_15 2037 /* 1/exp(5sec/15min) */
160 #define CALC_LOAD(load,exp,n) \
162 load += n*(FIXED_1-exp); \
165 extern unsigned long total_forks;
166 extern int nr_threads;
167 DECLARE_PER_CPU(unsigned long, process_counts);
168 extern int nr_processes(void);
169 extern unsigned long nr_running(void);
170 extern bool single_task_running(void);
171 extern unsigned long nr_iowait(void);
172 extern unsigned long nr_iowait_cpu(int cpu);
173 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
175 extern void calc_global_load(unsigned long ticks);
177 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
178 extern void update_cpu_load_nohz(void);
180 static inline void update_cpu_load_nohz(void) { }
183 extern unsigned long get_parent_ip(unsigned long addr);
185 extern void dump_cpu_task(int cpu);
190 #ifdef CONFIG_SCHED_DEBUG
191 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
192 extern void proc_sched_set_task(struct task_struct *p);
194 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
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 set_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 set_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 set_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 set_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(int pinned);
349 static inline void nohz_balance_enter_idle(int cpu) { }
350 static inline void set_cpu_sd_state_idle(void) { }
351 static inline int get_nohz_timer_target(int pinned)
353 return smp_processor_id();
358 * Only dump TASK_* tasks. (0 for all tasks)
360 extern void show_state_filter(unsigned long state_filter);
362 static inline void show_state(void)
364 show_state_filter(0);
367 extern void show_regs(struct pt_regs *);
370 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
371 * task), SP is the stack pointer of the first frame that should be shown in the back
372 * trace (or NULL if the entire call-chain of the task should be shown).
374 extern void show_stack(struct task_struct *task, unsigned long *sp);
376 extern void cpu_init (void);
377 extern void trap_init(void);
378 extern void update_process_times(int user);
379 extern void scheduler_tick(void);
381 extern void sched_show_task(struct task_struct *p);
383 #ifdef CONFIG_LOCKUP_DETECTOR
384 extern void touch_softlockup_watchdog(void);
385 extern void touch_softlockup_watchdog_sync(void);
386 extern void touch_all_softlockup_watchdogs(void);
387 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
389 size_t *lenp, loff_t *ppos);
390 extern unsigned int softlockup_panic;
391 void lockup_detector_init(void);
393 static inline void touch_softlockup_watchdog(void)
396 static inline void touch_softlockup_watchdog_sync(void)
399 static inline void touch_all_softlockup_watchdogs(void)
402 static inline void lockup_detector_init(void)
407 #ifdef CONFIG_DETECT_HUNG_TASK
408 void reset_hung_task_detector(void);
410 static inline void reset_hung_task_detector(void)
415 /* Attach to any functions which should be ignored in wchan output. */
416 #define __sched __attribute__((__section__(".sched.text")))
418 /* Linker adds these: start and end of __sched functions */
419 extern char __sched_text_start[], __sched_text_end[];
421 /* Is this address in the __sched functions? */
422 extern int in_sched_functions(unsigned long addr);
424 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
425 extern signed long schedule_timeout(signed long timeout);
426 extern signed long schedule_timeout_interruptible(signed long timeout);
427 extern signed long schedule_timeout_killable(signed long timeout);
428 extern signed long schedule_timeout_uninterruptible(signed long timeout);
429 asmlinkage void schedule(void);
430 extern void schedule_preempt_disabled(void);
432 extern long io_schedule_timeout(long timeout);
434 static inline void io_schedule(void)
436 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
440 struct user_namespace;
443 extern void arch_pick_mmap_layout(struct mm_struct *mm);
445 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
446 unsigned long, unsigned long);
448 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
449 unsigned long len, unsigned long pgoff,
450 unsigned long flags);
452 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
455 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
456 #define SUID_DUMP_USER 1 /* Dump as user of process */
457 #define SUID_DUMP_ROOT 2 /* Dump as root */
461 /* for SUID_DUMP_* above */
462 #define MMF_DUMPABLE_BITS 2
463 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
465 extern void set_dumpable(struct mm_struct *mm, int value);
467 * This returns the actual value of the suid_dumpable flag. For things
468 * that are using this for checking for privilege transitions, it must
469 * test against SUID_DUMP_USER rather than treating it as a boolean
472 static inline int __get_dumpable(unsigned long mm_flags)
474 return mm_flags & MMF_DUMPABLE_MASK;
477 static inline int get_dumpable(struct mm_struct *mm)
479 return __get_dumpable(mm->flags);
482 /* coredump filter bits */
483 #define MMF_DUMP_ANON_PRIVATE 2
484 #define MMF_DUMP_ANON_SHARED 3
485 #define MMF_DUMP_MAPPED_PRIVATE 4
486 #define MMF_DUMP_MAPPED_SHARED 5
487 #define MMF_DUMP_ELF_HEADERS 6
488 #define MMF_DUMP_HUGETLB_PRIVATE 7
489 #define MMF_DUMP_HUGETLB_SHARED 8
491 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
492 #define MMF_DUMP_FILTER_BITS 7
493 #define MMF_DUMP_FILTER_MASK \
494 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
495 #define MMF_DUMP_FILTER_DEFAULT \
496 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
497 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
499 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
500 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
502 # define MMF_DUMP_MASK_DEFAULT_ELF 0
504 /* leave room for more dump flags */
505 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
506 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
507 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
509 #define MMF_HAS_UPROBES 19 /* has uprobes */
510 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
512 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
514 struct sighand_struct {
516 struct k_sigaction action[_NSIG];
518 wait_queue_head_t signalfd_wqh;
521 struct pacct_struct {
524 unsigned long ac_mem;
525 cputime_t ac_utime, ac_stime;
526 unsigned long ac_minflt, ac_majflt;
537 * struct cputime - snaphsot of system and user cputime
538 * @utime: time spent in user mode
539 * @stime: time spent in system mode
541 * Gathers a generic snapshot of user and system time.
549 * struct task_cputime - collected CPU time counts
550 * @utime: time spent in user mode, in &cputime_t units
551 * @stime: time spent in kernel mode, in &cputime_t units
552 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
554 * This is an extension of struct cputime that includes the total runtime
555 * spent by the task from the scheduler point of view.
557 * As a result, this structure groups together three kinds of CPU time
558 * that are tracked for threads and thread groups. Most things considering
559 * CPU time want to group these counts together and treat all three
560 * of them in parallel.
562 struct task_cputime {
565 unsigned long long sum_exec_runtime;
567 /* Alternate field names when used to cache expirations. */
568 #define prof_exp stime
569 #define virt_exp utime
570 #define sched_exp sum_exec_runtime
572 #define INIT_CPUTIME \
573 (struct task_cputime) { \
576 .sum_exec_runtime = 0, \
580 * This is the atomic variant of task_cputime, which can be used for
581 * storing and updating task_cputime statistics without locking.
583 struct task_cputime_atomic {
586 atomic64_t sum_exec_runtime;
589 #define INIT_CPUTIME_ATOMIC \
590 (struct task_cputime_atomic) { \
591 .utime = ATOMIC64_INIT(0), \
592 .stime = ATOMIC64_INIT(0), \
593 .sum_exec_runtime = ATOMIC64_INIT(0), \
596 #ifdef CONFIG_PREEMPT_COUNT
597 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
599 #define PREEMPT_DISABLED PREEMPT_ENABLED
603 * Disable preemption until the scheduler is running.
604 * Reset by start_kernel()->sched_init()->init_idle().
606 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
607 * before the scheduler is active -- see should_resched().
609 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
612 * struct thread_group_cputimer - thread group interval timer counts
613 * @cputime_atomic: atomic thread group interval timers.
614 * @running: non-zero when there are timers running and
615 * @cputime receives updates.
617 * This structure contains the version of task_cputime, above, that is
618 * used for thread group CPU timer calculations.
620 struct thread_group_cputimer {
621 struct task_cputime_atomic cputime_atomic;
625 #include <linux/rwsem.h>
629 * NOTE! "signal_struct" does not have its own
630 * locking, because a shared signal_struct always
631 * implies a shared sighand_struct, so locking
632 * sighand_struct is always a proper superset of
633 * the locking of signal_struct.
635 struct signal_struct {
639 struct list_head thread_head;
641 wait_queue_head_t wait_chldexit; /* for wait4() */
643 /* current thread group signal load-balancing target: */
644 struct task_struct *curr_target;
646 /* shared signal handling: */
647 struct sigpending shared_pending;
649 /* thread group exit support */
652 * - notify group_exit_task when ->count is equal to notify_count
653 * - everyone except group_exit_task is stopped during signal delivery
654 * of fatal signals, group_exit_task processes the signal.
657 struct task_struct *group_exit_task;
659 /* thread group stop support, overloads group_exit_code too */
660 int group_stop_count;
661 unsigned int flags; /* see SIGNAL_* flags below */
664 * PR_SET_CHILD_SUBREAPER marks a process, like a service
665 * manager, to re-parent orphan (double-forking) child processes
666 * to this process instead of 'init'. The service manager is
667 * able to receive SIGCHLD signals and is able to investigate
668 * the process until it calls wait(). All children of this
669 * process will inherit a flag if they should look for a
670 * child_subreaper process at exit.
672 unsigned int is_child_subreaper:1;
673 unsigned int has_child_subreaper:1;
675 /* POSIX.1b Interval Timers */
677 struct list_head posix_timers;
679 /* ITIMER_REAL timer for the process */
680 struct hrtimer real_timer;
681 struct pid *leader_pid;
682 ktime_t it_real_incr;
685 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
686 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
687 * values are defined to 0 and 1 respectively
689 struct cpu_itimer it[2];
692 * Thread group totals for process CPU timers.
693 * See thread_group_cputimer(), et al, for details.
695 struct thread_group_cputimer cputimer;
697 /* Earliest-expiration cache. */
698 struct task_cputime cputime_expires;
700 struct list_head cpu_timers[3];
702 struct pid *tty_old_pgrp;
704 /* boolean value for session group leader */
707 struct tty_struct *tty; /* NULL if no tty */
709 #ifdef CONFIG_SCHED_AUTOGROUP
710 struct autogroup *autogroup;
713 * Cumulative resource counters for dead threads in the group,
714 * and for reaped dead child processes forked by this group.
715 * Live threads maintain their own counters and add to these
716 * in __exit_signal, except for the group leader.
718 seqlock_t stats_lock;
719 cputime_t utime, stime, cutime, cstime;
722 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
723 struct cputime prev_cputime;
725 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
726 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
727 unsigned long inblock, oublock, cinblock, coublock;
728 unsigned long maxrss, cmaxrss;
729 struct task_io_accounting ioac;
732 * Cumulative ns of schedule CPU time fo dead threads in the
733 * group, not including a zombie group leader, (This only differs
734 * from jiffies_to_ns(utime + stime) if sched_clock uses something
735 * other than jiffies.)
737 unsigned long long sum_sched_runtime;
740 * We don't bother to synchronize most readers of this at all,
741 * because there is no reader checking a limit that actually needs
742 * to get both rlim_cur and rlim_max atomically, and either one
743 * alone is a single word that can safely be read normally.
744 * getrlimit/setrlimit use task_lock(current->group_leader) to
745 * protect this instead of the siglock, because they really
746 * have no need to disable irqs.
748 struct rlimit rlim[RLIM_NLIMITS];
750 #ifdef CONFIG_BSD_PROCESS_ACCT
751 struct pacct_struct pacct; /* per-process accounting information */
753 #ifdef CONFIG_TASKSTATS
754 struct taskstats *stats;
758 unsigned audit_tty_log_passwd;
759 struct tty_audit_buf *tty_audit_buf;
761 #ifdef CONFIG_CGROUPS
763 * group_rwsem prevents new tasks from entering the threadgroup and
764 * member tasks from exiting,a more specifically, setting of
765 * PF_EXITING. fork and exit paths are protected with this rwsem
766 * using threadgroup_change_begin/end(). Users which require
767 * threadgroup to remain stable should use threadgroup_[un]lock()
768 * which also takes care of exec path. Currently, cgroup is the
771 struct rw_semaphore group_rwsem;
774 oom_flags_t oom_flags;
775 short oom_score_adj; /* OOM kill score adjustment */
776 short oom_score_adj_min; /* OOM kill score adjustment min value.
777 * Only settable by CAP_SYS_RESOURCE. */
779 struct mutex cred_guard_mutex; /* guard against foreign influences on
780 * credential calculations
781 * (notably. ptrace) */
785 * Bits in flags field of signal_struct.
787 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
788 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
789 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
790 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
792 * Pending notifications to parent.
794 #define SIGNAL_CLD_STOPPED 0x00000010
795 #define SIGNAL_CLD_CONTINUED 0x00000020
796 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
798 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
800 /* If true, all threads except ->group_exit_task have pending SIGKILL */
801 static inline int signal_group_exit(const struct signal_struct *sig)
803 return (sig->flags & SIGNAL_GROUP_EXIT) ||
804 (sig->group_exit_task != NULL);
808 * Some day this will be a full-fledged user tracking system..
811 atomic_t __count; /* reference count */
812 atomic_t processes; /* How many processes does this user have? */
813 atomic_t sigpending; /* How many pending signals does this user have? */
814 #ifdef CONFIG_INOTIFY_USER
815 atomic_t inotify_watches; /* How many inotify watches does this user have? */
816 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
818 #ifdef CONFIG_FANOTIFY
819 atomic_t fanotify_listeners;
822 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
824 #ifdef CONFIG_POSIX_MQUEUE
825 /* protected by mq_lock */
826 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
828 unsigned long locked_shm; /* How many pages of mlocked shm ? */
831 struct key *uid_keyring; /* UID specific keyring */
832 struct key *session_keyring; /* UID's default session keyring */
835 /* Hash table maintenance information */
836 struct hlist_node uidhash_node;
839 #ifdef CONFIG_PERF_EVENTS
840 atomic_long_t locked_vm;
844 extern int uids_sysfs_init(void);
846 extern struct user_struct *find_user(kuid_t);
848 extern struct user_struct root_user;
849 #define INIT_USER (&root_user)
852 struct backing_dev_info;
853 struct reclaim_state;
855 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
857 /* cumulative counters */
858 unsigned long pcount; /* # of times run on this cpu */
859 unsigned long long run_delay; /* time spent waiting on a runqueue */
862 unsigned long long last_arrival,/* when we last ran on a cpu */
863 last_queued; /* when we were last queued to run */
865 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
867 #ifdef CONFIG_TASK_DELAY_ACCT
868 struct task_delay_info {
870 unsigned int flags; /* Private per-task flags */
872 /* For each stat XXX, add following, aligned appropriately
874 * struct timespec XXX_start, XXX_end;
878 * Atomicity of updates to XXX_delay, XXX_count protected by
879 * single lock above (split into XXX_lock if contention is an issue).
883 * XXX_count is incremented on every XXX operation, the delay
884 * associated with the operation is added to XXX_delay.
885 * XXX_delay contains the accumulated delay time in nanoseconds.
887 u64 blkio_start; /* Shared by blkio, swapin */
888 u64 blkio_delay; /* wait for sync block io completion */
889 u64 swapin_delay; /* wait for swapin block io completion */
890 u32 blkio_count; /* total count of the number of sync block */
891 /* io operations performed */
892 u32 swapin_count; /* total count of the number of swapin block */
893 /* io operations performed */
896 u64 freepages_delay; /* wait for memory reclaim */
897 u32 freepages_count; /* total count of memory reclaim */
899 #endif /* CONFIG_TASK_DELAY_ACCT */
901 static inline int sched_info_on(void)
903 #ifdef CONFIG_SCHEDSTATS
905 #elif defined(CONFIG_TASK_DELAY_ACCT)
906 extern int delayacct_on;
921 * Increase resolution of cpu_capacity calculations
923 #define SCHED_CAPACITY_SHIFT 10
924 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
927 * Wake-queues are lists of tasks with a pending wakeup, whose
928 * callers have already marked the task as woken internally,
929 * and can thus carry on. A common use case is being able to
930 * do the wakeups once the corresponding user lock as been
933 * We hold reference to each task in the list across the wakeup,
934 * thus guaranteeing that the memory is still valid by the time
935 * the actual wakeups are performed in wake_up_q().
937 * One per task suffices, because there's never a need for a task to be
938 * in two wake queues simultaneously; it is forbidden to abandon a task
939 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
940 * already in a wake queue, the wakeup will happen soon and the second
941 * waker can just skip it.
943 * The WAKE_Q macro declares and initializes the list head.
944 * wake_up_q() does NOT reinitialize the list; it's expected to be
945 * called near the end of a function, where the fact that the queue is
946 * not used again will be easy to see by inspection.
948 * Note that this can cause spurious wakeups. schedule() callers
949 * must ensure the call is done inside a loop, confirming that the
950 * wakeup condition has in fact occurred.
953 struct wake_q_node *next;
957 struct wake_q_node *first;
958 struct wake_q_node **lastp;
961 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
963 #define WAKE_Q(name) \
964 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
966 extern void wake_q_add(struct wake_q_head *head,
967 struct task_struct *task);
968 extern void wake_up_q(struct wake_q_head *head);
971 * sched-domains (multiprocessor balancing) declarations:
974 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
975 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
976 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
977 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
978 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
979 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
980 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
981 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
982 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
983 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
984 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
985 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
986 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
987 #define SD_NUMA 0x4000 /* cross-node balancing */
989 #ifdef CONFIG_SCHED_SMT
990 static inline int cpu_smt_flags(void)
992 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
996 #ifdef CONFIG_SCHED_MC
997 static inline int cpu_core_flags(void)
999 return SD_SHARE_PKG_RESOURCES;
1004 static inline int cpu_numa_flags(void)
1010 struct sched_domain_attr {
1011 int relax_domain_level;
1014 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1015 .relax_domain_level = -1, \
1018 extern int sched_domain_level_max;
1022 struct sched_domain {
1023 /* These fields must be setup */
1024 struct sched_domain *parent; /* top domain must be null terminated */
1025 struct sched_domain *child; /* bottom domain must be null terminated */
1026 struct sched_group *groups; /* the balancing groups of the domain */
1027 unsigned long min_interval; /* Minimum balance interval ms */
1028 unsigned long max_interval; /* Maximum balance interval ms */
1029 unsigned int busy_factor; /* less balancing by factor if busy */
1030 unsigned int imbalance_pct; /* No balance until over watermark */
1031 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1032 unsigned int busy_idx;
1033 unsigned int idle_idx;
1034 unsigned int newidle_idx;
1035 unsigned int wake_idx;
1036 unsigned int forkexec_idx;
1037 unsigned int smt_gain;
1039 int nohz_idle; /* NOHZ IDLE status */
1040 int flags; /* See SD_* */
1043 /* Runtime fields. */
1044 unsigned long last_balance; /* init to jiffies. units in jiffies */
1045 unsigned int balance_interval; /* initialise to 1. units in ms. */
1046 unsigned int nr_balance_failed; /* initialise to 0 */
1048 /* idle_balance() stats */
1049 u64 max_newidle_lb_cost;
1050 unsigned long next_decay_max_lb_cost;
1052 #ifdef CONFIG_SCHEDSTATS
1053 /* load_balance() stats */
1054 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1055 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1056 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1057 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1058 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1059 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1060 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1061 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1063 /* Active load balancing */
1064 unsigned int alb_count;
1065 unsigned int alb_failed;
1066 unsigned int alb_pushed;
1068 /* SD_BALANCE_EXEC stats */
1069 unsigned int sbe_count;
1070 unsigned int sbe_balanced;
1071 unsigned int sbe_pushed;
1073 /* SD_BALANCE_FORK stats */
1074 unsigned int sbf_count;
1075 unsigned int sbf_balanced;
1076 unsigned int sbf_pushed;
1078 /* try_to_wake_up() stats */
1079 unsigned int ttwu_wake_remote;
1080 unsigned int ttwu_move_affine;
1081 unsigned int ttwu_move_balance;
1083 #ifdef CONFIG_SCHED_DEBUG
1087 void *private; /* used during construction */
1088 struct rcu_head rcu; /* used during destruction */
1091 unsigned int span_weight;
1093 * Span of all CPUs in this domain.
1095 * NOTE: this field is variable length. (Allocated dynamically
1096 * by attaching extra space to the end of the structure,
1097 * depending on how many CPUs the kernel has booted up with)
1099 unsigned long span[0];
1102 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1104 return to_cpumask(sd->span);
1107 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1108 struct sched_domain_attr *dattr_new);
1110 /* Allocate an array of sched domains, for partition_sched_domains(). */
1111 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1112 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1114 bool cpus_share_cache(int this_cpu, int that_cpu);
1116 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1117 typedef int (*sched_domain_flags_f)(void);
1119 #define SDTL_OVERLAP 0x01
1122 struct sched_domain **__percpu sd;
1123 struct sched_group **__percpu sg;
1124 struct sched_group_capacity **__percpu sgc;
1127 struct sched_domain_topology_level {
1128 sched_domain_mask_f mask;
1129 sched_domain_flags_f sd_flags;
1132 struct sd_data data;
1133 #ifdef CONFIG_SCHED_DEBUG
1138 extern struct sched_domain_topology_level *sched_domain_topology;
1140 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1141 extern void wake_up_if_idle(int cpu);
1143 #ifdef CONFIG_SCHED_DEBUG
1144 # define SD_INIT_NAME(type) .name = #type
1146 # define SD_INIT_NAME(type)
1149 #else /* CONFIG_SMP */
1151 struct sched_domain_attr;
1154 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1155 struct sched_domain_attr *dattr_new)
1159 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1164 #endif /* !CONFIG_SMP */
1167 struct io_context; /* See blkdev.h */
1170 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1171 extern void prefetch_stack(struct task_struct *t);
1173 static inline void prefetch_stack(struct task_struct *t) { }
1176 struct audit_context; /* See audit.c */
1178 struct pipe_inode_info;
1179 struct uts_namespace;
1181 struct load_weight {
1182 unsigned long weight;
1187 u64 last_runnable_update;
1190 * utilization_avg_contrib describes the amount of time that a
1191 * sched_entity is running on a CPU. It is based on running_avg_sum
1192 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1193 * load_avg_contrib described the amount of time that a sched_entity
1194 * is runnable on a rq. It is based on both runnable_avg_sum and the
1195 * weight of the task.
1197 unsigned long load_avg_contrib, utilization_avg_contrib;
1199 * These sums represent an infinite geometric series and so are bound
1200 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1201 * choices of y < 1-2^(-32)*1024.
1202 * running_avg_sum reflects the time that the sched_entity is
1203 * effectively running on the CPU.
1204 * runnable_avg_sum represents the amount of time a sched_entity is on
1205 * a runqueue which includes the running time that is monitored by
1208 u32 runnable_avg_sum, avg_period, running_avg_sum;
1211 #ifdef CONFIG_SCHEDSTATS
1212 struct sched_statistics {
1222 s64 sum_sleep_runtime;
1229 u64 nr_migrations_cold;
1230 u64 nr_failed_migrations_affine;
1231 u64 nr_failed_migrations_running;
1232 u64 nr_failed_migrations_hot;
1233 u64 nr_forced_migrations;
1236 u64 nr_wakeups_sync;
1237 u64 nr_wakeups_migrate;
1238 u64 nr_wakeups_local;
1239 u64 nr_wakeups_remote;
1240 u64 nr_wakeups_affine;
1241 u64 nr_wakeups_affine_attempts;
1242 u64 nr_wakeups_passive;
1243 u64 nr_wakeups_idle;
1247 struct sched_entity {
1248 struct load_weight load; /* for load-balancing */
1249 struct rb_node run_node;
1250 struct list_head group_node;
1254 u64 sum_exec_runtime;
1256 u64 prev_sum_exec_runtime;
1260 #ifdef CONFIG_SCHEDSTATS
1261 struct sched_statistics statistics;
1264 #ifdef CONFIG_FAIR_GROUP_SCHED
1266 struct sched_entity *parent;
1267 /* rq on which this entity is (to be) queued: */
1268 struct cfs_rq *cfs_rq;
1269 /* rq "owned" by this entity/group: */
1270 struct cfs_rq *my_q;
1274 /* Per-entity load-tracking */
1275 struct sched_avg avg;
1279 struct sched_rt_entity {
1280 struct list_head run_list;
1281 unsigned long timeout;
1282 unsigned long watchdog_stamp;
1283 unsigned int time_slice;
1285 struct sched_rt_entity *back;
1286 #ifdef CONFIG_RT_GROUP_SCHED
1287 struct sched_rt_entity *parent;
1288 /* rq on which this entity is (to be) queued: */
1289 struct rt_rq *rt_rq;
1290 /* rq "owned" by this entity/group: */
1295 struct sched_dl_entity {
1296 struct rb_node rb_node;
1299 * Original scheduling parameters. Copied here from sched_attr
1300 * during sched_setattr(), they will remain the same until
1301 * the next sched_setattr().
1303 u64 dl_runtime; /* maximum runtime for each instance */
1304 u64 dl_deadline; /* relative deadline of each instance */
1305 u64 dl_period; /* separation of two instances (period) */
1306 u64 dl_bw; /* dl_runtime / dl_deadline */
1309 * Actual scheduling parameters. Initialized with the values above,
1310 * they are continously updated during task execution. Note that
1311 * the remaining runtime could be < 0 in case we are in overrun.
1313 s64 runtime; /* remaining runtime for this instance */
1314 u64 deadline; /* absolute deadline for this instance */
1315 unsigned int flags; /* specifying the scheduler behaviour */
1320 * @dl_throttled tells if we exhausted the runtime. If so, the
1321 * task has to wait for a replenishment to be performed at the
1322 * next firing of dl_timer.
1324 * @dl_new tells if a new instance arrived. If so we must
1325 * start executing it with full runtime and reset its absolute
1328 * @dl_boosted tells if we are boosted due to DI. If so we are
1329 * outside bandwidth enforcement mechanism (but only until we
1330 * exit the critical section);
1332 * @dl_yielded tells if task gave up the cpu before consuming
1333 * all its available runtime during the last job.
1335 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1338 * Bandwidth enforcement timer. Each -deadline task has its
1339 * own bandwidth to be enforced, thus we need one timer per task.
1341 struct hrtimer dl_timer;
1353 enum perf_event_task_context {
1354 perf_invalid_context = -1,
1355 perf_hw_context = 0,
1357 perf_nr_task_contexts,
1360 struct task_struct {
1361 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1364 unsigned int flags; /* per process flags, defined below */
1365 unsigned int ptrace;
1368 struct llist_node wake_entry;
1370 struct task_struct *last_wakee;
1371 unsigned long wakee_flips;
1372 unsigned long wakee_flip_decay_ts;
1378 int prio, static_prio, normal_prio;
1379 unsigned int rt_priority;
1380 const struct sched_class *sched_class;
1381 struct sched_entity se;
1382 struct sched_rt_entity rt;
1383 #ifdef CONFIG_CGROUP_SCHED
1384 struct task_group *sched_task_group;
1386 struct sched_dl_entity dl;
1388 #ifdef CONFIG_PREEMPT_NOTIFIERS
1389 /* list of struct preempt_notifier: */
1390 struct hlist_head preempt_notifiers;
1393 #ifdef CONFIG_BLK_DEV_IO_TRACE
1394 unsigned int btrace_seq;
1397 unsigned int policy;
1398 int nr_cpus_allowed;
1399 cpumask_t cpus_allowed;
1401 #ifdef CONFIG_PREEMPT_RCU
1402 int rcu_read_lock_nesting;
1403 union rcu_special rcu_read_unlock_special;
1404 struct list_head rcu_node_entry;
1405 struct rcu_node *rcu_blocked_node;
1406 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1407 #ifdef CONFIG_TASKS_RCU
1408 unsigned long rcu_tasks_nvcsw;
1409 bool rcu_tasks_holdout;
1410 struct list_head rcu_tasks_holdout_list;
1411 int rcu_tasks_idle_cpu;
1412 #endif /* #ifdef CONFIG_TASKS_RCU */
1414 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1415 struct sched_info sched_info;
1418 struct list_head tasks;
1420 struct plist_node pushable_tasks;
1421 struct rb_node pushable_dl_tasks;
1424 struct mm_struct *mm, *active_mm;
1425 #ifdef CONFIG_COMPAT_BRK
1426 unsigned brk_randomized:1;
1428 /* per-thread vma caching */
1429 u32 vmacache_seqnum;
1430 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1431 #if defined(SPLIT_RSS_COUNTING)
1432 struct task_rss_stat rss_stat;
1436 int exit_code, exit_signal;
1437 int pdeath_signal; /* The signal sent when the parent dies */
1438 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1440 /* Used for emulating ABI behavior of previous Linux versions */
1441 unsigned int personality;
1443 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1445 unsigned in_iowait:1;
1447 /* Revert to default priority/policy when forking */
1448 unsigned sched_reset_on_fork:1;
1449 unsigned sched_contributes_to_load:1;
1451 #ifdef CONFIG_MEMCG_KMEM
1452 unsigned memcg_kmem_skip_account:1;
1455 unsigned long atomic_flags; /* Flags needing atomic access. */
1457 struct restart_block restart_block;
1462 #ifdef CONFIG_CC_STACKPROTECTOR
1463 /* Canary value for the -fstack-protector gcc feature */
1464 unsigned long stack_canary;
1467 * pointers to (original) parent process, youngest child, younger sibling,
1468 * older sibling, respectively. (p->father can be replaced with
1469 * p->real_parent->pid)
1471 struct task_struct __rcu *real_parent; /* real parent process */
1472 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1474 * children/sibling forms the list of my natural children
1476 struct list_head children; /* list of my children */
1477 struct list_head sibling; /* linkage in my parent's children list */
1478 struct task_struct *group_leader; /* threadgroup leader */
1481 * ptraced is the list of tasks this task is using ptrace on.
1482 * This includes both natural children and PTRACE_ATTACH targets.
1483 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1485 struct list_head ptraced;
1486 struct list_head ptrace_entry;
1488 /* PID/PID hash table linkage. */
1489 struct pid_link pids[PIDTYPE_MAX];
1490 struct list_head thread_group;
1491 struct list_head thread_node;
1493 struct completion *vfork_done; /* for vfork() */
1494 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1495 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1497 cputime_t utime, stime, utimescaled, stimescaled;
1499 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1500 struct cputime prev_cputime;
1502 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1503 seqlock_t vtime_seqlock;
1504 unsigned long long vtime_snap;
1509 } vtime_snap_whence;
1511 unsigned long nvcsw, nivcsw; /* context switch counts */
1512 u64 start_time; /* monotonic time in nsec */
1513 u64 real_start_time; /* boot based time in nsec */
1514 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1515 unsigned long min_flt, maj_flt;
1517 struct task_cputime cputime_expires;
1518 struct list_head cpu_timers[3];
1520 /* process credentials */
1521 const struct cred __rcu *real_cred; /* objective and real subjective task
1522 * credentials (COW) */
1523 const struct cred __rcu *cred; /* effective (overridable) subjective task
1524 * credentials (COW) */
1525 char comm[TASK_COMM_LEN]; /* executable name excluding path
1526 - access with [gs]et_task_comm (which lock
1527 it with task_lock())
1528 - initialized normally by setup_new_exec */
1529 /* file system info */
1530 int link_count, total_link_count;
1531 #ifdef CONFIG_SYSVIPC
1533 struct sysv_sem sysvsem;
1534 struct sysv_shm sysvshm;
1536 #ifdef CONFIG_DETECT_HUNG_TASK
1537 /* hung task detection */
1538 unsigned long last_switch_count;
1540 /* CPU-specific state of this task */
1541 struct thread_struct thread;
1542 /* filesystem information */
1543 struct fs_struct *fs;
1544 /* open file information */
1545 struct files_struct *files;
1547 struct nsproxy *nsproxy;
1548 /* signal handlers */
1549 struct signal_struct *signal;
1550 struct sighand_struct *sighand;
1552 sigset_t blocked, real_blocked;
1553 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1554 struct sigpending pending;
1556 unsigned long sas_ss_sp;
1558 int (*notifier)(void *priv);
1559 void *notifier_data;
1560 sigset_t *notifier_mask;
1561 struct callback_head *task_works;
1563 struct audit_context *audit_context;
1564 #ifdef CONFIG_AUDITSYSCALL
1566 unsigned int sessionid;
1568 struct seccomp seccomp;
1570 /* Thread group tracking */
1573 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1575 spinlock_t alloc_lock;
1577 /* Protection of the PI data structures: */
1578 raw_spinlock_t pi_lock;
1580 struct wake_q_node wake_q;
1582 #ifdef CONFIG_RT_MUTEXES
1583 /* PI waiters blocked on a rt_mutex held by this task */
1584 struct rb_root pi_waiters;
1585 struct rb_node *pi_waiters_leftmost;
1586 /* Deadlock detection and priority inheritance handling */
1587 struct rt_mutex_waiter *pi_blocked_on;
1590 #ifdef CONFIG_DEBUG_MUTEXES
1591 /* mutex deadlock detection */
1592 struct mutex_waiter *blocked_on;
1594 #ifdef CONFIG_TRACE_IRQFLAGS
1595 unsigned int irq_events;
1596 unsigned long hardirq_enable_ip;
1597 unsigned long hardirq_disable_ip;
1598 unsigned int hardirq_enable_event;
1599 unsigned int hardirq_disable_event;
1600 int hardirqs_enabled;
1601 int hardirq_context;
1602 unsigned long softirq_disable_ip;
1603 unsigned long softirq_enable_ip;
1604 unsigned int softirq_disable_event;
1605 unsigned int softirq_enable_event;
1606 int softirqs_enabled;
1607 int softirq_context;
1609 #ifdef CONFIG_LOCKDEP
1610 # define MAX_LOCK_DEPTH 48UL
1613 unsigned int lockdep_recursion;
1614 struct held_lock held_locks[MAX_LOCK_DEPTH];
1615 gfp_t lockdep_reclaim_gfp;
1618 /* journalling filesystem info */
1621 /* stacked block device info */
1622 struct bio_list *bio_list;
1625 /* stack plugging */
1626 struct blk_plug *plug;
1630 struct reclaim_state *reclaim_state;
1632 struct backing_dev_info *backing_dev_info;
1634 struct io_context *io_context;
1636 unsigned long ptrace_message;
1637 siginfo_t *last_siginfo; /* For ptrace use. */
1638 struct task_io_accounting ioac;
1639 #if defined(CONFIG_TASK_XACCT)
1640 u64 acct_rss_mem1; /* accumulated rss usage */
1641 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1642 cputime_t acct_timexpd; /* stime + utime since last update */
1644 #ifdef CONFIG_CPUSETS
1645 nodemask_t mems_allowed; /* Protected by alloc_lock */
1646 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1647 int cpuset_mem_spread_rotor;
1648 int cpuset_slab_spread_rotor;
1650 #ifdef CONFIG_CGROUPS
1651 /* Control Group info protected by css_set_lock */
1652 struct css_set __rcu *cgroups;
1653 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1654 struct list_head cg_list;
1657 struct robust_list_head __user *robust_list;
1658 #ifdef CONFIG_COMPAT
1659 struct compat_robust_list_head __user *compat_robust_list;
1661 struct list_head pi_state_list;
1662 struct futex_pi_state *pi_state_cache;
1664 #ifdef CONFIG_PERF_EVENTS
1665 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1666 struct mutex perf_event_mutex;
1667 struct list_head perf_event_list;
1669 #ifdef CONFIG_DEBUG_PREEMPT
1670 unsigned long preempt_disable_ip;
1673 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1675 short pref_node_fork;
1677 #ifdef CONFIG_NUMA_BALANCING
1679 unsigned int numa_scan_period;
1680 unsigned int numa_scan_period_max;
1681 int numa_preferred_nid;
1682 unsigned long numa_migrate_retry;
1683 u64 node_stamp; /* migration stamp */
1684 u64 last_task_numa_placement;
1685 u64 last_sum_exec_runtime;
1686 struct callback_head numa_work;
1688 struct list_head numa_entry;
1689 struct numa_group *numa_group;
1692 * numa_faults is an array split into four regions:
1693 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1694 * in this precise order.
1696 * faults_memory: Exponential decaying average of faults on a per-node
1697 * basis. Scheduling placement decisions are made based on these
1698 * counts. The values remain static for the duration of a PTE scan.
1699 * faults_cpu: Track the nodes the process was running on when a NUMA
1700 * hinting fault was incurred.
1701 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1702 * during the current scan window. When the scan completes, the counts
1703 * in faults_memory and faults_cpu decay and these values are copied.
1705 unsigned long *numa_faults;
1706 unsigned long total_numa_faults;
1709 * numa_faults_locality tracks if faults recorded during the last
1710 * scan window were remote/local or failed to migrate. The task scan
1711 * period is adapted based on the locality of the faults with different
1712 * weights depending on whether they were shared or private faults
1714 unsigned long numa_faults_locality[3];
1716 unsigned long numa_pages_migrated;
1717 #endif /* CONFIG_NUMA_BALANCING */
1719 struct rcu_head rcu;
1722 * cache last used pipe for splice
1724 struct pipe_inode_info *splice_pipe;
1726 struct page_frag task_frag;
1728 #ifdef CONFIG_TASK_DELAY_ACCT
1729 struct task_delay_info *delays;
1731 #ifdef CONFIG_FAULT_INJECTION
1735 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1736 * balance_dirty_pages() for some dirty throttling pause
1739 int nr_dirtied_pause;
1740 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1742 #ifdef CONFIG_LATENCYTOP
1743 int latency_record_count;
1744 struct latency_record latency_record[LT_SAVECOUNT];
1747 * time slack values; these are used to round up poll() and
1748 * select() etc timeout values. These are in nanoseconds.
1750 unsigned long timer_slack_ns;
1751 unsigned long default_timer_slack_ns;
1754 unsigned int kasan_depth;
1756 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1757 /* Index of current stored address in ret_stack */
1759 /* Stack of return addresses for return function tracing */
1760 struct ftrace_ret_stack *ret_stack;
1761 /* time stamp for last schedule */
1762 unsigned long long ftrace_timestamp;
1764 * Number of functions that haven't been traced
1765 * because of depth overrun.
1767 atomic_t trace_overrun;
1768 /* Pause for the tracing */
1769 atomic_t tracing_graph_pause;
1771 #ifdef CONFIG_TRACING
1772 /* state flags for use by tracers */
1773 unsigned long trace;
1774 /* bitmask and counter of trace recursion */
1775 unsigned long trace_recursion;
1776 #endif /* CONFIG_TRACING */
1778 struct memcg_oom_info {
1779 struct mem_cgroup *memcg;
1782 unsigned int may_oom:1;
1785 #ifdef CONFIG_UPROBES
1786 struct uprobe_task *utask;
1788 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1789 unsigned int sequential_io;
1790 unsigned int sequential_io_avg;
1792 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1793 unsigned long task_state_change;
1795 int pagefault_disabled;
1798 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1799 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1801 #define TNF_MIGRATED 0x01
1802 #define TNF_NO_GROUP 0x02
1803 #define TNF_SHARED 0x04
1804 #define TNF_FAULT_LOCAL 0x08
1805 #define TNF_MIGRATE_FAIL 0x10
1807 #ifdef CONFIG_NUMA_BALANCING
1808 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1809 extern pid_t task_numa_group_id(struct task_struct *p);
1810 extern void set_numabalancing_state(bool enabled);
1811 extern void task_numa_free(struct task_struct *p);
1812 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1813 int src_nid, int dst_cpu);
1815 static inline void task_numa_fault(int last_node, int node, int pages,
1819 static inline pid_t task_numa_group_id(struct task_struct *p)
1823 static inline void set_numabalancing_state(bool enabled)
1826 static inline void task_numa_free(struct task_struct *p)
1829 static inline bool should_numa_migrate_memory(struct task_struct *p,
1830 struct page *page, int src_nid, int dst_cpu)
1836 static inline struct pid *task_pid(struct task_struct *task)
1838 return task->pids[PIDTYPE_PID].pid;
1841 static inline struct pid *task_tgid(struct task_struct *task)
1843 return task->group_leader->pids[PIDTYPE_PID].pid;
1847 * Without tasklist or rcu lock it is not safe to dereference
1848 * the result of task_pgrp/task_session even if task == current,
1849 * we can race with another thread doing sys_setsid/sys_setpgid.
1851 static inline struct pid *task_pgrp(struct task_struct *task)
1853 return task->group_leader->pids[PIDTYPE_PGID].pid;
1856 static inline struct pid *task_session(struct task_struct *task)
1858 return task->group_leader->pids[PIDTYPE_SID].pid;
1861 struct pid_namespace;
1864 * the helpers to get the task's different pids as they are seen
1865 * from various namespaces
1867 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1868 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1870 * task_xid_nr_ns() : id seen from the ns specified;
1872 * set_task_vxid() : assigns a virtual id to a task;
1874 * see also pid_nr() etc in include/linux/pid.h
1876 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1877 struct pid_namespace *ns);
1879 static inline pid_t task_pid_nr(struct task_struct *tsk)
1884 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1885 struct pid_namespace *ns)
1887 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1890 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1892 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1896 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1901 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1903 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1905 return pid_vnr(task_tgid(tsk));
1909 static inline int pid_alive(const struct task_struct *p);
1910 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1916 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1922 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1924 return task_ppid_nr_ns(tsk, &init_pid_ns);
1927 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1928 struct pid_namespace *ns)
1930 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1933 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1935 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1939 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1940 struct pid_namespace *ns)
1942 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1945 static inline pid_t task_session_vnr(struct task_struct *tsk)
1947 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1950 /* obsolete, do not use */
1951 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1953 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1957 * pid_alive - check that a task structure is not stale
1958 * @p: Task structure to be checked.
1960 * Test if a process is not yet dead (at most zombie state)
1961 * If pid_alive fails, then pointers within the task structure
1962 * can be stale and must not be dereferenced.
1964 * Return: 1 if the process is alive. 0 otherwise.
1966 static inline int pid_alive(const struct task_struct *p)
1968 return p->pids[PIDTYPE_PID].pid != NULL;
1972 * is_global_init - check if a task structure is init
1973 * @tsk: Task structure to be checked.
1975 * Check if a task structure is the first user space task the kernel created.
1977 * Return: 1 if the task structure is init. 0 otherwise.
1979 static inline int is_global_init(struct task_struct *tsk)
1981 return tsk->pid == 1;
1984 extern struct pid *cad_pid;
1986 extern void free_task(struct task_struct *tsk);
1987 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1989 extern void __put_task_struct(struct task_struct *t);
1991 static inline void put_task_struct(struct task_struct *t)
1993 if (atomic_dec_and_test(&t->usage))
1994 __put_task_struct(t);
1997 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1998 extern void task_cputime(struct task_struct *t,
1999 cputime_t *utime, cputime_t *stime);
2000 extern void task_cputime_scaled(struct task_struct *t,
2001 cputime_t *utimescaled, cputime_t *stimescaled);
2002 extern cputime_t task_gtime(struct task_struct *t);
2004 static inline void task_cputime(struct task_struct *t,
2005 cputime_t *utime, cputime_t *stime)
2013 static inline void task_cputime_scaled(struct task_struct *t,
2014 cputime_t *utimescaled,
2015 cputime_t *stimescaled)
2018 *utimescaled = t->utimescaled;
2020 *stimescaled = t->stimescaled;
2023 static inline cputime_t task_gtime(struct task_struct *t)
2028 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2029 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2034 #define PF_EXITING 0x00000004 /* getting shut down */
2035 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2036 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2037 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2038 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2039 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2040 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2041 #define PF_DUMPCORE 0x00000200 /* dumped core */
2042 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2043 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2044 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2045 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2046 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2047 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2048 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2049 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2050 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2051 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2052 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2053 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2054 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2055 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2056 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2057 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2058 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2059 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2060 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2063 * Only the _current_ task can read/write to tsk->flags, but other
2064 * tasks can access tsk->flags in readonly mode for example
2065 * with tsk_used_math (like during threaded core dumping).
2066 * There is however an exception to this rule during ptrace
2067 * or during fork: the ptracer task is allowed to write to the
2068 * child->flags of its traced child (same goes for fork, the parent
2069 * can write to the child->flags), because we're guaranteed the
2070 * child is not running and in turn not changing child->flags
2071 * at the same time the parent does it.
2073 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2074 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2075 #define clear_used_math() clear_stopped_child_used_math(current)
2076 #define set_used_math() set_stopped_child_used_math(current)
2077 #define conditional_stopped_child_used_math(condition, child) \
2078 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2079 #define conditional_used_math(condition) \
2080 conditional_stopped_child_used_math(condition, current)
2081 #define copy_to_stopped_child_used_math(child) \
2082 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2083 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2084 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2085 #define used_math() tsk_used_math(current)
2087 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2088 * __GFP_FS is also cleared as it implies __GFP_IO.
2090 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2092 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2093 flags &= ~(__GFP_IO | __GFP_FS);
2097 static inline unsigned int memalloc_noio_save(void)
2099 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2100 current->flags |= PF_MEMALLOC_NOIO;
2104 static inline void memalloc_noio_restore(unsigned int flags)
2106 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2109 /* Per-process atomic flags. */
2110 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2111 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2112 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2115 #define TASK_PFA_TEST(name, func) \
2116 static inline bool task_##func(struct task_struct *p) \
2117 { return test_bit(PFA_##name, &p->atomic_flags); }
2118 #define TASK_PFA_SET(name, func) \
2119 static inline void task_set_##func(struct task_struct *p) \
2120 { set_bit(PFA_##name, &p->atomic_flags); }
2121 #define TASK_PFA_CLEAR(name, func) \
2122 static inline void task_clear_##func(struct task_struct *p) \
2123 { clear_bit(PFA_##name, &p->atomic_flags); }
2125 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2126 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2128 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2129 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2130 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2132 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2133 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2134 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2137 * task->jobctl flags
2139 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2141 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2142 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2143 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2144 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2145 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2146 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2147 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2149 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2150 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2151 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2152 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2153 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2154 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2155 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2157 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2158 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2160 extern bool task_set_jobctl_pending(struct task_struct *task,
2161 unsigned long mask);
2162 extern void task_clear_jobctl_trapping(struct task_struct *task);
2163 extern void task_clear_jobctl_pending(struct task_struct *task,
2164 unsigned long mask);
2166 static inline void rcu_copy_process(struct task_struct *p)
2168 #ifdef CONFIG_PREEMPT_RCU
2169 p->rcu_read_lock_nesting = 0;
2170 p->rcu_read_unlock_special.s = 0;
2171 p->rcu_blocked_node = NULL;
2172 INIT_LIST_HEAD(&p->rcu_node_entry);
2173 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2174 #ifdef CONFIG_TASKS_RCU
2175 p->rcu_tasks_holdout = false;
2176 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2177 p->rcu_tasks_idle_cpu = -1;
2178 #endif /* #ifdef CONFIG_TASKS_RCU */
2181 static inline void tsk_restore_flags(struct task_struct *task,
2182 unsigned long orig_flags, unsigned long flags)
2184 task->flags &= ~flags;
2185 task->flags |= orig_flags & flags;
2188 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2189 const struct cpumask *trial);
2190 extern int task_can_attach(struct task_struct *p,
2191 const struct cpumask *cs_cpus_allowed);
2193 extern void do_set_cpus_allowed(struct task_struct *p,
2194 const struct cpumask *new_mask);
2196 extern int set_cpus_allowed_ptr(struct task_struct *p,
2197 const struct cpumask *new_mask);
2199 static inline void do_set_cpus_allowed(struct task_struct *p,
2200 const struct cpumask *new_mask)
2203 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2204 const struct cpumask *new_mask)
2206 if (!cpumask_test_cpu(0, new_mask))
2212 #ifdef CONFIG_NO_HZ_COMMON
2213 void calc_load_enter_idle(void);
2214 void calc_load_exit_idle(void);
2216 static inline void calc_load_enter_idle(void) { }
2217 static inline void calc_load_exit_idle(void) { }
2218 #endif /* CONFIG_NO_HZ_COMMON */
2220 #ifndef CONFIG_CPUMASK_OFFSTACK
2221 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2223 return set_cpus_allowed_ptr(p, &new_mask);
2228 * Do not use outside of architecture code which knows its limitations.
2230 * sched_clock() has no promise of monotonicity or bounded drift between
2231 * CPUs, use (which you should not) requires disabling IRQs.
2233 * Please use one of the three interfaces below.
2235 extern unsigned long long notrace sched_clock(void);
2237 * See the comment in kernel/sched/clock.c
2239 extern u64 cpu_clock(int cpu);
2240 extern u64 local_clock(void);
2241 extern u64 running_clock(void);
2242 extern u64 sched_clock_cpu(int cpu);
2245 extern void sched_clock_init(void);
2247 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2248 static inline void sched_clock_tick(void)
2252 static inline void sched_clock_idle_sleep_event(void)
2256 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2261 * Architectures can set this to 1 if they have specified
2262 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2263 * but then during bootup it turns out that sched_clock()
2264 * is reliable after all:
2266 extern int sched_clock_stable(void);
2267 extern void set_sched_clock_stable(void);
2268 extern void clear_sched_clock_stable(void);
2270 extern void sched_clock_tick(void);
2271 extern void sched_clock_idle_sleep_event(void);
2272 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2275 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2277 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2278 * The reason for this explicit opt-in is not to have perf penalty with
2279 * slow sched_clocks.
2281 extern void enable_sched_clock_irqtime(void);
2282 extern void disable_sched_clock_irqtime(void);
2284 static inline void enable_sched_clock_irqtime(void) {}
2285 static inline void disable_sched_clock_irqtime(void) {}
2288 extern unsigned long long
2289 task_sched_runtime(struct task_struct *task);
2291 /* sched_exec is called by processes performing an exec */
2293 extern void sched_exec(void);
2295 #define sched_exec() {}
2298 extern void sched_clock_idle_sleep_event(void);
2299 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2301 #ifdef CONFIG_HOTPLUG_CPU
2302 extern void idle_task_exit(void);
2304 static inline void idle_task_exit(void) {}
2307 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2308 extern void wake_up_nohz_cpu(int cpu);
2310 static inline void wake_up_nohz_cpu(int cpu) { }
2313 #ifdef CONFIG_NO_HZ_FULL
2314 extern bool sched_can_stop_tick(void);
2315 extern u64 scheduler_tick_max_deferment(void);
2317 static inline bool sched_can_stop_tick(void) { return false; }
2320 #ifdef CONFIG_SCHED_AUTOGROUP
2321 extern void sched_autogroup_create_attach(struct task_struct *p);
2322 extern void sched_autogroup_detach(struct task_struct *p);
2323 extern void sched_autogroup_fork(struct signal_struct *sig);
2324 extern void sched_autogroup_exit(struct signal_struct *sig);
2325 #ifdef CONFIG_PROC_FS
2326 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2327 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2330 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2331 static inline void sched_autogroup_detach(struct task_struct *p) { }
2332 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2333 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2336 extern int yield_to(struct task_struct *p, bool preempt);
2337 extern void set_user_nice(struct task_struct *p, long nice);
2338 extern int task_prio(const struct task_struct *p);
2340 * task_nice - return the nice value of a given task.
2341 * @p: the task in question.
2343 * Return: The nice value [ -20 ... 0 ... 19 ].
2345 static inline int task_nice(const struct task_struct *p)
2347 return PRIO_TO_NICE((p)->static_prio);
2349 extern int can_nice(const struct task_struct *p, const int nice);
2350 extern int task_curr(const struct task_struct *p);
2351 extern int idle_cpu(int cpu);
2352 extern int sched_setscheduler(struct task_struct *, int,
2353 const struct sched_param *);
2354 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2355 const struct sched_param *);
2356 extern int sched_setattr(struct task_struct *,
2357 const struct sched_attr *);
2358 extern struct task_struct *idle_task(int cpu);
2360 * is_idle_task - is the specified task an idle task?
2361 * @p: the task in question.
2363 * Return: 1 if @p is an idle task. 0 otherwise.
2365 static inline bool is_idle_task(const struct task_struct *p)
2369 extern struct task_struct *curr_task(int cpu);
2370 extern void set_curr_task(int cpu, struct task_struct *p);
2374 union thread_union {
2375 struct thread_info thread_info;
2376 unsigned long stack[THREAD_SIZE/sizeof(long)];
2379 #ifndef __HAVE_ARCH_KSTACK_END
2380 static inline int kstack_end(void *addr)
2382 /* Reliable end of stack detection:
2383 * Some APM bios versions misalign the stack
2385 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2389 extern union thread_union init_thread_union;
2390 extern struct task_struct init_task;
2392 extern struct mm_struct init_mm;
2394 extern struct pid_namespace init_pid_ns;
2397 * find a task by one of its numerical ids
2399 * find_task_by_pid_ns():
2400 * finds a task by its pid in the specified namespace
2401 * find_task_by_vpid():
2402 * finds a task by its virtual pid
2404 * see also find_vpid() etc in include/linux/pid.h
2407 extern struct task_struct *find_task_by_vpid(pid_t nr);
2408 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2409 struct pid_namespace *ns);
2411 /* per-UID process charging. */
2412 extern struct user_struct * alloc_uid(kuid_t);
2413 static inline struct user_struct *get_uid(struct user_struct *u)
2415 atomic_inc(&u->__count);
2418 extern void free_uid(struct user_struct *);
2420 #include <asm/current.h>
2422 extern void xtime_update(unsigned long ticks);
2424 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2425 extern int wake_up_process(struct task_struct *tsk);
2426 extern void wake_up_new_task(struct task_struct *tsk);
2428 extern void kick_process(struct task_struct *tsk);
2430 static inline void kick_process(struct task_struct *tsk) { }
2432 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2433 extern void sched_dead(struct task_struct *p);
2435 extern void proc_caches_init(void);
2436 extern void flush_signals(struct task_struct *);
2437 extern void __flush_signals(struct task_struct *);
2438 extern void ignore_signals(struct task_struct *);
2439 extern void flush_signal_handlers(struct task_struct *, int force_default);
2440 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2442 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2444 unsigned long flags;
2447 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2448 ret = dequeue_signal(tsk, mask, info);
2449 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2454 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2456 extern void unblock_all_signals(void);
2457 extern void release_task(struct task_struct * p);
2458 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2459 extern int force_sigsegv(int, struct task_struct *);
2460 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2461 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2462 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2463 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2464 const struct cred *, u32);
2465 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2466 extern int kill_pid(struct pid *pid, int sig, int priv);
2467 extern int kill_proc_info(int, struct siginfo *, pid_t);
2468 extern __must_check bool do_notify_parent(struct task_struct *, int);
2469 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2470 extern void force_sig(int, struct task_struct *);
2471 extern int send_sig(int, struct task_struct *, int);
2472 extern int zap_other_threads(struct task_struct *p);
2473 extern struct sigqueue *sigqueue_alloc(void);
2474 extern void sigqueue_free(struct sigqueue *);
2475 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2476 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2478 static inline void restore_saved_sigmask(void)
2480 if (test_and_clear_restore_sigmask())
2481 __set_current_blocked(¤t->saved_sigmask);
2484 static inline sigset_t *sigmask_to_save(void)
2486 sigset_t *res = ¤t->blocked;
2487 if (unlikely(test_restore_sigmask()))
2488 res = ¤t->saved_sigmask;
2492 static inline int kill_cad_pid(int sig, int priv)
2494 return kill_pid(cad_pid, sig, priv);
2497 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2498 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2499 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2500 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2503 * True if we are on the alternate signal stack.
2505 static inline int on_sig_stack(unsigned long sp)
2507 #ifdef CONFIG_STACK_GROWSUP
2508 return sp >= current->sas_ss_sp &&
2509 sp - current->sas_ss_sp < current->sas_ss_size;
2511 return sp > current->sas_ss_sp &&
2512 sp - current->sas_ss_sp <= current->sas_ss_size;
2516 static inline int sas_ss_flags(unsigned long sp)
2518 if (!current->sas_ss_size)
2521 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2524 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2526 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2527 #ifdef CONFIG_STACK_GROWSUP
2528 return current->sas_ss_sp;
2530 return current->sas_ss_sp + current->sas_ss_size;
2536 * Routines for handling mm_structs
2538 extern struct mm_struct * mm_alloc(void);
2540 /* mmdrop drops the mm and the page tables */
2541 extern void __mmdrop(struct mm_struct *);
2542 static inline void mmdrop(struct mm_struct * mm)
2544 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2548 /* mmput gets rid of the mappings and all user-space */
2549 extern void mmput(struct mm_struct *);
2550 /* Grab a reference to a task's mm, if it is not already going away */
2551 extern struct mm_struct *get_task_mm(struct task_struct *task);
2553 * Grab a reference to a task's mm, if it is not already going away
2554 * and ptrace_may_access with the mode parameter passed to it
2557 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2558 /* Remove the current tasks stale references to the old mm_struct */
2559 extern void mm_release(struct task_struct *, struct mm_struct *);
2561 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2562 struct task_struct *);
2563 extern void flush_thread(void);
2564 extern void exit_thread(void);
2566 extern void exit_files(struct task_struct *);
2567 extern void __cleanup_sighand(struct sighand_struct *);
2569 extern void exit_itimers(struct signal_struct *);
2570 extern void flush_itimer_signals(void);
2572 extern void do_group_exit(int);
2574 extern int do_execve(struct filename *,
2575 const char __user * const __user *,
2576 const char __user * const __user *);
2577 extern int do_execveat(int, struct filename *,
2578 const char __user * const __user *,
2579 const char __user * const __user *,
2581 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2582 struct task_struct *fork_idle(int);
2583 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2585 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2586 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2588 __set_task_comm(tsk, from, false);
2590 extern char *get_task_comm(char *to, struct task_struct *tsk);
2593 void scheduler_ipi(void);
2594 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2596 static inline void scheduler_ipi(void) { }
2597 static inline unsigned long wait_task_inactive(struct task_struct *p,
2604 #define next_task(p) \
2605 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2607 #define for_each_process(p) \
2608 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2610 extern bool current_is_single_threaded(void);
2613 * Careful: do_each_thread/while_each_thread is a double loop so
2614 * 'break' will not work as expected - use goto instead.
2616 #define do_each_thread(g, t) \
2617 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2619 #define while_each_thread(g, t) \
2620 while ((t = next_thread(t)) != g)
2622 #define __for_each_thread(signal, t) \
2623 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2625 #define for_each_thread(p, t) \
2626 __for_each_thread((p)->signal, t)
2628 /* Careful: this is a double loop, 'break' won't work as expected. */
2629 #define for_each_process_thread(p, t) \
2630 for_each_process(p) for_each_thread(p, t)
2632 static inline int get_nr_threads(struct task_struct *tsk)
2634 return tsk->signal->nr_threads;
2637 static inline bool thread_group_leader(struct task_struct *p)
2639 return p->exit_signal >= 0;
2642 /* Do to the insanities of de_thread it is possible for a process
2643 * to have the pid of the thread group leader without actually being
2644 * the thread group leader. For iteration through the pids in proc
2645 * all we care about is that we have a task with the appropriate
2646 * pid, we don't actually care if we have the right task.
2648 static inline bool has_group_leader_pid(struct task_struct *p)
2650 return task_pid(p) == p->signal->leader_pid;
2654 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2656 return p1->signal == p2->signal;
2659 static inline struct task_struct *next_thread(const struct task_struct *p)
2661 return list_entry_rcu(p->thread_group.next,
2662 struct task_struct, thread_group);
2665 static inline int thread_group_empty(struct task_struct *p)
2667 return list_empty(&p->thread_group);
2670 #define delay_group_leader(p) \
2671 (thread_group_leader(p) && !thread_group_empty(p))
2674 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2675 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2676 * pins the final release of task.io_context. Also protects ->cpuset and
2677 * ->cgroup.subsys[]. And ->vfork_done.
2679 * Nests both inside and outside of read_lock(&tasklist_lock).
2680 * It must not be nested with write_lock_irq(&tasklist_lock),
2681 * neither inside nor outside.
2683 static inline void task_lock(struct task_struct *p)
2685 spin_lock(&p->alloc_lock);
2688 static inline void task_unlock(struct task_struct *p)
2690 spin_unlock(&p->alloc_lock);
2693 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2694 unsigned long *flags);
2696 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2697 unsigned long *flags)
2699 struct sighand_struct *ret;
2701 ret = __lock_task_sighand(tsk, flags);
2702 (void)__cond_lock(&tsk->sighand->siglock, ret);
2706 static inline void unlock_task_sighand(struct task_struct *tsk,
2707 unsigned long *flags)
2709 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2712 #ifdef CONFIG_CGROUPS
2713 static inline void threadgroup_change_begin(struct task_struct *tsk)
2715 down_read(&tsk->signal->group_rwsem);
2717 static inline void threadgroup_change_end(struct task_struct *tsk)
2719 up_read(&tsk->signal->group_rwsem);
2723 * threadgroup_lock - lock threadgroup
2724 * @tsk: member task of the threadgroup to lock
2726 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2727 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2728 * change ->group_leader/pid. This is useful for cases where the threadgroup
2729 * needs to stay stable across blockable operations.
2731 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2732 * synchronization. While held, no new task will be added to threadgroup
2733 * and no existing live task will have its PF_EXITING set.
2735 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2736 * sub-thread becomes a new leader.
2738 static inline void threadgroup_lock(struct task_struct *tsk)
2740 down_write(&tsk->signal->group_rwsem);
2744 * threadgroup_unlock - unlock threadgroup
2745 * @tsk: member task of the threadgroup to unlock
2747 * Reverse threadgroup_lock().
2749 static inline void threadgroup_unlock(struct task_struct *tsk)
2751 up_write(&tsk->signal->group_rwsem);
2754 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2755 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2756 static inline void threadgroup_lock(struct task_struct *tsk) {}
2757 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2760 #ifndef __HAVE_THREAD_FUNCTIONS
2762 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2763 #define task_stack_page(task) ((task)->stack)
2765 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2767 *task_thread_info(p) = *task_thread_info(org);
2768 task_thread_info(p)->task = p;
2772 * Return the address of the last usable long on the stack.
2774 * When the stack grows down, this is just above the thread
2775 * info struct. Going any lower will corrupt the threadinfo.
2777 * When the stack grows up, this is the highest address.
2778 * Beyond that position, we corrupt data on the next page.
2780 static inline unsigned long *end_of_stack(struct task_struct *p)
2782 #ifdef CONFIG_STACK_GROWSUP
2783 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2785 return (unsigned long *)(task_thread_info(p) + 1);
2790 #define task_stack_end_corrupted(task) \
2791 (*(end_of_stack(task)) != STACK_END_MAGIC)
2793 static inline int object_is_on_stack(void *obj)
2795 void *stack = task_stack_page(current);
2797 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2800 extern void thread_info_cache_init(void);
2802 #ifdef CONFIG_DEBUG_STACK_USAGE
2803 static inline unsigned long stack_not_used(struct task_struct *p)
2805 unsigned long *n = end_of_stack(p);
2807 do { /* Skip over canary */
2811 return (unsigned long)n - (unsigned long)end_of_stack(p);
2814 extern void set_task_stack_end_magic(struct task_struct *tsk);
2816 /* set thread flags in other task's structures
2817 * - see asm/thread_info.h for TIF_xxxx flags available
2819 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2821 set_ti_thread_flag(task_thread_info(tsk), flag);
2824 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2826 clear_ti_thread_flag(task_thread_info(tsk), flag);
2829 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2831 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2834 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2836 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2839 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2841 return test_ti_thread_flag(task_thread_info(tsk), flag);
2844 static inline void set_tsk_need_resched(struct task_struct *tsk)
2846 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2849 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2851 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2854 static inline int test_tsk_need_resched(struct task_struct *tsk)
2856 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2859 static inline int restart_syscall(void)
2861 set_tsk_thread_flag(current, TIF_SIGPENDING);
2862 return -ERESTARTNOINTR;
2865 static inline int signal_pending(struct task_struct *p)
2867 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2870 static inline int __fatal_signal_pending(struct task_struct *p)
2872 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2875 static inline int fatal_signal_pending(struct task_struct *p)
2877 return signal_pending(p) && __fatal_signal_pending(p);
2880 static inline int signal_pending_state(long state, struct task_struct *p)
2882 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2884 if (!signal_pending(p))
2887 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2891 * cond_resched() and cond_resched_lock(): latency reduction via
2892 * explicit rescheduling in places that are safe. The return
2893 * value indicates whether a reschedule was done in fact.
2894 * cond_resched_lock() will drop the spinlock before scheduling,
2895 * cond_resched_softirq() will enable bhs before scheduling.
2897 extern int _cond_resched(void);
2899 #define cond_resched() ({ \
2900 ___might_sleep(__FILE__, __LINE__, 0); \
2904 extern int __cond_resched_lock(spinlock_t *lock);
2906 #ifdef CONFIG_PREEMPT_COUNT
2907 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2909 #define PREEMPT_LOCK_OFFSET 0
2912 #define cond_resched_lock(lock) ({ \
2913 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2914 __cond_resched_lock(lock); \
2917 extern int __cond_resched_softirq(void);
2919 #define cond_resched_softirq() ({ \
2920 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2921 __cond_resched_softirq(); \
2924 static inline void cond_resched_rcu(void)
2926 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2934 * Does a critical section need to be broken due to another
2935 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2936 * but a general need for low latency)
2938 static inline int spin_needbreak(spinlock_t *lock)
2940 #ifdef CONFIG_PREEMPT
2941 return spin_is_contended(lock);
2948 * Idle thread specific functions to determine the need_resched
2951 #ifdef TIF_POLLING_NRFLAG
2952 static inline int tsk_is_polling(struct task_struct *p)
2954 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2957 static inline void __current_set_polling(void)
2959 set_thread_flag(TIF_POLLING_NRFLAG);
2962 static inline bool __must_check current_set_polling_and_test(void)
2964 __current_set_polling();
2967 * Polling state must be visible before we test NEED_RESCHED,
2968 * paired by resched_curr()
2970 smp_mb__after_atomic();
2972 return unlikely(tif_need_resched());
2975 static inline void __current_clr_polling(void)
2977 clear_thread_flag(TIF_POLLING_NRFLAG);
2980 static inline bool __must_check current_clr_polling_and_test(void)
2982 __current_clr_polling();
2985 * Polling state must be visible before we test NEED_RESCHED,
2986 * paired by resched_curr()
2988 smp_mb__after_atomic();
2990 return unlikely(tif_need_resched());
2994 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2995 static inline void __current_set_polling(void) { }
2996 static inline void __current_clr_polling(void) { }
2998 static inline bool __must_check current_set_polling_and_test(void)
3000 return unlikely(tif_need_resched());
3002 static inline bool __must_check current_clr_polling_and_test(void)
3004 return unlikely(tif_need_resched());
3008 static inline void current_clr_polling(void)
3010 __current_clr_polling();
3013 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3014 * Once the bit is cleared, we'll get IPIs with every new
3015 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3018 smp_mb(); /* paired with resched_curr() */
3020 preempt_fold_need_resched();
3023 static __always_inline bool need_resched(void)
3025 return unlikely(tif_need_resched());
3029 * Thread group CPU time accounting.
3031 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3032 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3035 * Reevaluate whether the task has signals pending delivery.
3036 * Wake the task if so.
3037 * This is required every time the blocked sigset_t changes.
3038 * callers must hold sighand->siglock.
3040 extern void recalc_sigpending_and_wake(struct task_struct *t);
3041 extern void recalc_sigpending(void);
3043 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3045 static inline void signal_wake_up(struct task_struct *t, bool resume)
3047 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3049 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3051 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3055 * Wrappers for p->thread_info->cpu access. No-op on UP.
3059 static inline unsigned int task_cpu(const struct task_struct *p)
3061 return task_thread_info(p)->cpu;
3064 static inline int task_node(const struct task_struct *p)
3066 return cpu_to_node(task_cpu(p));
3069 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3073 static inline unsigned int task_cpu(const struct task_struct *p)
3078 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3082 #endif /* CONFIG_SMP */
3084 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3085 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3087 #ifdef CONFIG_CGROUP_SCHED
3088 extern struct task_group root_task_group;
3089 #endif /* CONFIG_CGROUP_SCHED */
3091 extern int task_can_switch_user(struct user_struct *up,
3092 struct task_struct *tsk);
3094 #ifdef CONFIG_TASK_XACCT
3095 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3097 tsk->ioac.rchar += amt;
3100 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3102 tsk->ioac.wchar += amt;
3105 static inline void inc_syscr(struct task_struct *tsk)
3110 static inline void inc_syscw(struct task_struct *tsk)
3115 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3119 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3123 static inline void inc_syscr(struct task_struct *tsk)
3127 static inline void inc_syscw(struct task_struct *tsk)
3132 #ifndef TASK_SIZE_OF
3133 #define TASK_SIZE_OF(tsk) TASK_SIZE
3137 extern void mm_update_next_owner(struct mm_struct *mm);
3139 static inline void mm_update_next_owner(struct mm_struct *mm)
3142 #endif /* CONFIG_MEMCG */
3144 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3147 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3150 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3153 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3156 static inline unsigned long rlimit(unsigned int limit)
3158 return task_rlimit(current, limit);
3161 static inline unsigned long rlimit_max(unsigned int limit)
3163 return task_rlimit_max(current, limit);