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);
196 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
200 * Task state bitmask. NOTE! These bits are also
201 * encoded in fs/proc/array.c: get_task_state().
203 * We have two separate sets of flags: task->state
204 * is about runnability, while task->exit_state are
205 * about the task exiting. Confusing, but this way
206 * modifying one set can't modify the other one by
209 #define TASK_RUNNING 0
210 #define TASK_INTERRUPTIBLE 1
211 #define TASK_UNINTERRUPTIBLE 2
212 #define __TASK_STOPPED 4
213 #define __TASK_TRACED 8
214 /* in tsk->exit_state */
216 #define EXIT_ZOMBIE 32
217 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
218 /* in tsk->state again */
220 #define TASK_WAKEKILL 128
221 #define TASK_WAKING 256
222 #define TASK_PARKED 512
223 #define TASK_NOLOAD 1024
224 #define TASK_STATE_MAX 2048
226 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
228 extern char ___assert_task_state[1 - 2*!!(
229 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
231 /* Convenience macros for the sake of set_task_state */
232 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
233 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
234 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
236 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
238 /* Convenience macros for the sake of wake_up */
239 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
240 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
242 /* get_task_state() */
243 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
244 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
245 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
247 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
248 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
249 #define task_is_stopped_or_traced(task) \
250 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
251 #define task_contributes_to_load(task) \
252 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
253 (task->flags & PF_FROZEN) == 0 && \
254 (task->state & TASK_NOLOAD) == 0)
256 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
258 #define __set_task_state(tsk, state_value) \
260 (tsk)->task_state_change = _THIS_IP_; \
261 (tsk)->state = (state_value); \
263 #define set_task_state(tsk, state_value) \
265 (tsk)->task_state_change = _THIS_IP_; \
266 smp_store_mb((tsk)->state, (state_value)); \
270 * set_current_state() includes a barrier so that the write of current->state
271 * is correctly serialised wrt the caller's subsequent test of whether to
274 * set_current_state(TASK_UNINTERRUPTIBLE);
275 * if (do_i_need_to_sleep())
278 * If the caller does not need such serialisation then use __set_current_state()
280 #define __set_current_state(state_value) \
282 current->task_state_change = _THIS_IP_; \
283 current->state = (state_value); \
285 #define set_current_state(state_value) \
287 current->task_state_change = _THIS_IP_; \
288 smp_store_mb(current->state, (state_value)); \
293 #define __set_task_state(tsk, state_value) \
294 do { (tsk)->state = (state_value); } while (0)
295 #define set_task_state(tsk, state_value) \
296 smp_store_mb((tsk)->state, (state_value))
299 * set_current_state() includes a barrier so that the write of current->state
300 * is correctly serialised wrt the caller's subsequent test of whether to
303 * set_current_state(TASK_UNINTERRUPTIBLE);
304 * if (do_i_need_to_sleep())
307 * If the caller does not need such serialisation then use __set_current_state()
309 #define __set_current_state(state_value) \
310 do { current->state = (state_value); } while (0)
311 #define set_current_state(state_value) \
312 smp_store_mb(current->state, (state_value))
316 /* Task command name length */
317 #define TASK_COMM_LEN 16
319 #include <linux/spinlock.h>
322 * This serializes "schedule()" and also protects
323 * the run-queue from deletions/modifications (but
324 * _adding_ to the beginning of the run-queue has
327 extern rwlock_t tasklist_lock;
328 extern spinlock_t mmlist_lock;
332 #ifdef CONFIG_PROVE_RCU
333 extern int lockdep_tasklist_lock_is_held(void);
334 #endif /* #ifdef CONFIG_PROVE_RCU */
336 extern void sched_init(void);
337 extern void sched_init_smp(void);
338 extern asmlinkage void schedule_tail(struct task_struct *prev);
339 extern void init_idle(struct task_struct *idle, int cpu);
340 extern void init_idle_bootup_task(struct task_struct *idle);
342 extern cpumask_var_t cpu_isolated_map;
344 extern int runqueue_is_locked(int cpu);
346 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
347 extern void nohz_balance_enter_idle(int cpu);
348 extern void set_cpu_sd_state_idle(void);
349 extern int get_nohz_timer_target(void);
351 static inline void nohz_balance_enter_idle(int cpu) { }
352 static inline void set_cpu_sd_state_idle(void) { }
356 * Only dump TASK_* tasks. (0 for all tasks)
358 extern void show_state_filter(unsigned long state_filter);
360 static inline void show_state(void)
362 show_state_filter(0);
365 extern void show_regs(struct pt_regs *);
368 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
369 * task), SP is the stack pointer of the first frame that should be shown in the back
370 * trace (or NULL if the entire call-chain of the task should be shown).
372 extern void show_stack(struct task_struct *task, unsigned long *sp);
374 extern void cpu_init (void);
375 extern void trap_init(void);
376 extern void update_process_times(int user);
377 extern void scheduler_tick(void);
379 extern void sched_show_task(struct task_struct *p);
381 #ifdef CONFIG_LOCKUP_DETECTOR
382 extern void touch_softlockup_watchdog(void);
383 extern void touch_softlockup_watchdog_sync(void);
384 extern void touch_all_softlockup_watchdogs(void);
385 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
387 size_t *lenp, loff_t *ppos);
388 extern unsigned int softlockup_panic;
389 void lockup_detector_init(void);
391 static inline void touch_softlockup_watchdog(void)
394 static inline void touch_softlockup_watchdog_sync(void)
397 static inline void touch_all_softlockup_watchdogs(void)
400 static inline void lockup_detector_init(void)
405 #ifdef CONFIG_DETECT_HUNG_TASK
406 void reset_hung_task_detector(void);
408 static inline void reset_hung_task_detector(void)
413 /* Attach to any functions which should be ignored in wchan output. */
414 #define __sched __attribute__((__section__(".sched.text")))
416 /* Linker adds these: start and end of __sched functions */
417 extern char __sched_text_start[], __sched_text_end[];
419 /* Is this address in the __sched functions? */
420 extern int in_sched_functions(unsigned long addr);
422 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
423 extern signed long schedule_timeout(signed long timeout);
424 extern signed long schedule_timeout_interruptible(signed long timeout);
425 extern signed long schedule_timeout_killable(signed long timeout);
426 extern signed long schedule_timeout_uninterruptible(signed long timeout);
427 asmlinkage void schedule(void);
428 extern void schedule_preempt_disabled(void);
430 extern long io_schedule_timeout(long timeout);
432 static inline void io_schedule(void)
434 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
438 struct user_namespace;
441 extern void arch_pick_mmap_layout(struct mm_struct *mm);
443 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
444 unsigned long, unsigned long);
446 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
447 unsigned long len, unsigned long pgoff,
448 unsigned long flags);
450 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
453 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
454 #define SUID_DUMP_USER 1 /* Dump as user of process */
455 #define SUID_DUMP_ROOT 2 /* Dump as root */
459 /* for SUID_DUMP_* above */
460 #define MMF_DUMPABLE_BITS 2
461 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
463 extern void set_dumpable(struct mm_struct *mm, int value);
465 * This returns the actual value of the suid_dumpable flag. For things
466 * that are using this for checking for privilege transitions, it must
467 * test against SUID_DUMP_USER rather than treating it as a boolean
470 static inline int __get_dumpable(unsigned long mm_flags)
472 return mm_flags & MMF_DUMPABLE_MASK;
475 static inline int get_dumpable(struct mm_struct *mm)
477 return __get_dumpable(mm->flags);
480 /* coredump filter bits */
481 #define MMF_DUMP_ANON_PRIVATE 2
482 #define MMF_DUMP_ANON_SHARED 3
483 #define MMF_DUMP_MAPPED_PRIVATE 4
484 #define MMF_DUMP_MAPPED_SHARED 5
485 #define MMF_DUMP_ELF_HEADERS 6
486 #define MMF_DUMP_HUGETLB_PRIVATE 7
487 #define MMF_DUMP_HUGETLB_SHARED 8
489 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
490 #define MMF_DUMP_FILTER_BITS 7
491 #define MMF_DUMP_FILTER_MASK \
492 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
493 #define MMF_DUMP_FILTER_DEFAULT \
494 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
495 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
497 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
498 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
500 # define MMF_DUMP_MASK_DEFAULT_ELF 0
502 /* leave room for more dump flags */
503 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
504 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
505 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
507 #define MMF_HAS_UPROBES 19 /* has uprobes */
508 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
510 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
512 struct sighand_struct {
514 struct k_sigaction action[_NSIG];
516 wait_queue_head_t signalfd_wqh;
519 struct pacct_struct {
522 unsigned long ac_mem;
523 cputime_t ac_utime, ac_stime;
524 unsigned long ac_minflt, ac_majflt;
535 * struct cputime - snaphsot of system and user cputime
536 * @utime: time spent in user mode
537 * @stime: time spent in system mode
539 * Gathers a generic snapshot of user and system time.
547 * struct task_cputime - collected CPU time counts
548 * @utime: time spent in user mode, in &cputime_t units
549 * @stime: time spent in kernel mode, in &cputime_t units
550 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
552 * This is an extension of struct cputime that includes the total runtime
553 * spent by the task from the scheduler point of view.
555 * As a result, this structure groups together three kinds of CPU time
556 * that are tracked for threads and thread groups. Most things considering
557 * CPU time want to group these counts together and treat all three
558 * of them in parallel.
560 struct task_cputime {
563 unsigned long long sum_exec_runtime;
565 /* Alternate field names when used to cache expirations. */
566 #define prof_exp stime
567 #define virt_exp utime
568 #define sched_exp sum_exec_runtime
570 #define INIT_CPUTIME \
571 (struct task_cputime) { \
574 .sum_exec_runtime = 0, \
578 * This is the atomic variant of task_cputime, which can be used for
579 * storing and updating task_cputime statistics without locking.
581 struct task_cputime_atomic {
584 atomic64_t sum_exec_runtime;
587 #define INIT_CPUTIME_ATOMIC \
588 (struct task_cputime_atomic) { \
589 .utime = ATOMIC64_INIT(0), \
590 .stime = ATOMIC64_INIT(0), \
591 .sum_exec_runtime = ATOMIC64_INIT(0), \
594 #ifdef CONFIG_PREEMPT_COUNT
595 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
597 #define PREEMPT_DISABLED PREEMPT_ENABLED
601 * Disable preemption until the scheduler is running.
602 * Reset by start_kernel()->sched_init()->init_idle().
604 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
605 * before the scheduler is active -- see should_resched().
607 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
610 * struct thread_group_cputimer - thread group interval timer counts
611 * @cputime_atomic: atomic thread group interval timers.
612 * @running: non-zero when there are timers running and
613 * @cputime receives updates.
615 * This structure contains the version of task_cputime, above, that is
616 * used for thread group CPU timer calculations.
618 struct thread_group_cputimer {
619 struct task_cputime_atomic cputime_atomic;
623 #include <linux/rwsem.h>
627 * NOTE! "signal_struct" does not have its own
628 * locking, because a shared signal_struct always
629 * implies a shared sighand_struct, so locking
630 * sighand_struct is always a proper superset of
631 * the locking of signal_struct.
633 struct signal_struct {
637 struct list_head thread_head;
639 wait_queue_head_t wait_chldexit; /* for wait4() */
641 /* current thread group signal load-balancing target: */
642 struct task_struct *curr_target;
644 /* shared signal handling: */
645 struct sigpending shared_pending;
647 /* thread group exit support */
650 * - notify group_exit_task when ->count is equal to notify_count
651 * - everyone except group_exit_task is stopped during signal delivery
652 * of fatal signals, group_exit_task processes the signal.
655 struct task_struct *group_exit_task;
657 /* thread group stop support, overloads group_exit_code too */
658 int group_stop_count;
659 unsigned int flags; /* see SIGNAL_* flags below */
662 * PR_SET_CHILD_SUBREAPER marks a process, like a service
663 * manager, to re-parent orphan (double-forking) child processes
664 * to this process instead of 'init'. The service manager is
665 * able to receive SIGCHLD signals and is able to investigate
666 * the process until it calls wait(). All children of this
667 * process will inherit a flag if they should look for a
668 * child_subreaper process at exit.
670 unsigned int is_child_subreaper:1;
671 unsigned int has_child_subreaper:1;
673 /* POSIX.1b Interval Timers */
675 struct list_head posix_timers;
677 /* ITIMER_REAL timer for the process */
678 struct hrtimer real_timer;
679 struct pid *leader_pid;
680 ktime_t it_real_incr;
683 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
684 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
685 * values are defined to 0 and 1 respectively
687 struct cpu_itimer it[2];
690 * Thread group totals for process CPU timers.
691 * See thread_group_cputimer(), et al, for details.
693 struct thread_group_cputimer cputimer;
695 /* Earliest-expiration cache. */
696 struct task_cputime cputime_expires;
698 struct list_head cpu_timers[3];
700 struct pid *tty_old_pgrp;
702 /* boolean value for session group leader */
705 struct tty_struct *tty; /* NULL if no tty */
707 #ifdef CONFIG_SCHED_AUTOGROUP
708 struct autogroup *autogroup;
711 * Cumulative resource counters for dead threads in the group,
712 * and for reaped dead child processes forked by this group.
713 * Live threads maintain their own counters and add to these
714 * in __exit_signal, except for the group leader.
716 seqlock_t stats_lock;
717 cputime_t utime, stime, cutime, cstime;
720 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
721 struct cputime prev_cputime;
723 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
724 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
725 unsigned long inblock, oublock, cinblock, coublock;
726 unsigned long maxrss, cmaxrss;
727 struct task_io_accounting ioac;
730 * Cumulative ns of schedule CPU time fo dead threads in the
731 * group, not including a zombie group leader, (This only differs
732 * from jiffies_to_ns(utime + stime) if sched_clock uses something
733 * other than jiffies.)
735 unsigned long long sum_sched_runtime;
738 * We don't bother to synchronize most readers of this at all,
739 * because there is no reader checking a limit that actually needs
740 * to get both rlim_cur and rlim_max atomically, and either one
741 * alone is a single word that can safely be read normally.
742 * getrlimit/setrlimit use task_lock(current->group_leader) to
743 * protect this instead of the siglock, because they really
744 * have no need to disable irqs.
746 struct rlimit rlim[RLIM_NLIMITS];
748 #ifdef CONFIG_BSD_PROCESS_ACCT
749 struct pacct_struct pacct; /* per-process accounting information */
751 #ifdef CONFIG_TASKSTATS
752 struct taskstats *stats;
756 unsigned audit_tty_log_passwd;
757 struct tty_audit_buf *tty_audit_buf;
760 oom_flags_t oom_flags;
761 short oom_score_adj; /* OOM kill score adjustment */
762 short oom_score_adj_min; /* OOM kill score adjustment min value.
763 * Only settable by CAP_SYS_RESOURCE. */
765 struct mutex cred_guard_mutex; /* guard against foreign influences on
766 * credential calculations
767 * (notably. ptrace) */
771 * Bits in flags field of signal_struct.
773 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
774 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
775 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
776 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
778 * Pending notifications to parent.
780 #define SIGNAL_CLD_STOPPED 0x00000010
781 #define SIGNAL_CLD_CONTINUED 0x00000020
782 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
784 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
786 /* If true, all threads except ->group_exit_task have pending SIGKILL */
787 static inline int signal_group_exit(const struct signal_struct *sig)
789 return (sig->flags & SIGNAL_GROUP_EXIT) ||
790 (sig->group_exit_task != NULL);
794 * Some day this will be a full-fledged user tracking system..
797 atomic_t __count; /* reference count */
798 atomic_t processes; /* How many processes does this user have? */
799 atomic_t sigpending; /* How many pending signals does this user have? */
800 #ifdef CONFIG_INOTIFY_USER
801 atomic_t inotify_watches; /* How many inotify watches does this user have? */
802 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
804 #ifdef CONFIG_FANOTIFY
805 atomic_t fanotify_listeners;
808 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
810 #ifdef CONFIG_POSIX_MQUEUE
811 /* protected by mq_lock */
812 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
814 unsigned long locked_shm; /* How many pages of mlocked shm ? */
817 struct key *uid_keyring; /* UID specific keyring */
818 struct key *session_keyring; /* UID's default session keyring */
821 /* Hash table maintenance information */
822 struct hlist_node uidhash_node;
825 #ifdef CONFIG_PERF_EVENTS
826 atomic_long_t locked_vm;
830 extern int uids_sysfs_init(void);
832 extern struct user_struct *find_user(kuid_t);
834 extern struct user_struct root_user;
835 #define INIT_USER (&root_user)
838 struct backing_dev_info;
839 struct reclaim_state;
841 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
843 /* cumulative counters */
844 unsigned long pcount; /* # of times run on this cpu */
845 unsigned long long run_delay; /* time spent waiting on a runqueue */
848 unsigned long long last_arrival,/* when we last ran on a cpu */
849 last_queued; /* when we were last queued to run */
851 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
853 #ifdef CONFIG_TASK_DELAY_ACCT
854 struct task_delay_info {
856 unsigned int flags; /* Private per-task flags */
858 /* For each stat XXX, add following, aligned appropriately
860 * struct timespec XXX_start, XXX_end;
864 * Atomicity of updates to XXX_delay, XXX_count protected by
865 * single lock above (split into XXX_lock if contention is an issue).
869 * XXX_count is incremented on every XXX operation, the delay
870 * associated with the operation is added to XXX_delay.
871 * XXX_delay contains the accumulated delay time in nanoseconds.
873 u64 blkio_start; /* Shared by blkio, swapin */
874 u64 blkio_delay; /* wait for sync block io completion */
875 u64 swapin_delay; /* wait for swapin block io completion */
876 u32 blkio_count; /* total count of the number of sync block */
877 /* io operations performed */
878 u32 swapin_count; /* total count of the number of swapin block */
879 /* io operations performed */
882 u64 freepages_delay; /* wait for memory reclaim */
883 u32 freepages_count; /* total count of memory reclaim */
885 #endif /* CONFIG_TASK_DELAY_ACCT */
887 static inline int sched_info_on(void)
889 #ifdef CONFIG_SCHEDSTATS
891 #elif defined(CONFIG_TASK_DELAY_ACCT)
892 extern int delayacct_on;
907 * Increase resolution of cpu_capacity calculations
909 #define SCHED_CAPACITY_SHIFT 10
910 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
913 * Wake-queues are lists of tasks with a pending wakeup, whose
914 * callers have already marked the task as woken internally,
915 * and can thus carry on. A common use case is being able to
916 * do the wakeups once the corresponding user lock as been
919 * We hold reference to each task in the list across the wakeup,
920 * thus guaranteeing that the memory is still valid by the time
921 * the actual wakeups are performed in wake_up_q().
923 * One per task suffices, because there's never a need for a task to be
924 * in two wake queues simultaneously; it is forbidden to abandon a task
925 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
926 * already in a wake queue, the wakeup will happen soon and the second
927 * waker can just skip it.
929 * The WAKE_Q macro declares and initializes the list head.
930 * wake_up_q() does NOT reinitialize the list; it's expected to be
931 * called near the end of a function, where the fact that the queue is
932 * not used again will be easy to see by inspection.
934 * Note that this can cause spurious wakeups. schedule() callers
935 * must ensure the call is done inside a loop, confirming that the
936 * wakeup condition has in fact occurred.
939 struct wake_q_node *next;
943 struct wake_q_node *first;
944 struct wake_q_node **lastp;
947 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
949 #define WAKE_Q(name) \
950 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
952 extern void wake_q_add(struct wake_q_head *head,
953 struct task_struct *task);
954 extern void wake_up_q(struct wake_q_head *head);
957 * sched-domains (multiprocessor balancing) declarations:
960 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
961 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
962 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
963 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
964 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
965 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
966 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
967 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
968 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
969 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
970 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
971 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
972 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
973 #define SD_NUMA 0x4000 /* cross-node balancing */
975 #ifdef CONFIG_SCHED_SMT
976 static inline int cpu_smt_flags(void)
978 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
982 #ifdef CONFIG_SCHED_MC
983 static inline int cpu_core_flags(void)
985 return SD_SHARE_PKG_RESOURCES;
990 static inline int cpu_numa_flags(void)
996 struct sched_domain_attr {
997 int relax_domain_level;
1000 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1001 .relax_domain_level = -1, \
1004 extern int sched_domain_level_max;
1008 struct sched_domain {
1009 /* These fields must be setup */
1010 struct sched_domain *parent; /* top domain must be null terminated */
1011 struct sched_domain *child; /* bottom domain must be null terminated */
1012 struct sched_group *groups; /* the balancing groups of the domain */
1013 unsigned long min_interval; /* Minimum balance interval ms */
1014 unsigned long max_interval; /* Maximum balance interval ms */
1015 unsigned int busy_factor; /* less balancing by factor if busy */
1016 unsigned int imbalance_pct; /* No balance until over watermark */
1017 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1018 unsigned int busy_idx;
1019 unsigned int idle_idx;
1020 unsigned int newidle_idx;
1021 unsigned int wake_idx;
1022 unsigned int forkexec_idx;
1023 unsigned int smt_gain;
1025 int nohz_idle; /* NOHZ IDLE status */
1026 int flags; /* See SD_* */
1029 /* Runtime fields. */
1030 unsigned long last_balance; /* init to jiffies. units in jiffies */
1031 unsigned int balance_interval; /* initialise to 1. units in ms. */
1032 unsigned int nr_balance_failed; /* initialise to 0 */
1034 /* idle_balance() stats */
1035 u64 max_newidle_lb_cost;
1036 unsigned long next_decay_max_lb_cost;
1038 #ifdef CONFIG_SCHEDSTATS
1039 /* load_balance() stats */
1040 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1041 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1042 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1043 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1044 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1045 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1046 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1047 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1049 /* Active load balancing */
1050 unsigned int alb_count;
1051 unsigned int alb_failed;
1052 unsigned int alb_pushed;
1054 /* SD_BALANCE_EXEC stats */
1055 unsigned int sbe_count;
1056 unsigned int sbe_balanced;
1057 unsigned int sbe_pushed;
1059 /* SD_BALANCE_FORK stats */
1060 unsigned int sbf_count;
1061 unsigned int sbf_balanced;
1062 unsigned int sbf_pushed;
1064 /* try_to_wake_up() stats */
1065 unsigned int ttwu_wake_remote;
1066 unsigned int ttwu_move_affine;
1067 unsigned int ttwu_move_balance;
1069 #ifdef CONFIG_SCHED_DEBUG
1073 void *private; /* used during construction */
1074 struct rcu_head rcu; /* used during destruction */
1077 unsigned int span_weight;
1079 * Span of all CPUs in this domain.
1081 * NOTE: this field is variable length. (Allocated dynamically
1082 * by attaching extra space to the end of the structure,
1083 * depending on how many CPUs the kernel has booted up with)
1085 unsigned long span[0];
1088 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1090 return to_cpumask(sd->span);
1093 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1094 struct sched_domain_attr *dattr_new);
1096 /* Allocate an array of sched domains, for partition_sched_domains(). */
1097 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1098 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1100 bool cpus_share_cache(int this_cpu, int that_cpu);
1102 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1103 typedef int (*sched_domain_flags_f)(void);
1105 #define SDTL_OVERLAP 0x01
1108 struct sched_domain **__percpu sd;
1109 struct sched_group **__percpu sg;
1110 struct sched_group_capacity **__percpu sgc;
1113 struct sched_domain_topology_level {
1114 sched_domain_mask_f mask;
1115 sched_domain_flags_f sd_flags;
1118 struct sd_data data;
1119 #ifdef CONFIG_SCHED_DEBUG
1124 extern struct sched_domain_topology_level *sched_domain_topology;
1126 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1127 extern void wake_up_if_idle(int cpu);
1129 #ifdef CONFIG_SCHED_DEBUG
1130 # define SD_INIT_NAME(type) .name = #type
1132 # define SD_INIT_NAME(type)
1135 #else /* CONFIG_SMP */
1137 struct sched_domain_attr;
1140 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1141 struct sched_domain_attr *dattr_new)
1145 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1150 #endif /* !CONFIG_SMP */
1153 struct io_context; /* See blkdev.h */
1156 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1157 extern void prefetch_stack(struct task_struct *t);
1159 static inline void prefetch_stack(struct task_struct *t) { }
1162 struct audit_context; /* See audit.c */
1164 struct pipe_inode_info;
1165 struct uts_namespace;
1167 struct load_weight {
1168 unsigned long weight;
1173 u64 last_runnable_update;
1176 * utilization_avg_contrib describes the amount of time that a
1177 * sched_entity is running on a CPU. It is based on running_avg_sum
1178 * and is scaled in the range [0..SCHED_LOAD_SCALE].
1179 * load_avg_contrib described the amount of time that a sched_entity
1180 * is runnable on a rq. It is based on both runnable_avg_sum and the
1181 * weight of the task.
1183 unsigned long load_avg_contrib, utilization_avg_contrib;
1185 * These sums represent an infinite geometric series and so are bound
1186 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1187 * choices of y < 1-2^(-32)*1024.
1188 * running_avg_sum reflects the time that the sched_entity is
1189 * effectively running on the CPU.
1190 * runnable_avg_sum represents the amount of time a sched_entity is on
1191 * a runqueue which includes the running time that is monitored by
1194 u32 runnable_avg_sum, avg_period, running_avg_sum;
1197 #ifdef CONFIG_SCHEDSTATS
1198 struct sched_statistics {
1208 s64 sum_sleep_runtime;
1215 u64 nr_migrations_cold;
1216 u64 nr_failed_migrations_affine;
1217 u64 nr_failed_migrations_running;
1218 u64 nr_failed_migrations_hot;
1219 u64 nr_forced_migrations;
1222 u64 nr_wakeups_sync;
1223 u64 nr_wakeups_migrate;
1224 u64 nr_wakeups_local;
1225 u64 nr_wakeups_remote;
1226 u64 nr_wakeups_affine;
1227 u64 nr_wakeups_affine_attempts;
1228 u64 nr_wakeups_passive;
1229 u64 nr_wakeups_idle;
1233 struct sched_entity {
1234 struct load_weight load; /* for load-balancing */
1235 struct rb_node run_node;
1236 struct list_head group_node;
1240 u64 sum_exec_runtime;
1242 u64 prev_sum_exec_runtime;
1246 #ifdef CONFIG_SCHEDSTATS
1247 struct sched_statistics statistics;
1250 #ifdef CONFIG_FAIR_GROUP_SCHED
1252 struct sched_entity *parent;
1253 /* rq on which this entity is (to be) queued: */
1254 struct cfs_rq *cfs_rq;
1255 /* rq "owned" by this entity/group: */
1256 struct cfs_rq *my_q;
1260 /* Per-entity load-tracking */
1261 struct sched_avg avg;
1265 struct sched_rt_entity {
1266 struct list_head run_list;
1267 unsigned long timeout;
1268 unsigned long watchdog_stamp;
1269 unsigned int time_slice;
1271 struct sched_rt_entity *back;
1272 #ifdef CONFIG_RT_GROUP_SCHED
1273 struct sched_rt_entity *parent;
1274 /* rq on which this entity is (to be) queued: */
1275 struct rt_rq *rt_rq;
1276 /* rq "owned" by this entity/group: */
1281 struct sched_dl_entity {
1282 struct rb_node rb_node;
1285 * Original scheduling parameters. Copied here from sched_attr
1286 * during sched_setattr(), they will remain the same until
1287 * the next sched_setattr().
1289 u64 dl_runtime; /* maximum runtime for each instance */
1290 u64 dl_deadline; /* relative deadline of each instance */
1291 u64 dl_period; /* separation of two instances (period) */
1292 u64 dl_bw; /* dl_runtime / dl_deadline */
1295 * Actual scheduling parameters. Initialized with the values above,
1296 * they are continously updated during task execution. Note that
1297 * the remaining runtime could be < 0 in case we are in overrun.
1299 s64 runtime; /* remaining runtime for this instance */
1300 u64 deadline; /* absolute deadline for this instance */
1301 unsigned int flags; /* specifying the scheduler behaviour */
1306 * @dl_throttled tells if we exhausted the runtime. If so, the
1307 * task has to wait for a replenishment to be performed at the
1308 * next firing of dl_timer.
1310 * @dl_new tells if a new instance arrived. If so we must
1311 * start executing it with full runtime and reset its absolute
1314 * @dl_boosted tells if we are boosted due to DI. If so we are
1315 * outside bandwidth enforcement mechanism (but only until we
1316 * exit the critical section);
1318 * @dl_yielded tells if task gave up the cpu before consuming
1319 * all its available runtime during the last job.
1321 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1324 * Bandwidth enforcement timer. Each -deadline task has its
1325 * own bandwidth to be enforced, thus we need one timer per task.
1327 struct hrtimer dl_timer;
1339 enum perf_event_task_context {
1340 perf_invalid_context = -1,
1341 perf_hw_context = 0,
1343 perf_nr_task_contexts,
1346 struct task_struct {
1347 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1350 unsigned int flags; /* per process flags, defined below */
1351 unsigned int ptrace;
1354 struct llist_node wake_entry;
1356 struct task_struct *last_wakee;
1357 unsigned long wakee_flips;
1358 unsigned long wakee_flip_decay_ts;
1364 int prio, static_prio, normal_prio;
1365 unsigned int rt_priority;
1366 const struct sched_class *sched_class;
1367 struct sched_entity se;
1368 struct sched_rt_entity rt;
1369 #ifdef CONFIG_CGROUP_SCHED
1370 struct task_group *sched_task_group;
1372 struct sched_dl_entity dl;
1374 #ifdef CONFIG_PREEMPT_NOTIFIERS
1375 /* list of struct preempt_notifier: */
1376 struct hlist_head preempt_notifiers;
1379 #ifdef CONFIG_BLK_DEV_IO_TRACE
1380 unsigned int btrace_seq;
1383 unsigned int policy;
1384 int nr_cpus_allowed;
1385 cpumask_t cpus_allowed;
1387 #ifdef CONFIG_PREEMPT_RCU
1388 int rcu_read_lock_nesting;
1389 union rcu_special rcu_read_unlock_special;
1390 struct list_head rcu_node_entry;
1391 struct rcu_node *rcu_blocked_node;
1392 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1393 #ifdef CONFIG_TASKS_RCU
1394 unsigned long rcu_tasks_nvcsw;
1395 bool rcu_tasks_holdout;
1396 struct list_head rcu_tasks_holdout_list;
1397 int rcu_tasks_idle_cpu;
1398 #endif /* #ifdef CONFIG_TASKS_RCU */
1400 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1401 struct sched_info sched_info;
1404 struct list_head tasks;
1406 struct plist_node pushable_tasks;
1407 struct rb_node pushable_dl_tasks;
1410 struct mm_struct *mm, *active_mm;
1411 /* per-thread vma caching */
1412 u32 vmacache_seqnum;
1413 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1414 #if defined(SPLIT_RSS_COUNTING)
1415 struct task_rss_stat rss_stat;
1419 int exit_code, exit_signal;
1420 int pdeath_signal; /* The signal sent when the parent dies */
1421 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1423 /* Used for emulating ABI behavior of previous Linux versions */
1424 unsigned int personality;
1426 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1428 unsigned in_iowait:1;
1430 /* Revert to default priority/policy when forking */
1431 unsigned sched_reset_on_fork:1;
1432 unsigned sched_contributes_to_load:1;
1433 unsigned sched_migrated:1;
1435 #ifdef CONFIG_MEMCG_KMEM
1436 unsigned memcg_kmem_skip_account:1;
1438 #ifdef CONFIG_COMPAT_BRK
1439 unsigned brk_randomized:1;
1442 unsigned long atomic_flags; /* Flags needing atomic access. */
1444 struct restart_block restart_block;
1449 #ifdef CONFIG_CC_STACKPROTECTOR
1450 /* Canary value for the -fstack-protector gcc feature */
1451 unsigned long stack_canary;
1454 * pointers to (original) parent process, youngest child, younger sibling,
1455 * older sibling, respectively. (p->father can be replaced with
1456 * p->real_parent->pid)
1458 struct task_struct __rcu *real_parent; /* real parent process */
1459 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1461 * children/sibling forms the list of my natural children
1463 struct list_head children; /* list of my children */
1464 struct list_head sibling; /* linkage in my parent's children list */
1465 struct task_struct *group_leader; /* threadgroup leader */
1468 * ptraced is the list of tasks this task is using ptrace on.
1469 * This includes both natural children and PTRACE_ATTACH targets.
1470 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1472 struct list_head ptraced;
1473 struct list_head ptrace_entry;
1475 /* PID/PID hash table linkage. */
1476 struct pid_link pids[PIDTYPE_MAX];
1477 struct list_head thread_group;
1478 struct list_head thread_node;
1480 struct completion *vfork_done; /* for vfork() */
1481 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1482 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1484 cputime_t utime, stime, utimescaled, stimescaled;
1486 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1487 struct cputime prev_cputime;
1489 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1490 seqlock_t vtime_seqlock;
1491 unsigned long long vtime_snap;
1496 } vtime_snap_whence;
1498 unsigned long nvcsw, nivcsw; /* context switch counts */
1499 u64 start_time; /* monotonic time in nsec */
1500 u64 real_start_time; /* boot based time in nsec */
1501 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1502 unsigned long min_flt, maj_flt;
1504 struct task_cputime cputime_expires;
1505 struct list_head cpu_timers[3];
1507 /* process credentials */
1508 const struct cred __rcu *real_cred; /* objective and real subjective task
1509 * credentials (COW) */
1510 const struct cred __rcu *cred; /* effective (overridable) subjective task
1511 * credentials (COW) */
1512 char comm[TASK_COMM_LEN]; /* executable name excluding path
1513 - access with [gs]et_task_comm (which lock
1514 it with task_lock())
1515 - initialized normally by setup_new_exec */
1516 /* file system info */
1517 struct nameidata *nameidata;
1518 #ifdef CONFIG_SYSVIPC
1520 struct sysv_sem sysvsem;
1521 struct sysv_shm sysvshm;
1523 #ifdef CONFIG_DETECT_HUNG_TASK
1524 /* hung task detection */
1525 unsigned long last_switch_count;
1527 /* CPU-specific state of this task */
1528 struct thread_struct thread;
1529 /* filesystem information */
1530 struct fs_struct *fs;
1531 /* open file information */
1532 struct files_struct *files;
1534 struct nsproxy *nsproxy;
1535 /* signal handlers */
1536 struct signal_struct *signal;
1537 struct sighand_struct *sighand;
1539 sigset_t blocked, real_blocked;
1540 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1541 struct sigpending pending;
1543 unsigned long sas_ss_sp;
1545 int (*notifier)(void *priv);
1546 void *notifier_data;
1547 sigset_t *notifier_mask;
1548 struct callback_head *task_works;
1550 struct audit_context *audit_context;
1551 #ifdef CONFIG_AUDITSYSCALL
1553 unsigned int sessionid;
1555 struct seccomp seccomp;
1557 /* Thread group tracking */
1560 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1562 spinlock_t alloc_lock;
1564 /* Protection of the PI data structures: */
1565 raw_spinlock_t pi_lock;
1567 struct wake_q_node wake_q;
1569 #ifdef CONFIG_RT_MUTEXES
1570 /* PI waiters blocked on a rt_mutex held by this task */
1571 struct rb_root pi_waiters;
1572 struct rb_node *pi_waiters_leftmost;
1573 /* Deadlock detection and priority inheritance handling */
1574 struct rt_mutex_waiter *pi_blocked_on;
1577 #ifdef CONFIG_DEBUG_MUTEXES
1578 /* mutex deadlock detection */
1579 struct mutex_waiter *blocked_on;
1581 #ifdef CONFIG_TRACE_IRQFLAGS
1582 unsigned int irq_events;
1583 unsigned long hardirq_enable_ip;
1584 unsigned long hardirq_disable_ip;
1585 unsigned int hardirq_enable_event;
1586 unsigned int hardirq_disable_event;
1587 int hardirqs_enabled;
1588 int hardirq_context;
1589 unsigned long softirq_disable_ip;
1590 unsigned long softirq_enable_ip;
1591 unsigned int softirq_disable_event;
1592 unsigned int softirq_enable_event;
1593 int softirqs_enabled;
1594 int softirq_context;
1596 #ifdef CONFIG_LOCKDEP
1597 # define MAX_LOCK_DEPTH 48UL
1600 unsigned int lockdep_recursion;
1601 struct held_lock held_locks[MAX_LOCK_DEPTH];
1602 gfp_t lockdep_reclaim_gfp;
1605 /* journalling filesystem info */
1608 /* stacked block device info */
1609 struct bio_list *bio_list;
1612 /* stack plugging */
1613 struct blk_plug *plug;
1617 struct reclaim_state *reclaim_state;
1619 struct backing_dev_info *backing_dev_info;
1621 struct io_context *io_context;
1623 unsigned long ptrace_message;
1624 siginfo_t *last_siginfo; /* For ptrace use. */
1625 struct task_io_accounting ioac;
1626 #if defined(CONFIG_TASK_XACCT)
1627 u64 acct_rss_mem1; /* accumulated rss usage */
1628 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1629 cputime_t acct_timexpd; /* stime + utime since last update */
1631 #ifdef CONFIG_CPUSETS
1632 nodemask_t mems_allowed; /* Protected by alloc_lock */
1633 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1634 int cpuset_mem_spread_rotor;
1635 int cpuset_slab_spread_rotor;
1637 #ifdef CONFIG_CGROUPS
1638 /* Control Group info protected by css_set_lock */
1639 struct css_set __rcu *cgroups;
1640 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1641 struct list_head cg_list;
1644 struct robust_list_head __user *robust_list;
1645 #ifdef CONFIG_COMPAT
1646 struct compat_robust_list_head __user *compat_robust_list;
1648 struct list_head pi_state_list;
1649 struct futex_pi_state *pi_state_cache;
1651 #ifdef CONFIG_PERF_EVENTS
1652 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1653 struct mutex perf_event_mutex;
1654 struct list_head perf_event_list;
1656 #ifdef CONFIG_DEBUG_PREEMPT
1657 unsigned long preempt_disable_ip;
1660 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1662 short pref_node_fork;
1664 #ifdef CONFIG_NUMA_BALANCING
1666 unsigned int numa_scan_period;
1667 unsigned int numa_scan_period_max;
1668 int numa_preferred_nid;
1669 unsigned long numa_migrate_retry;
1670 u64 node_stamp; /* migration stamp */
1671 u64 last_task_numa_placement;
1672 u64 last_sum_exec_runtime;
1673 struct callback_head numa_work;
1675 struct list_head numa_entry;
1676 struct numa_group *numa_group;
1679 * numa_faults is an array split into four regions:
1680 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1681 * in this precise order.
1683 * faults_memory: Exponential decaying average of faults on a per-node
1684 * basis. Scheduling placement decisions are made based on these
1685 * counts. The values remain static for the duration of a PTE scan.
1686 * faults_cpu: Track the nodes the process was running on when a NUMA
1687 * hinting fault was incurred.
1688 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1689 * during the current scan window. When the scan completes, the counts
1690 * in faults_memory and faults_cpu decay and these values are copied.
1692 unsigned long *numa_faults;
1693 unsigned long total_numa_faults;
1696 * numa_faults_locality tracks if faults recorded during the last
1697 * scan window were remote/local or failed to migrate. The task scan
1698 * period is adapted based on the locality of the faults with different
1699 * weights depending on whether they were shared or private faults
1701 unsigned long numa_faults_locality[3];
1703 unsigned long numa_pages_migrated;
1704 #endif /* CONFIG_NUMA_BALANCING */
1706 struct rcu_head rcu;
1709 * cache last used pipe for splice
1711 struct pipe_inode_info *splice_pipe;
1713 struct page_frag task_frag;
1715 #ifdef CONFIG_TASK_DELAY_ACCT
1716 struct task_delay_info *delays;
1718 #ifdef CONFIG_FAULT_INJECTION
1722 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1723 * balance_dirty_pages() for some dirty throttling pause
1726 int nr_dirtied_pause;
1727 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1729 #ifdef CONFIG_LATENCYTOP
1730 int latency_record_count;
1731 struct latency_record latency_record[LT_SAVECOUNT];
1734 * time slack values; these are used to round up poll() and
1735 * select() etc timeout values. These are in nanoseconds.
1737 unsigned long timer_slack_ns;
1738 unsigned long default_timer_slack_ns;
1741 unsigned int kasan_depth;
1743 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1744 /* Index of current stored address in ret_stack */
1746 /* Stack of return addresses for return function tracing */
1747 struct ftrace_ret_stack *ret_stack;
1748 /* time stamp for last schedule */
1749 unsigned long long ftrace_timestamp;
1751 * Number of functions that haven't been traced
1752 * because of depth overrun.
1754 atomic_t trace_overrun;
1755 /* Pause for the tracing */
1756 atomic_t tracing_graph_pause;
1758 #ifdef CONFIG_TRACING
1759 /* state flags for use by tracers */
1760 unsigned long trace;
1761 /* bitmask and counter of trace recursion */
1762 unsigned long trace_recursion;
1763 #endif /* CONFIG_TRACING */
1765 struct memcg_oom_info {
1766 struct mem_cgroup *memcg;
1769 unsigned int may_oom:1;
1772 #ifdef CONFIG_UPROBES
1773 struct uprobe_task *utask;
1775 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1776 unsigned int sequential_io;
1777 unsigned int sequential_io_avg;
1779 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1780 unsigned long task_state_change;
1782 int pagefault_disabled;
1785 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1786 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1788 #define TNF_MIGRATED 0x01
1789 #define TNF_NO_GROUP 0x02
1790 #define TNF_SHARED 0x04
1791 #define TNF_FAULT_LOCAL 0x08
1792 #define TNF_MIGRATE_FAIL 0x10
1794 #ifdef CONFIG_NUMA_BALANCING
1795 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1796 extern pid_t task_numa_group_id(struct task_struct *p);
1797 extern void set_numabalancing_state(bool enabled);
1798 extern void task_numa_free(struct task_struct *p);
1799 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1800 int src_nid, int dst_cpu);
1802 static inline void task_numa_fault(int last_node, int node, int pages,
1806 static inline pid_t task_numa_group_id(struct task_struct *p)
1810 static inline void set_numabalancing_state(bool enabled)
1813 static inline void task_numa_free(struct task_struct *p)
1816 static inline bool should_numa_migrate_memory(struct task_struct *p,
1817 struct page *page, int src_nid, int dst_cpu)
1823 static inline struct pid *task_pid(struct task_struct *task)
1825 return task->pids[PIDTYPE_PID].pid;
1828 static inline struct pid *task_tgid(struct task_struct *task)
1830 return task->group_leader->pids[PIDTYPE_PID].pid;
1834 * Without tasklist or rcu lock it is not safe to dereference
1835 * the result of task_pgrp/task_session even if task == current,
1836 * we can race with another thread doing sys_setsid/sys_setpgid.
1838 static inline struct pid *task_pgrp(struct task_struct *task)
1840 return task->group_leader->pids[PIDTYPE_PGID].pid;
1843 static inline struct pid *task_session(struct task_struct *task)
1845 return task->group_leader->pids[PIDTYPE_SID].pid;
1848 struct pid_namespace;
1851 * the helpers to get the task's different pids as they are seen
1852 * from various namespaces
1854 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1855 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1857 * task_xid_nr_ns() : id seen from the ns specified;
1859 * set_task_vxid() : assigns a virtual id to a task;
1861 * see also pid_nr() etc in include/linux/pid.h
1863 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1864 struct pid_namespace *ns);
1866 static inline pid_t task_pid_nr(struct task_struct *tsk)
1871 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1872 struct pid_namespace *ns)
1874 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1877 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1879 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1883 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1888 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1890 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1892 return pid_vnr(task_tgid(tsk));
1896 static inline int pid_alive(const struct task_struct *p);
1897 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1903 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1909 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1911 return task_ppid_nr_ns(tsk, &init_pid_ns);
1914 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1915 struct pid_namespace *ns)
1917 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1920 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1922 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1926 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1927 struct pid_namespace *ns)
1929 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1932 static inline pid_t task_session_vnr(struct task_struct *tsk)
1934 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1937 /* obsolete, do not use */
1938 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1940 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1944 * pid_alive - check that a task structure is not stale
1945 * @p: Task structure to be checked.
1947 * Test if a process is not yet dead (at most zombie state)
1948 * If pid_alive fails, then pointers within the task structure
1949 * can be stale and must not be dereferenced.
1951 * Return: 1 if the process is alive. 0 otherwise.
1953 static inline int pid_alive(const struct task_struct *p)
1955 return p->pids[PIDTYPE_PID].pid != NULL;
1959 * is_global_init - check if a task structure is init
1960 * @tsk: Task structure to be checked.
1962 * Check if a task structure is the first user space task the kernel created.
1964 * Return: 1 if the task structure is init. 0 otherwise.
1966 static inline int is_global_init(struct task_struct *tsk)
1968 return tsk->pid == 1;
1971 extern struct pid *cad_pid;
1973 extern void free_task(struct task_struct *tsk);
1974 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1976 extern void __put_task_struct(struct task_struct *t);
1978 static inline void put_task_struct(struct task_struct *t)
1980 if (atomic_dec_and_test(&t->usage))
1981 __put_task_struct(t);
1984 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1985 extern void task_cputime(struct task_struct *t,
1986 cputime_t *utime, cputime_t *stime);
1987 extern void task_cputime_scaled(struct task_struct *t,
1988 cputime_t *utimescaled, cputime_t *stimescaled);
1989 extern cputime_t task_gtime(struct task_struct *t);
1991 static inline void task_cputime(struct task_struct *t,
1992 cputime_t *utime, cputime_t *stime)
2000 static inline void task_cputime_scaled(struct task_struct *t,
2001 cputime_t *utimescaled,
2002 cputime_t *stimescaled)
2005 *utimescaled = t->utimescaled;
2007 *stimescaled = t->stimescaled;
2010 static inline cputime_t task_gtime(struct task_struct *t)
2015 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2016 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2021 #define PF_EXITING 0x00000004 /* getting shut down */
2022 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2023 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2024 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2025 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2026 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2027 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2028 #define PF_DUMPCORE 0x00000200 /* dumped core */
2029 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2030 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2031 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2032 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2033 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2034 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2035 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2036 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2037 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2038 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2039 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2040 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2041 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2042 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2043 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2044 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2045 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2046 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2047 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2050 * Only the _current_ task can read/write to tsk->flags, but other
2051 * tasks can access tsk->flags in readonly mode for example
2052 * with tsk_used_math (like during threaded core dumping).
2053 * There is however an exception to this rule during ptrace
2054 * or during fork: the ptracer task is allowed to write to the
2055 * child->flags of its traced child (same goes for fork, the parent
2056 * can write to the child->flags), because we're guaranteed the
2057 * child is not running and in turn not changing child->flags
2058 * at the same time the parent does it.
2060 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2061 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2062 #define clear_used_math() clear_stopped_child_used_math(current)
2063 #define set_used_math() set_stopped_child_used_math(current)
2064 #define conditional_stopped_child_used_math(condition, child) \
2065 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2066 #define conditional_used_math(condition) \
2067 conditional_stopped_child_used_math(condition, current)
2068 #define copy_to_stopped_child_used_math(child) \
2069 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2070 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2071 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2072 #define used_math() tsk_used_math(current)
2074 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2075 * __GFP_FS is also cleared as it implies __GFP_IO.
2077 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2079 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2080 flags &= ~(__GFP_IO | __GFP_FS);
2084 static inline unsigned int memalloc_noio_save(void)
2086 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2087 current->flags |= PF_MEMALLOC_NOIO;
2091 static inline void memalloc_noio_restore(unsigned int flags)
2093 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2096 /* Per-process atomic flags. */
2097 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2098 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2099 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2102 #define TASK_PFA_TEST(name, func) \
2103 static inline bool task_##func(struct task_struct *p) \
2104 { return test_bit(PFA_##name, &p->atomic_flags); }
2105 #define TASK_PFA_SET(name, func) \
2106 static inline void task_set_##func(struct task_struct *p) \
2107 { set_bit(PFA_##name, &p->atomic_flags); }
2108 #define TASK_PFA_CLEAR(name, func) \
2109 static inline void task_clear_##func(struct task_struct *p) \
2110 { clear_bit(PFA_##name, &p->atomic_flags); }
2112 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2113 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2115 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2116 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2117 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2119 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2120 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2121 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2124 * task->jobctl flags
2126 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2128 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2129 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2130 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2131 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2132 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2133 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2134 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2136 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2137 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2138 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2139 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2140 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2141 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2142 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2144 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2145 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2147 extern bool task_set_jobctl_pending(struct task_struct *task,
2148 unsigned long mask);
2149 extern void task_clear_jobctl_trapping(struct task_struct *task);
2150 extern void task_clear_jobctl_pending(struct task_struct *task,
2151 unsigned long mask);
2153 static inline void rcu_copy_process(struct task_struct *p)
2155 #ifdef CONFIG_PREEMPT_RCU
2156 p->rcu_read_lock_nesting = 0;
2157 p->rcu_read_unlock_special.s = 0;
2158 p->rcu_blocked_node = NULL;
2159 INIT_LIST_HEAD(&p->rcu_node_entry);
2160 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2161 #ifdef CONFIG_TASKS_RCU
2162 p->rcu_tasks_holdout = false;
2163 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2164 p->rcu_tasks_idle_cpu = -1;
2165 #endif /* #ifdef CONFIG_TASKS_RCU */
2168 static inline void tsk_restore_flags(struct task_struct *task,
2169 unsigned long orig_flags, unsigned long flags)
2171 task->flags &= ~flags;
2172 task->flags |= orig_flags & flags;
2175 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2176 const struct cpumask *trial);
2177 extern int task_can_attach(struct task_struct *p,
2178 const struct cpumask *cs_cpus_allowed);
2180 extern void do_set_cpus_allowed(struct task_struct *p,
2181 const struct cpumask *new_mask);
2183 extern int set_cpus_allowed_ptr(struct task_struct *p,
2184 const struct cpumask *new_mask);
2186 static inline void do_set_cpus_allowed(struct task_struct *p,
2187 const struct cpumask *new_mask)
2190 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2191 const struct cpumask *new_mask)
2193 if (!cpumask_test_cpu(0, new_mask))
2199 #ifdef CONFIG_NO_HZ_COMMON
2200 void calc_load_enter_idle(void);
2201 void calc_load_exit_idle(void);
2203 static inline void calc_load_enter_idle(void) { }
2204 static inline void calc_load_exit_idle(void) { }
2205 #endif /* CONFIG_NO_HZ_COMMON */
2207 #ifndef CONFIG_CPUMASK_OFFSTACK
2208 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2210 return set_cpus_allowed_ptr(p, &new_mask);
2215 * Do not use outside of architecture code which knows its limitations.
2217 * sched_clock() has no promise of monotonicity or bounded drift between
2218 * CPUs, use (which you should not) requires disabling IRQs.
2220 * Please use one of the three interfaces below.
2222 extern unsigned long long notrace sched_clock(void);
2224 * See the comment in kernel/sched/clock.c
2226 extern u64 cpu_clock(int cpu);
2227 extern u64 local_clock(void);
2228 extern u64 running_clock(void);
2229 extern u64 sched_clock_cpu(int cpu);
2232 extern void sched_clock_init(void);
2234 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2235 static inline void sched_clock_tick(void)
2239 static inline void sched_clock_idle_sleep_event(void)
2243 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2248 * Architectures can set this to 1 if they have specified
2249 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2250 * but then during bootup it turns out that sched_clock()
2251 * is reliable after all:
2253 extern int sched_clock_stable(void);
2254 extern void set_sched_clock_stable(void);
2255 extern void clear_sched_clock_stable(void);
2257 extern void sched_clock_tick(void);
2258 extern void sched_clock_idle_sleep_event(void);
2259 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2262 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2264 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2265 * The reason for this explicit opt-in is not to have perf penalty with
2266 * slow sched_clocks.
2268 extern void enable_sched_clock_irqtime(void);
2269 extern void disable_sched_clock_irqtime(void);
2271 static inline void enable_sched_clock_irqtime(void) {}
2272 static inline void disable_sched_clock_irqtime(void) {}
2275 extern unsigned long long
2276 task_sched_runtime(struct task_struct *task);
2278 /* sched_exec is called by processes performing an exec */
2280 extern void sched_exec(void);
2282 #define sched_exec() {}
2285 extern void sched_clock_idle_sleep_event(void);
2286 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2288 #ifdef CONFIG_HOTPLUG_CPU
2289 extern void idle_task_exit(void);
2291 static inline void idle_task_exit(void) {}
2294 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2295 extern void wake_up_nohz_cpu(int cpu);
2297 static inline void wake_up_nohz_cpu(int cpu) { }
2300 #ifdef CONFIG_NO_HZ_FULL
2301 extern bool sched_can_stop_tick(void);
2302 extern u64 scheduler_tick_max_deferment(void);
2304 static inline bool sched_can_stop_tick(void) { return false; }
2307 #ifdef CONFIG_SCHED_AUTOGROUP
2308 extern void sched_autogroup_create_attach(struct task_struct *p);
2309 extern void sched_autogroup_detach(struct task_struct *p);
2310 extern void sched_autogroup_fork(struct signal_struct *sig);
2311 extern void sched_autogroup_exit(struct signal_struct *sig);
2312 #ifdef CONFIG_PROC_FS
2313 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2314 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2317 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2318 static inline void sched_autogroup_detach(struct task_struct *p) { }
2319 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2320 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2323 extern int yield_to(struct task_struct *p, bool preempt);
2324 extern void set_user_nice(struct task_struct *p, long nice);
2325 extern int task_prio(const struct task_struct *p);
2327 * task_nice - return the nice value of a given task.
2328 * @p: the task in question.
2330 * Return: The nice value [ -20 ... 0 ... 19 ].
2332 static inline int task_nice(const struct task_struct *p)
2334 return PRIO_TO_NICE((p)->static_prio);
2336 extern int can_nice(const struct task_struct *p, const int nice);
2337 extern int task_curr(const struct task_struct *p);
2338 extern int idle_cpu(int cpu);
2339 extern int sched_setscheduler(struct task_struct *, int,
2340 const struct sched_param *);
2341 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2342 const struct sched_param *);
2343 extern int sched_setattr(struct task_struct *,
2344 const struct sched_attr *);
2345 extern struct task_struct *idle_task(int cpu);
2347 * is_idle_task - is the specified task an idle task?
2348 * @p: the task in question.
2350 * Return: 1 if @p is an idle task. 0 otherwise.
2352 static inline bool is_idle_task(const struct task_struct *p)
2356 extern struct task_struct *curr_task(int cpu);
2357 extern void set_curr_task(int cpu, struct task_struct *p);
2361 union thread_union {
2362 struct thread_info thread_info;
2363 unsigned long stack[THREAD_SIZE/sizeof(long)];
2366 #ifndef __HAVE_ARCH_KSTACK_END
2367 static inline int kstack_end(void *addr)
2369 /* Reliable end of stack detection:
2370 * Some APM bios versions misalign the stack
2372 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2376 extern union thread_union init_thread_union;
2377 extern struct task_struct init_task;
2379 extern struct mm_struct init_mm;
2381 extern struct pid_namespace init_pid_ns;
2384 * find a task by one of its numerical ids
2386 * find_task_by_pid_ns():
2387 * finds a task by its pid in the specified namespace
2388 * find_task_by_vpid():
2389 * finds a task by its virtual pid
2391 * see also find_vpid() etc in include/linux/pid.h
2394 extern struct task_struct *find_task_by_vpid(pid_t nr);
2395 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2396 struct pid_namespace *ns);
2398 /* per-UID process charging. */
2399 extern struct user_struct * alloc_uid(kuid_t);
2400 static inline struct user_struct *get_uid(struct user_struct *u)
2402 atomic_inc(&u->__count);
2405 extern void free_uid(struct user_struct *);
2407 #include <asm/current.h>
2409 extern void xtime_update(unsigned long ticks);
2411 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2412 extern int wake_up_process(struct task_struct *tsk);
2413 extern void wake_up_new_task(struct task_struct *tsk);
2415 extern void kick_process(struct task_struct *tsk);
2417 static inline void kick_process(struct task_struct *tsk) { }
2419 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2420 extern void sched_dead(struct task_struct *p);
2422 extern void proc_caches_init(void);
2423 extern void flush_signals(struct task_struct *);
2424 extern void ignore_signals(struct task_struct *);
2425 extern void flush_signal_handlers(struct task_struct *, int force_default);
2426 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2428 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2430 unsigned long flags;
2433 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2434 ret = dequeue_signal(tsk, mask, info);
2435 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2440 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2442 extern void unblock_all_signals(void);
2443 extern void release_task(struct task_struct * p);
2444 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2445 extern int force_sigsegv(int, struct task_struct *);
2446 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2447 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2448 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2449 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2450 const struct cred *, u32);
2451 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2452 extern int kill_pid(struct pid *pid, int sig, int priv);
2453 extern int kill_proc_info(int, struct siginfo *, pid_t);
2454 extern __must_check bool do_notify_parent(struct task_struct *, int);
2455 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2456 extern void force_sig(int, struct task_struct *);
2457 extern int send_sig(int, struct task_struct *, int);
2458 extern int zap_other_threads(struct task_struct *p);
2459 extern struct sigqueue *sigqueue_alloc(void);
2460 extern void sigqueue_free(struct sigqueue *);
2461 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2462 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2464 static inline void restore_saved_sigmask(void)
2466 if (test_and_clear_restore_sigmask())
2467 __set_current_blocked(¤t->saved_sigmask);
2470 static inline sigset_t *sigmask_to_save(void)
2472 sigset_t *res = ¤t->blocked;
2473 if (unlikely(test_restore_sigmask()))
2474 res = ¤t->saved_sigmask;
2478 static inline int kill_cad_pid(int sig, int priv)
2480 return kill_pid(cad_pid, sig, priv);
2483 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2484 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2485 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2486 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2489 * True if we are on the alternate signal stack.
2491 static inline int on_sig_stack(unsigned long sp)
2493 #ifdef CONFIG_STACK_GROWSUP
2494 return sp >= current->sas_ss_sp &&
2495 sp - current->sas_ss_sp < current->sas_ss_size;
2497 return sp > current->sas_ss_sp &&
2498 sp - current->sas_ss_sp <= current->sas_ss_size;
2502 static inline int sas_ss_flags(unsigned long sp)
2504 if (!current->sas_ss_size)
2507 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2510 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2512 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2513 #ifdef CONFIG_STACK_GROWSUP
2514 return current->sas_ss_sp;
2516 return current->sas_ss_sp + current->sas_ss_size;
2522 * Routines for handling mm_structs
2524 extern struct mm_struct * mm_alloc(void);
2526 /* mmdrop drops the mm and the page tables */
2527 extern void __mmdrop(struct mm_struct *);
2528 static inline void mmdrop(struct mm_struct * mm)
2530 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2534 /* mmput gets rid of the mappings and all user-space */
2535 extern void mmput(struct mm_struct *);
2536 /* Grab a reference to a task's mm, if it is not already going away */
2537 extern struct mm_struct *get_task_mm(struct task_struct *task);
2539 * Grab a reference to a task's mm, if it is not already going away
2540 * and ptrace_may_access with the mode parameter passed to it
2543 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2544 /* Remove the current tasks stale references to the old mm_struct */
2545 extern void mm_release(struct task_struct *, struct mm_struct *);
2547 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2548 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2549 struct task_struct *, unsigned long);
2551 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2552 struct task_struct *);
2554 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2555 * via pt_regs, so ignore the tls argument passed via C. */
2556 static inline int copy_thread_tls(
2557 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2558 struct task_struct *p, unsigned long tls)
2560 return copy_thread(clone_flags, sp, arg, p);
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 *, unsigned long);
2582 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2583 struct task_struct *fork_idle(int);
2584 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2586 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2587 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2589 __set_task_comm(tsk, from, false);
2591 extern char *get_task_comm(char *to, struct task_struct *tsk);
2594 void scheduler_ipi(void);
2595 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2597 static inline void scheduler_ipi(void) { }
2598 static inline unsigned long wait_task_inactive(struct task_struct *p,
2605 #define tasklist_empty() \
2606 list_empty(&init_task.tasks)
2608 #define next_task(p) \
2609 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2611 #define for_each_process(p) \
2612 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2614 extern bool current_is_single_threaded(void);
2617 * Careful: do_each_thread/while_each_thread is a double loop so
2618 * 'break' will not work as expected - use goto instead.
2620 #define do_each_thread(g, t) \
2621 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2623 #define while_each_thread(g, t) \
2624 while ((t = next_thread(t)) != g)
2626 #define __for_each_thread(signal, t) \
2627 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2629 #define for_each_thread(p, t) \
2630 __for_each_thread((p)->signal, t)
2632 /* Careful: this is a double loop, 'break' won't work as expected. */
2633 #define for_each_process_thread(p, t) \
2634 for_each_process(p) for_each_thread(p, t)
2636 static inline int get_nr_threads(struct task_struct *tsk)
2638 return tsk->signal->nr_threads;
2641 static inline bool thread_group_leader(struct task_struct *p)
2643 return p->exit_signal >= 0;
2646 /* Do to the insanities of de_thread it is possible for a process
2647 * to have the pid of the thread group leader without actually being
2648 * the thread group leader. For iteration through the pids in proc
2649 * all we care about is that we have a task with the appropriate
2650 * pid, we don't actually care if we have the right task.
2652 static inline bool has_group_leader_pid(struct task_struct *p)
2654 return task_pid(p) == p->signal->leader_pid;
2658 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2660 return p1->signal == p2->signal;
2663 static inline struct task_struct *next_thread(const struct task_struct *p)
2665 return list_entry_rcu(p->thread_group.next,
2666 struct task_struct, thread_group);
2669 static inline int thread_group_empty(struct task_struct *p)
2671 return list_empty(&p->thread_group);
2674 #define delay_group_leader(p) \
2675 (thread_group_leader(p) && !thread_group_empty(p))
2678 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2679 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2680 * pins the final release of task.io_context. Also protects ->cpuset and
2681 * ->cgroup.subsys[]. And ->vfork_done.
2683 * Nests both inside and outside of read_lock(&tasklist_lock).
2684 * It must not be nested with write_lock_irq(&tasklist_lock),
2685 * neither inside nor outside.
2687 static inline void task_lock(struct task_struct *p)
2689 spin_lock(&p->alloc_lock);
2692 static inline void task_unlock(struct task_struct *p)
2694 spin_unlock(&p->alloc_lock);
2697 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2698 unsigned long *flags);
2700 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2701 unsigned long *flags)
2703 struct sighand_struct *ret;
2705 ret = __lock_task_sighand(tsk, flags);
2706 (void)__cond_lock(&tsk->sighand->siglock, ret);
2710 static inline void unlock_task_sighand(struct task_struct *tsk,
2711 unsigned long *flags)
2713 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2717 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2718 * @tsk: task causing the changes
2720 * All operations which modify a threadgroup - a new thread joining the
2721 * group, death of a member thread (the assertion of PF_EXITING) and
2722 * exec(2) dethreading the process and replacing the leader - are wrapped
2723 * by threadgroup_change_{begin|end}(). This is to provide a place which
2724 * subsystems needing threadgroup stability can hook into for
2727 static inline void threadgroup_change_begin(struct task_struct *tsk)
2730 cgroup_threadgroup_change_begin(tsk);
2734 * threadgroup_change_end - mark the end of changes to a threadgroup
2735 * @tsk: task causing the changes
2737 * See threadgroup_change_begin().
2739 static inline void threadgroup_change_end(struct task_struct *tsk)
2741 cgroup_threadgroup_change_end(tsk);
2744 #ifndef __HAVE_THREAD_FUNCTIONS
2746 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2747 #define task_stack_page(task) ((task)->stack)
2749 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2751 *task_thread_info(p) = *task_thread_info(org);
2752 task_thread_info(p)->task = p;
2756 * Return the address of the last usable long on the stack.
2758 * When the stack grows down, this is just above the thread
2759 * info struct. Going any lower will corrupt the threadinfo.
2761 * When the stack grows up, this is the highest address.
2762 * Beyond that position, we corrupt data on the next page.
2764 static inline unsigned long *end_of_stack(struct task_struct *p)
2766 #ifdef CONFIG_STACK_GROWSUP
2767 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2769 return (unsigned long *)(task_thread_info(p) + 1);
2774 #define task_stack_end_corrupted(task) \
2775 (*(end_of_stack(task)) != STACK_END_MAGIC)
2777 static inline int object_is_on_stack(void *obj)
2779 void *stack = task_stack_page(current);
2781 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2784 extern void thread_info_cache_init(void);
2786 #ifdef CONFIG_DEBUG_STACK_USAGE
2787 static inline unsigned long stack_not_used(struct task_struct *p)
2789 unsigned long *n = end_of_stack(p);
2791 do { /* Skip over canary */
2795 return (unsigned long)n - (unsigned long)end_of_stack(p);
2798 extern void set_task_stack_end_magic(struct task_struct *tsk);
2800 /* set thread flags in other task's structures
2801 * - see asm/thread_info.h for TIF_xxxx flags available
2803 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2805 set_ti_thread_flag(task_thread_info(tsk), flag);
2808 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2810 clear_ti_thread_flag(task_thread_info(tsk), flag);
2813 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2815 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2818 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2820 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2823 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2825 return test_ti_thread_flag(task_thread_info(tsk), flag);
2828 static inline void set_tsk_need_resched(struct task_struct *tsk)
2830 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2833 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2835 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2838 static inline int test_tsk_need_resched(struct task_struct *tsk)
2840 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2843 static inline int restart_syscall(void)
2845 set_tsk_thread_flag(current, TIF_SIGPENDING);
2846 return -ERESTARTNOINTR;
2849 static inline int signal_pending(struct task_struct *p)
2851 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2854 static inline int __fatal_signal_pending(struct task_struct *p)
2856 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2859 static inline int fatal_signal_pending(struct task_struct *p)
2861 return signal_pending(p) && __fatal_signal_pending(p);
2864 static inline int signal_pending_state(long state, struct task_struct *p)
2866 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2868 if (!signal_pending(p))
2871 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2875 * cond_resched() and cond_resched_lock(): latency reduction via
2876 * explicit rescheduling in places that are safe. The return
2877 * value indicates whether a reschedule was done in fact.
2878 * cond_resched_lock() will drop the spinlock before scheduling,
2879 * cond_resched_softirq() will enable bhs before scheduling.
2881 extern int _cond_resched(void);
2883 #define cond_resched() ({ \
2884 ___might_sleep(__FILE__, __LINE__, 0); \
2888 extern int __cond_resched_lock(spinlock_t *lock);
2890 #ifdef CONFIG_PREEMPT_COUNT
2891 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2893 #define PREEMPT_LOCK_OFFSET 0
2896 #define cond_resched_lock(lock) ({ \
2897 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2898 __cond_resched_lock(lock); \
2901 extern int __cond_resched_softirq(void);
2903 #define cond_resched_softirq() ({ \
2904 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2905 __cond_resched_softirq(); \
2908 static inline void cond_resched_rcu(void)
2910 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2918 * Does a critical section need to be broken due to another
2919 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2920 * but a general need for low latency)
2922 static inline int spin_needbreak(spinlock_t *lock)
2924 #ifdef CONFIG_PREEMPT
2925 return spin_is_contended(lock);
2932 * Idle thread specific functions to determine the need_resched
2935 #ifdef TIF_POLLING_NRFLAG
2936 static inline int tsk_is_polling(struct task_struct *p)
2938 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2941 static inline void __current_set_polling(void)
2943 set_thread_flag(TIF_POLLING_NRFLAG);
2946 static inline bool __must_check current_set_polling_and_test(void)
2948 __current_set_polling();
2951 * Polling state must be visible before we test NEED_RESCHED,
2952 * paired by resched_curr()
2954 smp_mb__after_atomic();
2956 return unlikely(tif_need_resched());
2959 static inline void __current_clr_polling(void)
2961 clear_thread_flag(TIF_POLLING_NRFLAG);
2964 static inline bool __must_check current_clr_polling_and_test(void)
2966 __current_clr_polling();
2969 * Polling state must be visible before we test NEED_RESCHED,
2970 * paired by resched_curr()
2972 smp_mb__after_atomic();
2974 return unlikely(tif_need_resched());
2978 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2979 static inline void __current_set_polling(void) { }
2980 static inline void __current_clr_polling(void) { }
2982 static inline bool __must_check current_set_polling_and_test(void)
2984 return unlikely(tif_need_resched());
2986 static inline bool __must_check current_clr_polling_and_test(void)
2988 return unlikely(tif_need_resched());
2992 static inline void current_clr_polling(void)
2994 __current_clr_polling();
2997 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2998 * Once the bit is cleared, we'll get IPIs with every new
2999 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3002 smp_mb(); /* paired with resched_curr() */
3004 preempt_fold_need_resched();
3007 static __always_inline bool need_resched(void)
3009 return unlikely(tif_need_resched());
3013 * Thread group CPU time accounting.
3015 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3016 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3019 * Reevaluate whether the task has signals pending delivery.
3020 * Wake the task if so.
3021 * This is required every time the blocked sigset_t changes.
3022 * callers must hold sighand->siglock.
3024 extern void recalc_sigpending_and_wake(struct task_struct *t);
3025 extern void recalc_sigpending(void);
3027 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3029 static inline void signal_wake_up(struct task_struct *t, bool resume)
3031 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3033 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3035 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3039 * Wrappers for p->thread_info->cpu access. No-op on UP.
3043 static inline unsigned int task_cpu(const struct task_struct *p)
3045 return task_thread_info(p)->cpu;
3048 static inline int task_node(const struct task_struct *p)
3050 return cpu_to_node(task_cpu(p));
3053 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3057 static inline unsigned int task_cpu(const struct task_struct *p)
3062 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3066 #endif /* CONFIG_SMP */
3068 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3069 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3071 #ifdef CONFIG_CGROUP_SCHED
3072 extern struct task_group root_task_group;
3073 #endif /* CONFIG_CGROUP_SCHED */
3075 extern int task_can_switch_user(struct user_struct *up,
3076 struct task_struct *tsk);
3078 #ifdef CONFIG_TASK_XACCT
3079 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3081 tsk->ioac.rchar += amt;
3084 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3086 tsk->ioac.wchar += amt;
3089 static inline void inc_syscr(struct task_struct *tsk)
3094 static inline void inc_syscw(struct task_struct *tsk)
3099 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3103 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3107 static inline void inc_syscr(struct task_struct *tsk)
3111 static inline void inc_syscw(struct task_struct *tsk)
3116 #ifndef TASK_SIZE_OF
3117 #define TASK_SIZE_OF(tsk) TASK_SIZE
3121 extern void mm_update_next_owner(struct mm_struct *mm);
3123 static inline void mm_update_next_owner(struct mm_struct *mm)
3126 #endif /* CONFIG_MEMCG */
3128 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3131 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3134 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3137 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3140 static inline unsigned long rlimit(unsigned int limit)
3142 return task_rlimit(current, limit);
3145 static inline unsigned long rlimit_max(unsigned int limit)
3147 return task_rlimit_max(current, limit);