4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
61 #include <linux/cgroup-defs.h>
63 #include <asm/processor.h>
65 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
68 * Extended scheduling parameters data structure.
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
91 * This is reflected by the actual fields of the sched_attr structure:
93 * @size size of the structure, for fwd/bwd compat.
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
117 /* SCHED_NORMAL, SCHED_BATCH */
120 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state;
130 struct robust_list_head;
133 struct perf_event_context;
138 #define VMACACHE_BITS 2
139 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
140 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
152 extern unsigned long avenrun[]; /* Load averages */
153 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
155 #define FSHIFT 11 /* nr of bits of precision */
156 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
157 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
158 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159 #define EXP_5 2014 /* 1/exp(5sec/5min) */
160 #define EXP_15 2037 /* 1/exp(5sec/15min) */
162 #define CALC_LOAD(load,exp,n) \
164 load += n*(FIXED_1-exp); \
167 extern unsigned long total_forks;
168 extern int nr_threads;
169 DECLARE_PER_CPU(unsigned long, process_counts);
170 extern int nr_processes(void);
171 extern unsigned long nr_running(void);
172 extern bool single_task_running(void);
173 extern unsigned long nr_iowait(void);
174 extern unsigned long nr_iowait_cpu(int cpu);
175 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
177 extern void calc_global_load(unsigned long ticks);
179 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
180 extern void update_cpu_load_nohz(void);
182 static inline void update_cpu_load_nohz(void) { }
185 extern unsigned long get_parent_ip(unsigned long addr);
187 extern void dump_cpu_task(int cpu);
192 #ifdef CONFIG_SCHED_DEBUG
193 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
194 extern void proc_sched_set_task(struct task_struct *p);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_NOLOAD 1024
222 #define TASK_STATE_MAX 2048
224 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
226 extern char ___assert_task_state[1 - 2*!!(
227 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
229 /* Convenience macros for the sake of set_task_state */
230 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
231 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
232 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
234 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
236 /* Convenience macros for the sake of wake_up */
237 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
238 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
240 /* get_task_state() */
241 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
242 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
243 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
245 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
246 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
247 #define task_is_stopped_or_traced(task) \
248 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
249 #define task_contributes_to_load(task) \
250 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
251 (task->flags & PF_FROZEN) == 0 && \
252 (task->state & TASK_NOLOAD) == 0)
254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256 #define __set_task_state(tsk, state_value) \
258 (tsk)->task_state_change = _THIS_IP_; \
259 (tsk)->state = (state_value); \
261 #define set_task_state(tsk, state_value) \
263 (tsk)->task_state_change = _THIS_IP_; \
264 smp_store_mb((tsk)->state, (state_value)); \
268 * set_current_state() includes a barrier so that the write of current->state
269 * is correctly serialised wrt the caller's subsequent test of whether to
272 * set_current_state(TASK_UNINTERRUPTIBLE);
273 * if (do_i_need_to_sleep())
276 * If the caller does not need such serialisation then use __set_current_state()
278 #define __set_current_state(state_value) \
280 current->task_state_change = _THIS_IP_; \
281 current->state = (state_value); \
283 #define set_current_state(state_value) \
285 current->task_state_change = _THIS_IP_; \
286 smp_store_mb(current->state, (state_value)); \
291 #define __set_task_state(tsk, state_value) \
292 do { (tsk)->state = (state_value); } while (0)
293 #define set_task_state(tsk, state_value) \
294 smp_store_mb((tsk)->state, (state_value))
297 * set_current_state() includes a barrier so that the write of current->state
298 * is correctly serialised wrt the caller's subsequent test of whether to
301 * set_current_state(TASK_UNINTERRUPTIBLE);
302 * if (do_i_need_to_sleep())
305 * If the caller does not need such serialisation then use __set_current_state()
307 #define __set_current_state(state_value) \
308 do { current->state = (state_value); } while (0)
309 #define set_current_state(state_value) \
310 smp_store_mb(current->state, (state_value))
314 /* Task command name length */
315 #define TASK_COMM_LEN 16
317 #include <linux/spinlock.h>
320 * This serializes "schedule()" and also protects
321 * the run-queue from deletions/modifications (but
322 * _adding_ to the beginning of the run-queue has
325 extern rwlock_t tasklist_lock;
326 extern spinlock_t mmlist_lock;
330 #ifdef CONFIG_PROVE_RCU
331 extern int lockdep_tasklist_lock_is_held(void);
332 #endif /* #ifdef CONFIG_PROVE_RCU */
334 extern void sched_init(void);
335 extern void sched_init_smp(void);
336 extern asmlinkage void schedule_tail(struct task_struct *prev);
337 extern void init_idle(struct task_struct *idle, int cpu);
338 extern void init_idle_bootup_task(struct task_struct *idle);
340 extern cpumask_var_t cpu_isolated_map;
342 extern int runqueue_is_locked(int cpu);
344 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
345 extern void nohz_balance_enter_idle(int cpu);
346 extern void set_cpu_sd_state_idle(void);
347 extern int get_nohz_timer_target(void);
349 static inline void nohz_balance_enter_idle(int cpu) { }
350 static inline void set_cpu_sd_state_idle(void) { }
354 * Only dump TASK_* tasks. (0 for all tasks)
356 extern void show_state_filter(unsigned long state_filter);
358 static inline void show_state(void)
360 show_state_filter(0);
363 extern void show_regs(struct pt_regs *);
366 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
367 * task), SP is the stack pointer of the first frame that should be shown in the back
368 * trace (or NULL if the entire call-chain of the task should be shown).
370 extern void show_stack(struct task_struct *task, unsigned long *sp);
372 extern void cpu_init (void);
373 extern void trap_init(void);
374 extern void update_process_times(int user);
375 extern void scheduler_tick(void);
377 extern void sched_show_task(struct task_struct *p);
379 #ifdef CONFIG_LOCKUP_DETECTOR
380 extern void touch_softlockup_watchdog(void);
381 extern void touch_softlockup_watchdog_sync(void);
382 extern void touch_all_softlockup_watchdogs(void);
383 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
385 size_t *lenp, loff_t *ppos);
386 extern unsigned int softlockup_panic;
387 void lockup_detector_init(void);
389 static inline void touch_softlockup_watchdog(void)
392 static inline void touch_softlockup_watchdog_sync(void)
395 static inline void touch_all_softlockup_watchdogs(void)
398 static inline void lockup_detector_init(void)
403 #ifdef CONFIG_DETECT_HUNG_TASK
404 void reset_hung_task_detector(void);
406 static inline void reset_hung_task_detector(void)
411 /* Attach to any functions which should be ignored in wchan output. */
412 #define __sched __attribute__((__section__(".sched.text")))
414 /* Linker adds these: start and end of __sched functions */
415 extern char __sched_text_start[], __sched_text_end[];
417 /* Is this address in the __sched functions? */
418 extern int in_sched_functions(unsigned long addr);
420 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
421 extern signed long schedule_timeout(signed long timeout);
422 extern signed long schedule_timeout_interruptible(signed long timeout);
423 extern signed long schedule_timeout_killable(signed long timeout);
424 extern signed long schedule_timeout_uninterruptible(signed long timeout);
425 asmlinkage void schedule(void);
426 extern void schedule_preempt_disabled(void);
428 extern long io_schedule_timeout(long timeout);
430 static inline void io_schedule(void)
432 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
436 struct user_namespace;
439 extern void arch_pick_mmap_layout(struct mm_struct *mm);
441 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
442 unsigned long, unsigned long);
444 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
445 unsigned long len, unsigned long pgoff,
446 unsigned long flags);
448 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
451 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
452 #define SUID_DUMP_USER 1 /* Dump as user of process */
453 #define SUID_DUMP_ROOT 2 /* Dump as root */
457 /* for SUID_DUMP_* above */
458 #define MMF_DUMPABLE_BITS 2
459 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
461 extern void set_dumpable(struct mm_struct *mm, int value);
463 * This returns the actual value of the suid_dumpable flag. For things
464 * that are using this for checking for privilege transitions, it must
465 * test against SUID_DUMP_USER rather than treating it as a boolean
468 static inline int __get_dumpable(unsigned long mm_flags)
470 return mm_flags & MMF_DUMPABLE_MASK;
473 static inline int get_dumpable(struct mm_struct *mm)
475 return __get_dumpable(mm->flags);
478 /* coredump filter bits */
479 #define MMF_DUMP_ANON_PRIVATE 2
480 #define MMF_DUMP_ANON_SHARED 3
481 #define MMF_DUMP_MAPPED_PRIVATE 4
482 #define MMF_DUMP_MAPPED_SHARED 5
483 #define MMF_DUMP_ELF_HEADERS 6
484 #define MMF_DUMP_HUGETLB_PRIVATE 7
485 #define MMF_DUMP_HUGETLB_SHARED 8
487 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
488 #define MMF_DUMP_FILTER_BITS 7
489 #define MMF_DUMP_FILTER_MASK \
490 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
491 #define MMF_DUMP_FILTER_DEFAULT \
492 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
493 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
495 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
496 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
498 # define MMF_DUMP_MASK_DEFAULT_ELF 0
500 /* leave room for more dump flags */
501 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
502 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
503 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
505 #define MMF_HAS_UPROBES 19 /* has uprobes */
506 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
508 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
510 struct sighand_struct {
512 struct k_sigaction action[_NSIG];
514 wait_queue_head_t signalfd_wqh;
517 struct pacct_struct {
520 unsigned long ac_mem;
521 cputime_t ac_utime, ac_stime;
522 unsigned long ac_minflt, ac_majflt;
533 * struct prev_cputime - snaphsot of system and user cputime
534 * @utime: time spent in user mode
535 * @stime: time spent in system mode
536 * @lock: protects the above two fields
538 * Stores previous user/system time values such that we can guarantee
541 struct prev_cputime {
542 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
549 static inline void prev_cputime_init(struct prev_cputime *prev)
551 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
552 prev->utime = prev->stime = 0;
553 raw_spin_lock_init(&prev->lock);
558 * struct task_cputime - collected CPU time counts
559 * @utime: time spent in user mode, in &cputime_t units
560 * @stime: time spent in kernel mode, in &cputime_t units
561 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
563 * This structure groups together three kinds of CPU time that are tracked for
564 * threads and thread groups. Most things considering CPU time want to group
565 * these counts together and treat all three of them in parallel.
567 struct task_cputime {
570 unsigned long long sum_exec_runtime;
573 /* Alternate field names when used to cache expirations. */
574 #define virt_exp utime
575 #define prof_exp stime
576 #define sched_exp sum_exec_runtime
578 #define INIT_CPUTIME \
579 (struct task_cputime) { \
582 .sum_exec_runtime = 0, \
586 * This is the atomic variant of task_cputime, which can be used for
587 * storing and updating task_cputime statistics without locking.
589 struct task_cputime_atomic {
592 atomic64_t sum_exec_runtime;
595 #define INIT_CPUTIME_ATOMIC \
596 (struct task_cputime_atomic) { \
597 .utime = ATOMIC64_INIT(0), \
598 .stime = ATOMIC64_INIT(0), \
599 .sum_exec_runtime = ATOMIC64_INIT(0), \
602 #ifdef CONFIG_PREEMPT_COUNT
603 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
605 #define PREEMPT_DISABLED PREEMPT_ENABLED
609 * Disable preemption until the scheduler is running.
610 * Reset by start_kernel()->sched_init()->init_idle().
612 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
613 * before the scheduler is active -- see should_resched().
615 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
618 * struct thread_group_cputimer - thread group interval timer counts
619 * @cputime_atomic: atomic thread group interval timers.
620 * @running: non-zero when there are timers running and
621 * @cputime receives updates.
623 * This structure contains the version of task_cputime, above, that is
624 * used for thread group CPU timer calculations.
626 struct thread_group_cputimer {
627 struct task_cputime_atomic cputime_atomic;
631 #include <linux/rwsem.h>
635 * NOTE! "signal_struct" does not have its own
636 * locking, because a shared signal_struct always
637 * implies a shared sighand_struct, so locking
638 * sighand_struct is always a proper superset of
639 * the locking of signal_struct.
641 struct signal_struct {
645 struct list_head thread_head;
647 wait_queue_head_t wait_chldexit; /* for wait4() */
649 /* current thread group signal load-balancing target: */
650 struct task_struct *curr_target;
652 /* shared signal handling: */
653 struct sigpending shared_pending;
655 /* thread group exit support */
658 * - notify group_exit_task when ->count is equal to notify_count
659 * - everyone except group_exit_task is stopped during signal delivery
660 * of fatal signals, group_exit_task processes the signal.
663 struct task_struct *group_exit_task;
665 /* thread group stop support, overloads group_exit_code too */
666 int group_stop_count;
667 unsigned int flags; /* see SIGNAL_* flags below */
670 * PR_SET_CHILD_SUBREAPER marks a process, like a service
671 * manager, to re-parent orphan (double-forking) child processes
672 * to this process instead of 'init'. The service manager is
673 * able to receive SIGCHLD signals and is able to investigate
674 * the process until it calls wait(). All children of this
675 * process will inherit a flag if they should look for a
676 * child_subreaper process at exit.
678 unsigned int is_child_subreaper:1;
679 unsigned int has_child_subreaper:1;
681 /* POSIX.1b Interval Timers */
683 struct list_head posix_timers;
685 /* ITIMER_REAL timer for the process */
686 struct hrtimer real_timer;
687 struct pid *leader_pid;
688 ktime_t it_real_incr;
691 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
692 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
693 * values are defined to 0 and 1 respectively
695 struct cpu_itimer it[2];
698 * Thread group totals for process CPU timers.
699 * See thread_group_cputimer(), et al, for details.
701 struct thread_group_cputimer cputimer;
703 /* Earliest-expiration cache. */
704 struct task_cputime cputime_expires;
706 struct list_head cpu_timers[3];
708 struct pid *tty_old_pgrp;
710 /* boolean value for session group leader */
713 struct tty_struct *tty; /* NULL if no tty */
715 #ifdef CONFIG_SCHED_AUTOGROUP
716 struct autogroup *autogroup;
719 * Cumulative resource counters for dead threads in the group,
720 * and for reaped dead child processes forked by this group.
721 * Live threads maintain their own counters and add to these
722 * in __exit_signal, except for the group leader.
724 seqlock_t stats_lock;
725 cputime_t utime, stime, cutime, cstime;
728 struct prev_cputime prev_cputime;
729 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
730 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
731 unsigned long inblock, oublock, cinblock, coublock;
732 unsigned long maxrss, cmaxrss;
733 struct task_io_accounting ioac;
736 * Cumulative ns of schedule CPU time fo dead threads in the
737 * group, not including a zombie group leader, (This only differs
738 * from jiffies_to_ns(utime + stime) if sched_clock uses something
739 * other than jiffies.)
741 unsigned long long sum_sched_runtime;
744 * We don't bother to synchronize most readers of this at all,
745 * because there is no reader checking a limit that actually needs
746 * to get both rlim_cur and rlim_max atomically, and either one
747 * alone is a single word that can safely be read normally.
748 * getrlimit/setrlimit use task_lock(current->group_leader) to
749 * protect this instead of the siglock, because they really
750 * have no need to disable irqs.
752 struct rlimit rlim[RLIM_NLIMITS];
754 #ifdef CONFIG_BSD_PROCESS_ACCT
755 struct pacct_struct pacct; /* per-process accounting information */
757 #ifdef CONFIG_TASKSTATS
758 struct taskstats *stats;
762 unsigned audit_tty_log_passwd;
763 struct tty_audit_buf *tty_audit_buf;
766 oom_flags_t oom_flags;
767 short oom_score_adj; /* OOM kill score adjustment */
768 short oom_score_adj_min; /* OOM kill score adjustment min value.
769 * Only settable by CAP_SYS_RESOURCE. */
771 struct mutex cred_guard_mutex; /* guard against foreign influences on
772 * credential calculations
773 * (notably. ptrace) */
777 * Bits in flags field of signal_struct.
779 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
780 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
781 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
782 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
784 * Pending notifications to parent.
786 #define SIGNAL_CLD_STOPPED 0x00000010
787 #define SIGNAL_CLD_CONTINUED 0x00000020
788 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
790 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
792 /* If true, all threads except ->group_exit_task have pending SIGKILL */
793 static inline int signal_group_exit(const struct signal_struct *sig)
795 return (sig->flags & SIGNAL_GROUP_EXIT) ||
796 (sig->group_exit_task != NULL);
800 * Some day this will be a full-fledged user tracking system..
803 atomic_t __count; /* reference count */
804 atomic_t processes; /* How many processes does this user have? */
805 atomic_t sigpending; /* How many pending signals does this user have? */
806 #ifdef CONFIG_INOTIFY_USER
807 atomic_t inotify_watches; /* How many inotify watches does this user have? */
808 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
810 #ifdef CONFIG_FANOTIFY
811 atomic_t fanotify_listeners;
814 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
816 #ifdef CONFIG_POSIX_MQUEUE
817 /* protected by mq_lock */
818 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
820 unsigned long locked_shm; /* How many pages of mlocked shm ? */
823 struct key *uid_keyring; /* UID specific keyring */
824 struct key *session_keyring; /* UID's default session keyring */
827 /* Hash table maintenance information */
828 struct hlist_node uidhash_node;
831 #ifdef CONFIG_PERF_EVENTS
832 atomic_long_t locked_vm;
836 extern int uids_sysfs_init(void);
838 extern struct user_struct *find_user(kuid_t);
840 extern struct user_struct root_user;
841 #define INIT_USER (&root_user)
844 struct backing_dev_info;
845 struct reclaim_state;
847 #ifdef CONFIG_SCHED_INFO
849 /* cumulative counters */
850 unsigned long pcount; /* # of times run on this cpu */
851 unsigned long long run_delay; /* time spent waiting on a runqueue */
854 unsigned long long last_arrival,/* when we last ran on a cpu */
855 last_queued; /* when we were last queued to run */
857 #endif /* CONFIG_SCHED_INFO */
859 #ifdef CONFIG_TASK_DELAY_ACCT
860 struct task_delay_info {
862 unsigned int flags; /* Private per-task flags */
864 /* For each stat XXX, add following, aligned appropriately
866 * struct timespec XXX_start, XXX_end;
870 * Atomicity of updates to XXX_delay, XXX_count protected by
871 * single lock above (split into XXX_lock if contention is an issue).
875 * XXX_count is incremented on every XXX operation, the delay
876 * associated with the operation is added to XXX_delay.
877 * XXX_delay contains the accumulated delay time in nanoseconds.
879 u64 blkio_start; /* Shared by blkio, swapin */
880 u64 blkio_delay; /* wait for sync block io completion */
881 u64 swapin_delay; /* wait for swapin block io completion */
882 u32 blkio_count; /* total count of the number of sync block */
883 /* io operations performed */
884 u32 swapin_count; /* total count of the number of swapin block */
885 /* io operations performed */
888 u64 freepages_delay; /* wait for memory reclaim */
889 u32 freepages_count; /* total count of memory reclaim */
891 #endif /* CONFIG_TASK_DELAY_ACCT */
893 static inline int sched_info_on(void)
895 #ifdef CONFIG_SCHEDSTATS
897 #elif defined(CONFIG_TASK_DELAY_ACCT)
898 extern int delayacct_on;
913 * Increase resolution of cpu_capacity calculations
915 #define SCHED_CAPACITY_SHIFT 10
916 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
919 * Wake-queues are lists of tasks with a pending wakeup, whose
920 * callers have already marked the task as woken internally,
921 * and can thus carry on. A common use case is being able to
922 * do the wakeups once the corresponding user lock as been
925 * We hold reference to each task in the list across the wakeup,
926 * thus guaranteeing that the memory is still valid by the time
927 * the actual wakeups are performed in wake_up_q().
929 * One per task suffices, because there's never a need for a task to be
930 * in two wake queues simultaneously; it is forbidden to abandon a task
931 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
932 * already in a wake queue, the wakeup will happen soon and the second
933 * waker can just skip it.
935 * The WAKE_Q macro declares and initializes the list head.
936 * wake_up_q() does NOT reinitialize the list; it's expected to be
937 * called near the end of a function, where the fact that the queue is
938 * not used again will be easy to see by inspection.
940 * Note that this can cause spurious wakeups. schedule() callers
941 * must ensure the call is done inside a loop, confirming that the
942 * wakeup condition has in fact occurred.
945 struct wake_q_node *next;
949 struct wake_q_node *first;
950 struct wake_q_node **lastp;
953 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
955 #define WAKE_Q(name) \
956 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
958 extern void wake_q_add(struct wake_q_head *head,
959 struct task_struct *task);
960 extern void wake_up_q(struct wake_q_head *head);
963 * sched-domains (multiprocessor balancing) declarations:
966 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
967 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
968 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
969 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
970 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
971 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
972 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
973 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
974 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
975 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
976 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
977 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
978 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
979 #define SD_NUMA 0x4000 /* cross-node balancing */
981 #ifdef CONFIG_SCHED_SMT
982 static inline int cpu_smt_flags(void)
984 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
988 #ifdef CONFIG_SCHED_MC
989 static inline int cpu_core_flags(void)
991 return SD_SHARE_PKG_RESOURCES;
996 static inline int cpu_numa_flags(void)
1002 struct sched_domain_attr {
1003 int relax_domain_level;
1006 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1007 .relax_domain_level = -1, \
1010 extern int sched_domain_level_max;
1014 struct sched_domain {
1015 /* These fields must be setup */
1016 struct sched_domain *parent; /* top domain must be null terminated */
1017 struct sched_domain *child; /* bottom domain must be null terminated */
1018 struct sched_group *groups; /* the balancing groups of the domain */
1019 unsigned long min_interval; /* Minimum balance interval ms */
1020 unsigned long max_interval; /* Maximum balance interval ms */
1021 unsigned int busy_factor; /* less balancing by factor if busy */
1022 unsigned int imbalance_pct; /* No balance until over watermark */
1023 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1024 unsigned int busy_idx;
1025 unsigned int idle_idx;
1026 unsigned int newidle_idx;
1027 unsigned int wake_idx;
1028 unsigned int forkexec_idx;
1029 unsigned int smt_gain;
1031 int nohz_idle; /* NOHZ IDLE status */
1032 int flags; /* See SD_* */
1035 /* Runtime fields. */
1036 unsigned long last_balance; /* init to jiffies. units in jiffies */
1037 unsigned int balance_interval; /* initialise to 1. units in ms. */
1038 unsigned int nr_balance_failed; /* initialise to 0 */
1040 /* idle_balance() stats */
1041 u64 max_newidle_lb_cost;
1042 unsigned long next_decay_max_lb_cost;
1044 #ifdef CONFIG_SCHEDSTATS
1045 /* load_balance() stats */
1046 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1047 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1048 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1049 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1050 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1051 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1052 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1053 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1055 /* Active load balancing */
1056 unsigned int alb_count;
1057 unsigned int alb_failed;
1058 unsigned int alb_pushed;
1060 /* SD_BALANCE_EXEC stats */
1061 unsigned int sbe_count;
1062 unsigned int sbe_balanced;
1063 unsigned int sbe_pushed;
1065 /* SD_BALANCE_FORK stats */
1066 unsigned int sbf_count;
1067 unsigned int sbf_balanced;
1068 unsigned int sbf_pushed;
1070 /* try_to_wake_up() stats */
1071 unsigned int ttwu_wake_remote;
1072 unsigned int ttwu_move_affine;
1073 unsigned int ttwu_move_balance;
1075 #ifdef CONFIG_SCHED_DEBUG
1079 void *private; /* used during construction */
1080 struct rcu_head rcu; /* used during destruction */
1083 unsigned int span_weight;
1085 * Span of all CPUs in this domain.
1087 * NOTE: this field is variable length. (Allocated dynamically
1088 * by attaching extra space to the end of the structure,
1089 * depending on how many CPUs the kernel has booted up with)
1091 unsigned long span[0];
1094 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1096 return to_cpumask(sd->span);
1099 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1100 struct sched_domain_attr *dattr_new);
1102 /* Allocate an array of sched domains, for partition_sched_domains(). */
1103 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1104 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1106 bool cpus_share_cache(int this_cpu, int that_cpu);
1108 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1109 typedef int (*sched_domain_flags_f)(void);
1111 #define SDTL_OVERLAP 0x01
1114 struct sched_domain **__percpu sd;
1115 struct sched_group **__percpu sg;
1116 struct sched_group_capacity **__percpu sgc;
1119 struct sched_domain_topology_level {
1120 sched_domain_mask_f mask;
1121 sched_domain_flags_f sd_flags;
1124 struct sd_data data;
1125 #ifdef CONFIG_SCHED_DEBUG
1130 extern struct sched_domain_topology_level *sched_domain_topology;
1132 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1133 extern void wake_up_if_idle(int cpu);
1135 #ifdef CONFIG_SCHED_DEBUG
1136 # define SD_INIT_NAME(type) .name = #type
1138 # define SD_INIT_NAME(type)
1141 #else /* CONFIG_SMP */
1143 struct sched_domain_attr;
1146 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1147 struct sched_domain_attr *dattr_new)
1151 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1156 #endif /* !CONFIG_SMP */
1159 struct io_context; /* See blkdev.h */
1162 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1163 extern void prefetch_stack(struct task_struct *t);
1165 static inline void prefetch_stack(struct task_struct *t) { }
1168 struct audit_context; /* See audit.c */
1170 struct pipe_inode_info;
1171 struct uts_namespace;
1173 struct load_weight {
1174 unsigned long weight;
1179 * The load_avg/util_avg accumulates an infinite geometric series.
1180 * 1) load_avg factors the amount of time that a sched_entity is
1181 * runnable on a rq into its weight. For cfs_rq, it is the aggregated
1182 * such weights of all runnable and blocked sched_entities.
1183 * 2) util_avg factors frequency scaling into the amount of time
1184 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1185 * For cfs_rq, it is the aggregated such times of all runnable and
1186 * blocked sched_entities.
1187 * The 64 bit load_sum can:
1188 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1189 * the highest weight (=88761) always runnable, we should not overflow
1190 * 2) for entity, support any load.weight always runnable
1193 u64 last_update_time, load_sum;
1194 u32 util_sum, period_contrib;
1195 unsigned long load_avg, util_avg;
1198 #ifdef CONFIG_SCHEDSTATS
1199 struct sched_statistics {
1209 s64 sum_sleep_runtime;
1216 u64 nr_migrations_cold;
1217 u64 nr_failed_migrations_affine;
1218 u64 nr_failed_migrations_running;
1219 u64 nr_failed_migrations_hot;
1220 u64 nr_forced_migrations;
1223 u64 nr_wakeups_sync;
1224 u64 nr_wakeups_migrate;
1225 u64 nr_wakeups_local;
1226 u64 nr_wakeups_remote;
1227 u64 nr_wakeups_affine;
1228 u64 nr_wakeups_affine_attempts;
1229 u64 nr_wakeups_passive;
1230 u64 nr_wakeups_idle;
1234 struct sched_entity {
1235 struct load_weight load; /* for load-balancing */
1236 struct rb_node run_node;
1237 struct list_head group_node;
1241 u64 sum_exec_runtime;
1243 u64 prev_sum_exec_runtime;
1247 #ifdef CONFIG_SCHEDSTATS
1248 struct sched_statistics statistics;
1251 #ifdef CONFIG_FAIR_GROUP_SCHED
1253 struct sched_entity *parent;
1254 /* rq on which this entity is (to be) queued: */
1255 struct cfs_rq *cfs_rq;
1256 /* rq "owned" by this entity/group: */
1257 struct cfs_rq *my_q;
1261 /* Per entity load average tracking */
1262 struct sched_avg avg;
1266 struct sched_rt_entity {
1267 struct list_head run_list;
1268 unsigned long timeout;
1269 unsigned long watchdog_stamp;
1270 unsigned int time_slice;
1272 struct sched_rt_entity *back;
1273 #ifdef CONFIG_RT_GROUP_SCHED
1274 struct sched_rt_entity *parent;
1275 /* rq on which this entity is (to be) queued: */
1276 struct rt_rq *rt_rq;
1277 /* rq "owned" by this entity/group: */
1282 struct sched_dl_entity {
1283 struct rb_node rb_node;
1286 * Original scheduling parameters. Copied here from sched_attr
1287 * during sched_setattr(), they will remain the same until
1288 * the next sched_setattr().
1290 u64 dl_runtime; /* maximum runtime for each instance */
1291 u64 dl_deadline; /* relative deadline of each instance */
1292 u64 dl_period; /* separation of two instances (period) */
1293 u64 dl_bw; /* dl_runtime / dl_deadline */
1296 * Actual scheduling parameters. Initialized with the values above,
1297 * they are continously updated during task execution. Note that
1298 * the remaining runtime could be < 0 in case we are in overrun.
1300 s64 runtime; /* remaining runtime for this instance */
1301 u64 deadline; /* absolute deadline for this instance */
1302 unsigned int flags; /* specifying the scheduler behaviour */
1307 * @dl_throttled tells if we exhausted the runtime. If so, the
1308 * task has to wait for a replenishment to be performed at the
1309 * next firing of dl_timer.
1311 * @dl_new tells if a new instance arrived. If so we must
1312 * start executing it with full runtime and reset its absolute
1315 * @dl_boosted tells if we are boosted due to DI. If so we are
1316 * outside bandwidth enforcement mechanism (but only until we
1317 * exit the critical section);
1319 * @dl_yielded tells if task gave up the cpu before consuming
1320 * all its available runtime during the last job.
1322 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1325 * Bandwidth enforcement timer. Each -deadline task has its
1326 * own bandwidth to be enforced, thus we need one timer per task.
1328 struct hrtimer dl_timer;
1340 enum perf_event_task_context {
1341 perf_invalid_context = -1,
1342 perf_hw_context = 0,
1344 perf_nr_task_contexts,
1347 struct task_struct {
1348 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1351 unsigned int flags; /* per process flags, defined below */
1352 unsigned int ptrace;
1355 struct llist_node wake_entry;
1357 unsigned int wakee_flips;
1358 unsigned long wakee_flip_decay_ts;
1359 struct task_struct *last_wakee;
1365 int prio, static_prio, normal_prio;
1366 unsigned int rt_priority;
1367 const struct sched_class *sched_class;
1368 struct sched_entity se;
1369 struct sched_rt_entity rt;
1370 #ifdef CONFIG_CGROUP_SCHED
1371 struct task_group *sched_task_group;
1373 struct sched_dl_entity dl;
1375 #ifdef CONFIG_PREEMPT_NOTIFIERS
1376 /* list of struct preempt_notifier: */
1377 struct hlist_head preempt_notifiers;
1380 #ifdef CONFIG_BLK_DEV_IO_TRACE
1381 unsigned int btrace_seq;
1384 unsigned int policy;
1385 int nr_cpus_allowed;
1386 cpumask_t cpus_allowed;
1388 #ifdef CONFIG_PREEMPT_RCU
1389 int rcu_read_lock_nesting;
1390 union rcu_special rcu_read_unlock_special;
1391 struct list_head rcu_node_entry;
1392 struct rcu_node *rcu_blocked_node;
1393 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1394 #ifdef CONFIG_TASKS_RCU
1395 unsigned long rcu_tasks_nvcsw;
1396 bool rcu_tasks_holdout;
1397 struct list_head rcu_tasks_holdout_list;
1398 int rcu_tasks_idle_cpu;
1399 #endif /* #ifdef CONFIG_TASKS_RCU */
1401 #ifdef CONFIG_SCHED_INFO
1402 struct sched_info sched_info;
1405 struct list_head tasks;
1407 struct plist_node pushable_tasks;
1408 struct rb_node pushable_dl_tasks;
1411 struct mm_struct *mm, *active_mm;
1412 /* per-thread vma caching */
1413 u32 vmacache_seqnum;
1414 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1415 #if defined(SPLIT_RSS_COUNTING)
1416 struct task_rss_stat rss_stat;
1420 int exit_code, exit_signal;
1421 int pdeath_signal; /* The signal sent when the parent dies */
1422 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1424 /* Used for emulating ABI behavior of previous Linux versions */
1425 unsigned int personality;
1427 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1429 unsigned in_iowait:1;
1431 /* Revert to default priority/policy when forking */
1432 unsigned sched_reset_on_fork:1;
1433 unsigned sched_contributes_to_load:1;
1434 unsigned sched_migrated:1;
1436 #ifdef CONFIG_MEMCG_KMEM
1437 unsigned memcg_kmem_skip_account:1;
1439 #ifdef CONFIG_COMPAT_BRK
1440 unsigned brk_randomized:1;
1443 unsigned long atomic_flags; /* Flags needing atomic access. */
1445 struct restart_block restart_block;
1450 #ifdef CONFIG_CC_STACKPROTECTOR
1451 /* Canary value for the -fstack-protector gcc feature */
1452 unsigned long stack_canary;
1455 * pointers to (original) parent process, youngest child, younger sibling,
1456 * older sibling, respectively. (p->father can be replaced with
1457 * p->real_parent->pid)
1459 struct task_struct __rcu *real_parent; /* real parent process */
1460 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1462 * children/sibling forms the list of my natural children
1464 struct list_head children; /* list of my children */
1465 struct list_head sibling; /* linkage in my parent's children list */
1466 struct task_struct *group_leader; /* threadgroup leader */
1469 * ptraced is the list of tasks this task is using ptrace on.
1470 * This includes both natural children and PTRACE_ATTACH targets.
1471 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1473 struct list_head ptraced;
1474 struct list_head ptrace_entry;
1476 /* PID/PID hash table linkage. */
1477 struct pid_link pids[PIDTYPE_MAX];
1478 struct list_head thread_group;
1479 struct list_head thread_node;
1481 struct completion *vfork_done; /* for vfork() */
1482 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1483 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1485 cputime_t utime, stime, utimescaled, stimescaled;
1487 struct prev_cputime prev_cputime;
1488 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1489 seqlock_t vtime_seqlock;
1490 unsigned long long vtime_snap;
1495 } vtime_snap_whence;
1497 unsigned long nvcsw, nivcsw; /* context switch counts */
1498 u64 start_time; /* monotonic time in nsec */
1499 u64 real_start_time; /* boot based time in nsec */
1500 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1501 unsigned long min_flt, maj_flt;
1503 struct task_cputime cputime_expires;
1504 struct list_head cpu_timers[3];
1506 /* process credentials */
1507 const struct cred __rcu *real_cred; /* objective and real subjective task
1508 * credentials (COW) */
1509 const struct cred __rcu *cred; /* effective (overridable) subjective task
1510 * credentials (COW) */
1511 char comm[TASK_COMM_LEN]; /* executable name excluding path
1512 - access with [gs]et_task_comm (which lock
1513 it with task_lock())
1514 - initialized normally by setup_new_exec */
1515 /* file system info */
1516 struct nameidata *nameidata;
1517 #ifdef CONFIG_SYSVIPC
1519 struct sysv_sem sysvsem;
1520 struct sysv_shm sysvshm;
1522 #ifdef CONFIG_DETECT_HUNG_TASK
1523 /* hung task detection */
1524 unsigned long last_switch_count;
1526 /* filesystem information */
1527 struct fs_struct *fs;
1528 /* open file information */
1529 struct files_struct *files;
1531 struct nsproxy *nsproxy;
1532 /* signal handlers */
1533 struct signal_struct *signal;
1534 struct sighand_struct *sighand;
1536 sigset_t blocked, real_blocked;
1537 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1538 struct sigpending pending;
1540 unsigned long sas_ss_sp;
1542 int (*notifier)(void *priv);
1543 void *notifier_data;
1544 sigset_t *notifier_mask;
1545 struct callback_head *task_works;
1547 struct audit_context *audit_context;
1548 #ifdef CONFIG_AUDITSYSCALL
1550 unsigned int sessionid;
1552 struct seccomp seccomp;
1554 /* Thread group tracking */
1557 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1559 spinlock_t alloc_lock;
1561 /* Protection of the PI data structures: */
1562 raw_spinlock_t pi_lock;
1564 struct wake_q_node wake_q;
1566 #ifdef CONFIG_RT_MUTEXES
1567 /* PI waiters blocked on a rt_mutex held by this task */
1568 struct rb_root pi_waiters;
1569 struct rb_node *pi_waiters_leftmost;
1570 /* Deadlock detection and priority inheritance handling */
1571 struct rt_mutex_waiter *pi_blocked_on;
1574 #ifdef CONFIG_DEBUG_MUTEXES
1575 /* mutex deadlock detection */
1576 struct mutex_waiter *blocked_on;
1578 #ifdef CONFIG_TRACE_IRQFLAGS
1579 unsigned int irq_events;
1580 unsigned long hardirq_enable_ip;
1581 unsigned long hardirq_disable_ip;
1582 unsigned int hardirq_enable_event;
1583 unsigned int hardirq_disable_event;
1584 int hardirqs_enabled;
1585 int hardirq_context;
1586 unsigned long softirq_disable_ip;
1587 unsigned long softirq_enable_ip;
1588 unsigned int softirq_disable_event;
1589 unsigned int softirq_enable_event;
1590 int softirqs_enabled;
1591 int softirq_context;
1593 #ifdef CONFIG_LOCKDEP
1594 # define MAX_LOCK_DEPTH 48UL
1597 unsigned int lockdep_recursion;
1598 struct held_lock held_locks[MAX_LOCK_DEPTH];
1599 gfp_t lockdep_reclaim_gfp;
1602 /* journalling filesystem info */
1605 /* stacked block device info */
1606 struct bio_list *bio_list;
1609 /* stack plugging */
1610 struct blk_plug *plug;
1614 struct reclaim_state *reclaim_state;
1616 struct backing_dev_info *backing_dev_info;
1618 struct io_context *io_context;
1620 unsigned long ptrace_message;
1621 siginfo_t *last_siginfo; /* For ptrace use. */
1622 struct task_io_accounting ioac;
1623 #if defined(CONFIG_TASK_XACCT)
1624 u64 acct_rss_mem1; /* accumulated rss usage */
1625 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1626 cputime_t acct_timexpd; /* stime + utime since last update */
1628 #ifdef CONFIG_CPUSETS
1629 nodemask_t mems_allowed; /* Protected by alloc_lock */
1630 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1631 int cpuset_mem_spread_rotor;
1632 int cpuset_slab_spread_rotor;
1634 #ifdef CONFIG_CGROUPS
1635 /* Control Group info protected by css_set_lock */
1636 struct css_set __rcu *cgroups;
1637 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1638 struct list_head cg_list;
1641 struct robust_list_head __user *robust_list;
1642 #ifdef CONFIG_COMPAT
1643 struct compat_robust_list_head __user *compat_robust_list;
1645 struct list_head pi_state_list;
1646 struct futex_pi_state *pi_state_cache;
1648 #ifdef CONFIG_PERF_EVENTS
1649 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1650 struct mutex perf_event_mutex;
1651 struct list_head perf_event_list;
1653 #ifdef CONFIG_DEBUG_PREEMPT
1654 unsigned long preempt_disable_ip;
1657 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1659 short pref_node_fork;
1661 #ifdef CONFIG_NUMA_BALANCING
1663 unsigned int numa_scan_period;
1664 unsigned int numa_scan_period_max;
1665 int numa_preferred_nid;
1666 unsigned long numa_migrate_retry;
1667 u64 node_stamp; /* migration stamp */
1668 u64 last_task_numa_placement;
1669 u64 last_sum_exec_runtime;
1670 struct callback_head numa_work;
1672 struct list_head numa_entry;
1673 struct numa_group *numa_group;
1676 * numa_faults is an array split into four regions:
1677 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1678 * in this precise order.
1680 * faults_memory: Exponential decaying average of faults on a per-node
1681 * basis. Scheduling placement decisions are made based on these
1682 * counts. The values remain static for the duration of a PTE scan.
1683 * faults_cpu: Track the nodes the process was running on when a NUMA
1684 * hinting fault was incurred.
1685 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1686 * during the current scan window. When the scan completes, the counts
1687 * in faults_memory and faults_cpu decay and these values are copied.
1689 unsigned long *numa_faults;
1690 unsigned long total_numa_faults;
1693 * numa_faults_locality tracks if faults recorded during the last
1694 * scan window were remote/local or failed to migrate. The task scan
1695 * period is adapted based on the locality of the faults with different
1696 * weights depending on whether they were shared or private faults
1698 unsigned long numa_faults_locality[3];
1700 unsigned long numa_pages_migrated;
1701 #endif /* CONFIG_NUMA_BALANCING */
1703 struct rcu_head rcu;
1706 * cache last used pipe for splice
1708 struct pipe_inode_info *splice_pipe;
1710 struct page_frag task_frag;
1712 #ifdef CONFIG_TASK_DELAY_ACCT
1713 struct task_delay_info *delays;
1715 #ifdef CONFIG_FAULT_INJECTION
1719 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1720 * balance_dirty_pages() for some dirty throttling pause
1723 int nr_dirtied_pause;
1724 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1726 #ifdef CONFIG_LATENCYTOP
1727 int latency_record_count;
1728 struct latency_record latency_record[LT_SAVECOUNT];
1731 * time slack values; these are used to round up poll() and
1732 * select() etc timeout values. These are in nanoseconds.
1734 unsigned long timer_slack_ns;
1735 unsigned long default_timer_slack_ns;
1738 unsigned int kasan_depth;
1740 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1741 /* Index of current stored address in ret_stack */
1743 /* Stack of return addresses for return function tracing */
1744 struct ftrace_ret_stack *ret_stack;
1745 /* time stamp for last schedule */
1746 unsigned long long ftrace_timestamp;
1748 * Number of functions that haven't been traced
1749 * because of depth overrun.
1751 atomic_t trace_overrun;
1752 /* Pause for the tracing */
1753 atomic_t tracing_graph_pause;
1755 #ifdef CONFIG_TRACING
1756 /* state flags for use by tracers */
1757 unsigned long trace;
1758 /* bitmask and counter of trace recursion */
1759 unsigned long trace_recursion;
1760 #endif /* CONFIG_TRACING */
1762 struct memcg_oom_info {
1763 struct mem_cgroup *memcg;
1766 unsigned int may_oom:1;
1769 #ifdef CONFIG_UPROBES
1770 struct uprobe_task *utask;
1772 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1773 unsigned int sequential_io;
1774 unsigned int sequential_io_avg;
1776 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1777 unsigned long task_state_change;
1779 int pagefault_disabled;
1780 /* CPU-specific state of this task */
1781 struct thread_struct thread;
1783 * WARNING: on x86, 'thread_struct' contains a variable-sized
1784 * structure. It *MUST* be at the end of 'task_struct'.
1786 * Do not put anything below here!
1790 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1791 extern int arch_task_struct_size __read_mostly;
1793 # define arch_task_struct_size (sizeof(struct task_struct))
1796 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1797 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1799 #define TNF_MIGRATED 0x01
1800 #define TNF_NO_GROUP 0x02
1801 #define TNF_SHARED 0x04
1802 #define TNF_FAULT_LOCAL 0x08
1803 #define TNF_MIGRATE_FAIL 0x10
1805 #ifdef CONFIG_NUMA_BALANCING
1806 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1807 extern pid_t task_numa_group_id(struct task_struct *p);
1808 extern void set_numabalancing_state(bool enabled);
1809 extern void task_numa_free(struct task_struct *p);
1810 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1811 int src_nid, int dst_cpu);
1813 static inline void task_numa_fault(int last_node, int node, int pages,
1817 static inline pid_t task_numa_group_id(struct task_struct *p)
1821 static inline void set_numabalancing_state(bool enabled)
1824 static inline void task_numa_free(struct task_struct *p)
1827 static inline bool should_numa_migrate_memory(struct task_struct *p,
1828 struct page *page, int src_nid, int dst_cpu)
1834 static inline struct pid *task_pid(struct task_struct *task)
1836 return task->pids[PIDTYPE_PID].pid;
1839 static inline struct pid *task_tgid(struct task_struct *task)
1841 return task->group_leader->pids[PIDTYPE_PID].pid;
1845 * Without tasklist or rcu lock it is not safe to dereference
1846 * the result of task_pgrp/task_session even if task == current,
1847 * we can race with another thread doing sys_setsid/sys_setpgid.
1849 static inline struct pid *task_pgrp(struct task_struct *task)
1851 return task->group_leader->pids[PIDTYPE_PGID].pid;
1854 static inline struct pid *task_session(struct task_struct *task)
1856 return task->group_leader->pids[PIDTYPE_SID].pid;
1859 struct pid_namespace;
1862 * the helpers to get the task's different pids as they are seen
1863 * from various namespaces
1865 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1866 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1868 * task_xid_nr_ns() : id seen from the ns specified;
1870 * set_task_vxid() : assigns a virtual id to a task;
1872 * see also pid_nr() etc in include/linux/pid.h
1874 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1875 struct pid_namespace *ns);
1877 static inline pid_t task_pid_nr(struct task_struct *tsk)
1882 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1883 struct pid_namespace *ns)
1885 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1888 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1890 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1894 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1899 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1901 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1903 return pid_vnr(task_tgid(tsk));
1907 static inline int pid_alive(const struct task_struct *p);
1908 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1914 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1920 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1922 return task_ppid_nr_ns(tsk, &init_pid_ns);
1925 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1926 struct pid_namespace *ns)
1928 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1931 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1933 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1937 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1938 struct pid_namespace *ns)
1940 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1943 static inline pid_t task_session_vnr(struct task_struct *tsk)
1945 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1948 /* obsolete, do not use */
1949 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1951 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1955 * pid_alive - check that a task structure is not stale
1956 * @p: Task structure to be checked.
1958 * Test if a process is not yet dead (at most zombie state)
1959 * If pid_alive fails, then pointers within the task structure
1960 * can be stale and must not be dereferenced.
1962 * Return: 1 if the process is alive. 0 otherwise.
1964 static inline int pid_alive(const struct task_struct *p)
1966 return p->pids[PIDTYPE_PID].pid != NULL;
1970 * is_global_init - check if a task structure is init
1971 * @tsk: Task structure to be checked.
1973 * Check if a task structure is the first user space task the kernel created.
1975 * Return: 1 if the task structure is init. 0 otherwise.
1977 static inline int is_global_init(struct task_struct *tsk)
1979 return tsk->pid == 1;
1982 extern struct pid *cad_pid;
1984 extern void free_task(struct task_struct *tsk);
1985 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1987 extern void __put_task_struct(struct task_struct *t);
1989 static inline void put_task_struct(struct task_struct *t)
1991 if (atomic_dec_and_test(&t->usage))
1992 __put_task_struct(t);
1995 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1996 extern void task_cputime(struct task_struct *t,
1997 cputime_t *utime, cputime_t *stime);
1998 extern void task_cputime_scaled(struct task_struct *t,
1999 cputime_t *utimescaled, cputime_t *stimescaled);
2000 extern cputime_t task_gtime(struct task_struct *t);
2002 static inline void task_cputime(struct task_struct *t,
2003 cputime_t *utime, cputime_t *stime)
2011 static inline void task_cputime_scaled(struct task_struct *t,
2012 cputime_t *utimescaled,
2013 cputime_t *stimescaled)
2016 *utimescaled = t->utimescaled;
2018 *stimescaled = t->stimescaled;
2021 static inline cputime_t task_gtime(struct task_struct *t)
2026 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2027 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2032 #define PF_EXITING 0x00000004 /* getting shut down */
2033 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2034 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2035 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2036 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2037 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2038 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2039 #define PF_DUMPCORE 0x00000200 /* dumped core */
2040 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2041 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2042 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2043 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2044 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2045 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2046 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2047 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2048 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2049 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2050 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2051 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2052 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2053 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2054 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2055 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2056 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2057 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2058 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2061 * Only the _current_ task can read/write to tsk->flags, but other
2062 * tasks can access tsk->flags in readonly mode for example
2063 * with tsk_used_math (like during threaded core dumping).
2064 * There is however an exception to this rule during ptrace
2065 * or during fork: the ptracer task is allowed to write to the
2066 * child->flags of its traced child (same goes for fork, the parent
2067 * can write to the child->flags), because we're guaranteed the
2068 * child is not running and in turn not changing child->flags
2069 * at the same time the parent does it.
2071 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2072 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2073 #define clear_used_math() clear_stopped_child_used_math(current)
2074 #define set_used_math() set_stopped_child_used_math(current)
2075 #define conditional_stopped_child_used_math(condition, child) \
2076 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2077 #define conditional_used_math(condition) \
2078 conditional_stopped_child_used_math(condition, current)
2079 #define copy_to_stopped_child_used_math(child) \
2080 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2081 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2082 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2083 #define used_math() tsk_used_math(current)
2085 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2086 * __GFP_FS is also cleared as it implies __GFP_IO.
2088 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2090 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2091 flags &= ~(__GFP_IO | __GFP_FS);
2095 static inline unsigned int memalloc_noio_save(void)
2097 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2098 current->flags |= PF_MEMALLOC_NOIO;
2102 static inline void memalloc_noio_restore(unsigned int flags)
2104 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2107 /* Per-process atomic flags. */
2108 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2109 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2110 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2113 #define TASK_PFA_TEST(name, func) \
2114 static inline bool task_##func(struct task_struct *p) \
2115 { return test_bit(PFA_##name, &p->atomic_flags); }
2116 #define TASK_PFA_SET(name, func) \
2117 static inline void task_set_##func(struct task_struct *p) \
2118 { set_bit(PFA_##name, &p->atomic_flags); }
2119 #define TASK_PFA_CLEAR(name, func) \
2120 static inline void task_clear_##func(struct task_struct *p) \
2121 { clear_bit(PFA_##name, &p->atomic_flags); }
2123 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2124 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2126 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2127 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2128 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2130 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2131 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2132 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2135 * task->jobctl flags
2137 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2139 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2140 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2141 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2142 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2143 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2144 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2145 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2147 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2148 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2149 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2150 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2151 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2152 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2153 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2155 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2156 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2158 extern bool task_set_jobctl_pending(struct task_struct *task,
2159 unsigned long mask);
2160 extern void task_clear_jobctl_trapping(struct task_struct *task);
2161 extern void task_clear_jobctl_pending(struct task_struct *task,
2162 unsigned long mask);
2164 static inline void rcu_copy_process(struct task_struct *p)
2166 #ifdef CONFIG_PREEMPT_RCU
2167 p->rcu_read_lock_nesting = 0;
2168 p->rcu_read_unlock_special.s = 0;
2169 p->rcu_blocked_node = NULL;
2170 INIT_LIST_HEAD(&p->rcu_node_entry);
2171 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2172 #ifdef CONFIG_TASKS_RCU
2173 p->rcu_tasks_holdout = false;
2174 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2175 p->rcu_tasks_idle_cpu = -1;
2176 #endif /* #ifdef CONFIG_TASKS_RCU */
2179 static inline void tsk_restore_flags(struct task_struct *task,
2180 unsigned long orig_flags, unsigned long flags)
2182 task->flags &= ~flags;
2183 task->flags |= orig_flags & flags;
2186 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2187 const struct cpumask *trial);
2188 extern int task_can_attach(struct task_struct *p,
2189 const struct cpumask *cs_cpus_allowed);
2191 extern void do_set_cpus_allowed(struct task_struct *p,
2192 const struct cpumask *new_mask);
2194 extern int set_cpus_allowed_ptr(struct task_struct *p,
2195 const struct cpumask *new_mask);
2197 static inline void do_set_cpus_allowed(struct task_struct *p,
2198 const struct cpumask *new_mask)
2201 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2202 const struct cpumask *new_mask)
2204 if (!cpumask_test_cpu(0, new_mask))
2210 #ifdef CONFIG_NO_HZ_COMMON
2211 void calc_load_enter_idle(void);
2212 void calc_load_exit_idle(void);
2214 static inline void calc_load_enter_idle(void) { }
2215 static inline void calc_load_exit_idle(void) { }
2216 #endif /* CONFIG_NO_HZ_COMMON */
2219 * Do not use outside of architecture code which knows its limitations.
2221 * sched_clock() has no promise of monotonicity or bounded drift between
2222 * CPUs, use (which you should not) requires disabling IRQs.
2224 * Please use one of the three interfaces below.
2226 extern unsigned long long notrace sched_clock(void);
2228 * See the comment in kernel/sched/clock.c
2230 extern u64 cpu_clock(int cpu);
2231 extern u64 local_clock(void);
2232 extern u64 running_clock(void);
2233 extern u64 sched_clock_cpu(int cpu);
2236 extern void sched_clock_init(void);
2238 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2239 static inline void sched_clock_tick(void)
2243 static inline void sched_clock_idle_sleep_event(void)
2247 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2252 * Architectures can set this to 1 if they have specified
2253 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2254 * but then during bootup it turns out that sched_clock()
2255 * is reliable after all:
2257 extern int sched_clock_stable(void);
2258 extern void set_sched_clock_stable(void);
2259 extern void clear_sched_clock_stable(void);
2261 extern void sched_clock_tick(void);
2262 extern void sched_clock_idle_sleep_event(void);
2263 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2266 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2268 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2269 * The reason for this explicit opt-in is not to have perf penalty with
2270 * slow sched_clocks.
2272 extern void enable_sched_clock_irqtime(void);
2273 extern void disable_sched_clock_irqtime(void);
2275 static inline void enable_sched_clock_irqtime(void) {}
2276 static inline void disable_sched_clock_irqtime(void) {}
2279 extern unsigned long long
2280 task_sched_runtime(struct task_struct *task);
2282 /* sched_exec is called by processes performing an exec */
2284 extern void sched_exec(void);
2286 #define sched_exec() {}
2289 extern void sched_clock_idle_sleep_event(void);
2290 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2292 #ifdef CONFIG_HOTPLUG_CPU
2293 extern void idle_task_exit(void);
2295 static inline void idle_task_exit(void) {}
2298 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2299 extern void wake_up_nohz_cpu(int cpu);
2301 static inline void wake_up_nohz_cpu(int cpu) { }
2304 #ifdef CONFIG_NO_HZ_FULL
2305 extern bool sched_can_stop_tick(void);
2306 extern u64 scheduler_tick_max_deferment(void);
2308 static inline bool sched_can_stop_tick(void) { return false; }
2311 #ifdef CONFIG_SCHED_AUTOGROUP
2312 extern void sched_autogroup_create_attach(struct task_struct *p);
2313 extern void sched_autogroup_detach(struct task_struct *p);
2314 extern void sched_autogroup_fork(struct signal_struct *sig);
2315 extern void sched_autogroup_exit(struct signal_struct *sig);
2316 #ifdef CONFIG_PROC_FS
2317 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2318 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2321 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2322 static inline void sched_autogroup_detach(struct task_struct *p) { }
2323 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2324 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2327 extern int yield_to(struct task_struct *p, bool preempt);
2328 extern void set_user_nice(struct task_struct *p, long nice);
2329 extern int task_prio(const struct task_struct *p);
2331 * task_nice - return the nice value of a given task.
2332 * @p: the task in question.
2334 * Return: The nice value [ -20 ... 0 ... 19 ].
2336 static inline int task_nice(const struct task_struct *p)
2338 return PRIO_TO_NICE((p)->static_prio);
2340 extern int can_nice(const struct task_struct *p, const int nice);
2341 extern int task_curr(const struct task_struct *p);
2342 extern int idle_cpu(int cpu);
2343 extern int sched_setscheduler(struct task_struct *, int,
2344 const struct sched_param *);
2345 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2346 const struct sched_param *);
2347 extern int sched_setattr(struct task_struct *,
2348 const struct sched_attr *);
2349 extern struct task_struct *idle_task(int cpu);
2351 * is_idle_task - is the specified task an idle task?
2352 * @p: the task in question.
2354 * Return: 1 if @p is an idle task. 0 otherwise.
2356 static inline bool is_idle_task(const struct task_struct *p)
2360 extern struct task_struct *curr_task(int cpu);
2361 extern void set_curr_task(int cpu, struct task_struct *p);
2365 union thread_union {
2366 struct thread_info thread_info;
2367 unsigned long stack[THREAD_SIZE/sizeof(long)];
2370 #ifndef __HAVE_ARCH_KSTACK_END
2371 static inline int kstack_end(void *addr)
2373 /* Reliable end of stack detection:
2374 * Some APM bios versions misalign the stack
2376 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2380 extern union thread_union init_thread_union;
2381 extern struct task_struct init_task;
2383 extern struct mm_struct init_mm;
2385 extern struct pid_namespace init_pid_ns;
2388 * find a task by one of its numerical ids
2390 * find_task_by_pid_ns():
2391 * finds a task by its pid in the specified namespace
2392 * find_task_by_vpid():
2393 * finds a task by its virtual pid
2395 * see also find_vpid() etc in include/linux/pid.h
2398 extern struct task_struct *find_task_by_vpid(pid_t nr);
2399 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2400 struct pid_namespace *ns);
2402 /* per-UID process charging. */
2403 extern struct user_struct * alloc_uid(kuid_t);
2404 static inline struct user_struct *get_uid(struct user_struct *u)
2406 atomic_inc(&u->__count);
2409 extern void free_uid(struct user_struct *);
2411 #include <asm/current.h>
2413 extern void xtime_update(unsigned long ticks);
2415 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2416 extern int wake_up_process(struct task_struct *tsk);
2417 extern void wake_up_new_task(struct task_struct *tsk);
2419 extern void kick_process(struct task_struct *tsk);
2421 static inline void kick_process(struct task_struct *tsk) { }
2423 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2424 extern void sched_dead(struct task_struct *p);
2426 extern void proc_caches_init(void);
2427 extern void flush_signals(struct task_struct *);
2428 extern void ignore_signals(struct task_struct *);
2429 extern void flush_signal_handlers(struct task_struct *, int force_default);
2430 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2432 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2434 unsigned long flags;
2437 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2438 ret = dequeue_signal(tsk, mask, info);
2439 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2444 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2446 extern void unblock_all_signals(void);
2447 extern void release_task(struct task_struct * p);
2448 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2449 extern int force_sigsegv(int, struct task_struct *);
2450 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2451 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2452 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2453 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2454 const struct cred *, u32);
2455 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2456 extern int kill_pid(struct pid *pid, int sig, int priv);
2457 extern int kill_proc_info(int, struct siginfo *, pid_t);
2458 extern __must_check bool do_notify_parent(struct task_struct *, int);
2459 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2460 extern void force_sig(int, struct task_struct *);
2461 extern int send_sig(int, struct task_struct *, int);
2462 extern int zap_other_threads(struct task_struct *p);
2463 extern struct sigqueue *sigqueue_alloc(void);
2464 extern void sigqueue_free(struct sigqueue *);
2465 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2466 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2468 static inline void restore_saved_sigmask(void)
2470 if (test_and_clear_restore_sigmask())
2471 __set_current_blocked(¤t->saved_sigmask);
2474 static inline sigset_t *sigmask_to_save(void)
2476 sigset_t *res = ¤t->blocked;
2477 if (unlikely(test_restore_sigmask()))
2478 res = ¤t->saved_sigmask;
2482 static inline int kill_cad_pid(int sig, int priv)
2484 return kill_pid(cad_pid, sig, priv);
2487 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2488 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2489 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2490 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2493 * True if we are on the alternate signal stack.
2495 static inline int on_sig_stack(unsigned long sp)
2497 #ifdef CONFIG_STACK_GROWSUP
2498 return sp >= current->sas_ss_sp &&
2499 sp - current->sas_ss_sp < current->sas_ss_size;
2501 return sp > current->sas_ss_sp &&
2502 sp - current->sas_ss_sp <= current->sas_ss_size;
2506 static inline int sas_ss_flags(unsigned long sp)
2508 if (!current->sas_ss_size)
2511 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2514 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2516 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2517 #ifdef CONFIG_STACK_GROWSUP
2518 return current->sas_ss_sp;
2520 return current->sas_ss_sp + current->sas_ss_size;
2526 * Routines for handling mm_structs
2528 extern struct mm_struct * mm_alloc(void);
2530 /* mmdrop drops the mm and the page tables */
2531 extern void __mmdrop(struct mm_struct *);
2532 static inline void mmdrop(struct mm_struct * mm)
2534 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2538 /* mmput gets rid of the mappings and all user-space */
2539 extern void mmput(struct mm_struct *);
2540 /* Grab a reference to a task's mm, if it is not already going away */
2541 extern struct mm_struct *get_task_mm(struct task_struct *task);
2543 * Grab a reference to a task's mm, if it is not already going away
2544 * and ptrace_may_access with the mode parameter passed to it
2547 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2548 /* Remove the current tasks stale references to the old mm_struct */
2549 extern void mm_release(struct task_struct *, struct mm_struct *);
2551 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2552 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2553 struct task_struct *, unsigned long);
2555 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2556 struct task_struct *);
2558 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2559 * via pt_regs, so ignore the tls argument passed via C. */
2560 static inline int copy_thread_tls(
2561 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2562 struct task_struct *p, unsigned long tls)
2564 return copy_thread(clone_flags, sp, arg, p);
2567 extern void flush_thread(void);
2568 extern void exit_thread(void);
2570 extern void exit_files(struct task_struct *);
2571 extern void __cleanup_sighand(struct sighand_struct *);
2573 extern void exit_itimers(struct signal_struct *);
2574 extern void flush_itimer_signals(void);
2576 extern void do_group_exit(int);
2578 extern int do_execve(struct filename *,
2579 const char __user * const __user *,
2580 const char __user * const __user *);
2581 extern int do_execveat(int, struct filename *,
2582 const char __user * const __user *,
2583 const char __user * const __user *,
2585 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2586 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2587 struct task_struct *fork_idle(int);
2588 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2590 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2591 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2593 __set_task_comm(tsk, from, false);
2595 extern char *get_task_comm(char *to, struct task_struct *tsk);
2598 void scheduler_ipi(void);
2599 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2601 static inline void scheduler_ipi(void) { }
2602 static inline unsigned long wait_task_inactive(struct task_struct *p,
2609 #define tasklist_empty() \
2610 list_empty(&init_task.tasks)
2612 #define next_task(p) \
2613 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2615 #define for_each_process(p) \
2616 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2618 extern bool current_is_single_threaded(void);
2621 * Careful: do_each_thread/while_each_thread is a double loop so
2622 * 'break' will not work as expected - use goto instead.
2624 #define do_each_thread(g, t) \
2625 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2627 #define while_each_thread(g, t) \
2628 while ((t = next_thread(t)) != g)
2630 #define __for_each_thread(signal, t) \
2631 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2633 #define for_each_thread(p, t) \
2634 __for_each_thread((p)->signal, t)
2636 /* Careful: this is a double loop, 'break' won't work as expected. */
2637 #define for_each_process_thread(p, t) \
2638 for_each_process(p) for_each_thread(p, t)
2640 static inline int get_nr_threads(struct task_struct *tsk)
2642 return tsk->signal->nr_threads;
2645 static inline bool thread_group_leader(struct task_struct *p)
2647 return p->exit_signal >= 0;
2650 /* Do to the insanities of de_thread it is possible for a process
2651 * to have the pid of the thread group leader without actually being
2652 * the thread group leader. For iteration through the pids in proc
2653 * all we care about is that we have a task with the appropriate
2654 * pid, we don't actually care if we have the right task.
2656 static inline bool has_group_leader_pid(struct task_struct *p)
2658 return task_pid(p) == p->signal->leader_pid;
2662 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2664 return p1->signal == p2->signal;
2667 static inline struct task_struct *next_thread(const struct task_struct *p)
2669 return list_entry_rcu(p->thread_group.next,
2670 struct task_struct, thread_group);
2673 static inline int thread_group_empty(struct task_struct *p)
2675 return list_empty(&p->thread_group);
2678 #define delay_group_leader(p) \
2679 (thread_group_leader(p) && !thread_group_empty(p))
2682 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2683 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2684 * pins the final release of task.io_context. Also protects ->cpuset and
2685 * ->cgroup.subsys[]. And ->vfork_done.
2687 * Nests both inside and outside of read_lock(&tasklist_lock).
2688 * It must not be nested with write_lock_irq(&tasklist_lock),
2689 * neither inside nor outside.
2691 static inline void task_lock(struct task_struct *p)
2693 spin_lock(&p->alloc_lock);
2696 static inline void task_unlock(struct task_struct *p)
2698 spin_unlock(&p->alloc_lock);
2701 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2702 unsigned long *flags);
2704 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2705 unsigned long *flags)
2707 struct sighand_struct *ret;
2709 ret = __lock_task_sighand(tsk, flags);
2710 (void)__cond_lock(&tsk->sighand->siglock, ret);
2714 static inline void unlock_task_sighand(struct task_struct *tsk,
2715 unsigned long *flags)
2717 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2721 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2722 * @tsk: task causing the changes
2724 * All operations which modify a threadgroup - a new thread joining the
2725 * group, death of a member thread (the assertion of PF_EXITING) and
2726 * exec(2) dethreading the process and replacing the leader - are wrapped
2727 * by threadgroup_change_{begin|end}(). This is to provide a place which
2728 * subsystems needing threadgroup stability can hook into for
2731 static inline void threadgroup_change_begin(struct task_struct *tsk)
2734 cgroup_threadgroup_change_begin(tsk);
2738 * threadgroup_change_end - mark the end of changes to a threadgroup
2739 * @tsk: task causing the changes
2741 * See threadgroup_change_begin().
2743 static inline void threadgroup_change_end(struct task_struct *tsk)
2745 cgroup_threadgroup_change_end(tsk);
2748 #ifndef __HAVE_THREAD_FUNCTIONS
2750 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2751 #define task_stack_page(task) ((task)->stack)
2753 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2755 *task_thread_info(p) = *task_thread_info(org);
2756 task_thread_info(p)->task = p;
2760 * Return the address of the last usable long on the stack.
2762 * When the stack grows down, this is just above the thread
2763 * info struct. Going any lower will corrupt the threadinfo.
2765 * When the stack grows up, this is the highest address.
2766 * Beyond that position, we corrupt data on the next page.
2768 static inline unsigned long *end_of_stack(struct task_struct *p)
2770 #ifdef CONFIG_STACK_GROWSUP
2771 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2773 return (unsigned long *)(task_thread_info(p) + 1);
2778 #define task_stack_end_corrupted(task) \
2779 (*(end_of_stack(task)) != STACK_END_MAGIC)
2781 static inline int object_is_on_stack(void *obj)
2783 void *stack = task_stack_page(current);
2785 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2788 extern void thread_info_cache_init(void);
2790 #ifdef CONFIG_DEBUG_STACK_USAGE
2791 static inline unsigned long stack_not_used(struct task_struct *p)
2793 unsigned long *n = end_of_stack(p);
2795 do { /* Skip over canary */
2799 return (unsigned long)n - (unsigned long)end_of_stack(p);
2802 extern void set_task_stack_end_magic(struct task_struct *tsk);
2804 /* set thread flags in other task's structures
2805 * - see asm/thread_info.h for TIF_xxxx flags available
2807 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2809 set_ti_thread_flag(task_thread_info(tsk), flag);
2812 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2814 clear_ti_thread_flag(task_thread_info(tsk), flag);
2817 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2819 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2822 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2824 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2827 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2829 return test_ti_thread_flag(task_thread_info(tsk), flag);
2832 static inline void set_tsk_need_resched(struct task_struct *tsk)
2834 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2837 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2839 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2842 static inline int test_tsk_need_resched(struct task_struct *tsk)
2844 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2847 static inline int restart_syscall(void)
2849 set_tsk_thread_flag(current, TIF_SIGPENDING);
2850 return -ERESTARTNOINTR;
2853 static inline int signal_pending(struct task_struct *p)
2855 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2858 static inline int __fatal_signal_pending(struct task_struct *p)
2860 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2863 static inline int fatal_signal_pending(struct task_struct *p)
2865 return signal_pending(p) && __fatal_signal_pending(p);
2868 static inline int signal_pending_state(long state, struct task_struct *p)
2870 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2872 if (!signal_pending(p))
2875 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2879 * cond_resched() and cond_resched_lock(): latency reduction via
2880 * explicit rescheduling in places that are safe. The return
2881 * value indicates whether a reschedule was done in fact.
2882 * cond_resched_lock() will drop the spinlock before scheduling,
2883 * cond_resched_softirq() will enable bhs before scheduling.
2885 extern int _cond_resched(void);
2887 #define cond_resched() ({ \
2888 ___might_sleep(__FILE__, __LINE__, 0); \
2892 extern int __cond_resched_lock(spinlock_t *lock);
2894 #define cond_resched_lock(lock) ({ \
2895 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2896 __cond_resched_lock(lock); \
2899 extern int __cond_resched_softirq(void);
2901 #define cond_resched_softirq() ({ \
2902 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2903 __cond_resched_softirq(); \
2906 static inline void cond_resched_rcu(void)
2908 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2916 * Does a critical section need to be broken due to another
2917 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2918 * but a general need for low latency)
2920 static inline int spin_needbreak(spinlock_t *lock)
2922 #ifdef CONFIG_PREEMPT
2923 return spin_is_contended(lock);
2930 * Idle thread specific functions to determine the need_resched
2933 #ifdef TIF_POLLING_NRFLAG
2934 static inline int tsk_is_polling(struct task_struct *p)
2936 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2939 static inline void __current_set_polling(void)
2941 set_thread_flag(TIF_POLLING_NRFLAG);
2944 static inline bool __must_check current_set_polling_and_test(void)
2946 __current_set_polling();
2949 * Polling state must be visible before we test NEED_RESCHED,
2950 * paired by resched_curr()
2952 smp_mb__after_atomic();
2954 return unlikely(tif_need_resched());
2957 static inline void __current_clr_polling(void)
2959 clear_thread_flag(TIF_POLLING_NRFLAG);
2962 static inline bool __must_check current_clr_polling_and_test(void)
2964 __current_clr_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());
2976 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2977 static inline void __current_set_polling(void) { }
2978 static inline void __current_clr_polling(void) { }
2980 static inline bool __must_check current_set_polling_and_test(void)
2982 return unlikely(tif_need_resched());
2984 static inline bool __must_check current_clr_polling_and_test(void)
2986 return unlikely(tif_need_resched());
2990 static inline void current_clr_polling(void)
2992 __current_clr_polling();
2995 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2996 * Once the bit is cleared, we'll get IPIs with every new
2997 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3000 smp_mb(); /* paired with resched_curr() */
3002 preempt_fold_need_resched();
3005 static __always_inline bool need_resched(void)
3007 return unlikely(tif_need_resched());
3011 * Thread group CPU time accounting.
3013 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3014 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3017 * Reevaluate whether the task has signals pending delivery.
3018 * Wake the task if so.
3019 * This is required every time the blocked sigset_t changes.
3020 * callers must hold sighand->siglock.
3022 extern void recalc_sigpending_and_wake(struct task_struct *t);
3023 extern void recalc_sigpending(void);
3025 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3027 static inline void signal_wake_up(struct task_struct *t, bool resume)
3029 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3031 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3033 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3037 * Wrappers for p->thread_info->cpu access. No-op on UP.
3041 static inline unsigned int task_cpu(const struct task_struct *p)
3043 return task_thread_info(p)->cpu;
3046 static inline int task_node(const struct task_struct *p)
3048 return cpu_to_node(task_cpu(p));
3051 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3055 static inline unsigned int task_cpu(const struct task_struct *p)
3060 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3064 #endif /* CONFIG_SMP */
3066 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3067 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3069 #ifdef CONFIG_CGROUP_SCHED
3070 extern struct task_group root_task_group;
3071 #endif /* CONFIG_CGROUP_SCHED */
3073 extern int task_can_switch_user(struct user_struct *up,
3074 struct task_struct *tsk);
3076 #ifdef CONFIG_TASK_XACCT
3077 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3079 tsk->ioac.rchar += amt;
3082 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3084 tsk->ioac.wchar += amt;
3087 static inline void inc_syscr(struct task_struct *tsk)
3092 static inline void inc_syscw(struct task_struct *tsk)
3097 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3101 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3105 static inline void inc_syscr(struct task_struct *tsk)
3109 static inline void inc_syscw(struct task_struct *tsk)
3114 #ifndef TASK_SIZE_OF
3115 #define TASK_SIZE_OF(tsk) TASK_SIZE
3119 extern void mm_update_next_owner(struct mm_struct *mm);
3121 static inline void mm_update_next_owner(struct mm_struct *mm)
3124 #endif /* CONFIG_MEMCG */
3126 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3129 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3132 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3135 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3138 static inline unsigned long rlimit(unsigned int limit)
3140 return task_rlimit(current, limit);
3143 static inline unsigned long rlimit_max(unsigned int limit)
3145 return task_rlimit_max(current, limit);