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>
29 #include <asm/kmap_types.h>
32 #include <asm/ptrace.h>
33 #include <linux/cputime.h>
35 #include <linux/smp.h>
36 #include <linux/sem.h>
37 #include <linux/shm.h>
38 #include <linux/signal.h>
39 #include <linux/compiler.h>
40 #include <linux/completion.h>
41 #include <linux/pid.h>
42 #include <linux/percpu.h>
43 #include <linux/topology.h>
44 #include <linux/proportions.h>
45 #include <linux/seccomp.h>
46 #include <linux/rcupdate.h>
47 #include <linux/rculist.h>
48 #include <linux/rtmutex.h>
50 #include <linux/time.h>
51 #include <linux/param.h>
52 #include <linux/resource.h>
53 #include <linux/timer.h>
54 #include <linux/hrtimer.h>
55 #include <linux/task_io_accounting.h>
56 #include <linux/latencytop.h>
57 #include <linux/cred.h>
58 #include <linux/llist.h>
59 #include <linux/uidgid.h>
60 #include <linux/gfp.h>
61 #include <linux/magic.h>
62 #include <linux/cgroup-defs.h>
64 #include <asm/processor.h>
66 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
69 * Extended scheduling parameters data structure.
71 * This is needed because the original struct sched_param can not be
72 * altered without introducing ABI issues with legacy applications
73 * (e.g., in sched_getparam()).
75 * However, the possibility of specifying more than just a priority for
76 * the tasks may be useful for a wide variety of application fields, e.g.,
77 * multimedia, streaming, automation and control, and many others.
79 * This variant (sched_attr) is meant at describing a so-called
80 * sporadic time-constrained task. In such model a task is specified by:
81 * - the activation period or minimum instance inter-arrival time;
82 * - the maximum (or average, depending on the actual scheduling
83 * discipline) computation time of all instances, a.k.a. runtime;
84 * - the deadline (relative to the actual activation time) of each
86 * Very briefly, a periodic (sporadic) task asks for the execution of
87 * some specific computation --which is typically called an instance--
88 * (at most) every period. Moreover, each instance typically lasts no more
89 * than the runtime and must be completed by time instant t equal to
90 * the instance activation time + the deadline.
92 * This is reflected by the actual fields of the sched_attr structure:
94 * @size size of the structure, for fwd/bwd compat.
96 * @sched_policy task's scheduling policy
97 * @sched_flags for customizing the scheduler behaviour
98 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
99 * @sched_priority task's static priority (SCHED_FIFO/RR)
100 * @sched_deadline representative of the task's deadline
101 * @sched_runtime representative of the task's runtime
102 * @sched_period representative of the task's period
104 * Given this task model, there are a multiplicity of scheduling algorithms
105 * and policies, that can be used to ensure all the tasks will make their
106 * timing constraints.
108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
109 * only user of this new interface. More information about the algorithm
110 * available in the scheduling class file or in Documentation/.
118 /* SCHED_NORMAL, SCHED_BATCH */
121 /* SCHED_FIFO, SCHED_RR */
130 struct futex_pi_state;
131 struct robust_list_head;
134 struct perf_event_context;
139 #define VMACACHE_BITS 2
140 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
141 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
144 * These are the constant used to fake the fixed-point load-average
145 * counting. Some notes:
146 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
147 * a load-average precision of 10 bits integer + 11 bits fractional
148 * - if you want to count load-averages more often, you need more
149 * precision, or rounding will get you. With 2-second counting freq,
150 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 extern unsigned long avenrun[]; /* Load averages */
154 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
156 #define FSHIFT 11 /* nr of bits of precision */
157 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
158 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
159 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
160 #define EXP_5 2014 /* 1/exp(5sec/5min) */
161 #define EXP_15 2037 /* 1/exp(5sec/15min) */
163 #define CALC_LOAD(load,exp,n) \
165 load += n*(FIXED_1-exp); \
168 extern unsigned long total_forks;
169 extern int nr_threads;
170 DECLARE_PER_CPU(unsigned long, process_counts);
171 extern int nr_processes(void);
172 extern unsigned long nr_running(void);
173 extern bool single_task_running(void);
174 extern unsigned long nr_iowait(void);
175 extern unsigned long nr_iowait_cpu(int cpu);
176 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
178 extern void calc_global_load(unsigned long ticks);
180 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
181 extern void update_cpu_load_nohz(void);
183 static inline void update_cpu_load_nohz(void) { }
186 extern void dump_cpu_task(int cpu);
191 #ifdef CONFIG_SCHED_DEBUG
192 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
193 extern void proc_sched_set_task(struct task_struct *p);
197 * Task state bitmask. NOTE! These bits are also
198 * encoded in fs/proc/array.c: get_task_state().
200 * We have two separate sets of flags: task->state
201 * is about runnability, while task->exit_state are
202 * about the task exiting. Confusing, but this way
203 * modifying one set can't modify the other one by
206 #define TASK_RUNNING 0
207 #define TASK_INTERRUPTIBLE 1
208 #define TASK_UNINTERRUPTIBLE 2
209 #define __TASK_STOPPED 4
210 #define __TASK_TRACED 8
211 /* in tsk->exit_state */
213 #define EXIT_ZOMBIE 32
214 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
215 /* in tsk->state again */
217 #define TASK_WAKEKILL 128
218 #define TASK_WAKING 256
219 #define TASK_PARKED 512
220 #define TASK_NOLOAD 1024
221 #define TASK_STATE_MAX 2048
223 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
225 extern char ___assert_task_state[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
233 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
235 /* Convenience macros for the sake of wake_up */
236 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
237 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
239 /* get_task_state() */
240 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
241 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
242 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
244 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
245 #define task_contributes_to_load(task) \
246 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
247 (task->flags & PF_FROZEN) == 0 && \
248 (task->state & TASK_NOLOAD) == 0)
250 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
252 #define __set_task_state(tsk, state_value) \
254 (tsk)->task_state_change = _THIS_IP_; \
255 (tsk)->state = (state_value); \
257 #define set_task_state(tsk, state_value) \
259 (tsk)->task_state_change = _THIS_IP_; \
260 smp_store_mb((tsk)->state, (state_value)); \
264 * set_current_state() includes a barrier so that the write of current->state
265 * is correctly serialised wrt the caller's subsequent test of whether to
268 * set_current_state(TASK_UNINTERRUPTIBLE);
269 * if (do_i_need_to_sleep())
272 * If the caller does not need such serialisation then use __set_current_state()
274 #define __set_current_state(state_value) \
276 current->task_state_change = _THIS_IP_; \
277 current->state = (state_value); \
279 #define set_current_state(state_value) \
281 current->task_state_change = _THIS_IP_; \
282 smp_store_mb(current->state, (state_value)); \
287 #define __set_task_state(tsk, state_value) \
288 do { (tsk)->state = (state_value); } while (0)
289 #define set_task_state(tsk, state_value) \
290 smp_store_mb((tsk)->state, (state_value))
293 * set_current_state() includes a barrier so that the write of current->state
294 * is correctly serialised wrt the caller's subsequent test of whether to
297 * set_current_state(TASK_UNINTERRUPTIBLE);
298 * if (do_i_need_to_sleep())
301 * If the caller does not need such serialisation then use __set_current_state()
303 #define __set_current_state(state_value) \
304 do { current->state = (state_value); } while (0)
305 #define set_current_state(state_value) \
306 smp_store_mb(current->state, (state_value))
310 #define __set_current_state_no_track(state_value) \
311 do { current->state = (state_value); } while (0)
312 #define set_current_state_no_track(state_value) \
313 set_mb(current->state, (state_value))
315 /* Task command name length */
316 #define TASK_COMM_LEN 16
318 #include <linux/spinlock.h>
321 * This serializes "schedule()" and also protects
322 * the run-queue from deletions/modifications (but
323 * _adding_ to the beginning of the run-queue has
326 extern rwlock_t tasklist_lock;
327 extern spinlock_t mmlist_lock;
331 #ifdef CONFIG_PROVE_RCU
332 extern int lockdep_tasklist_lock_is_held(void);
333 #endif /* #ifdef CONFIG_PROVE_RCU */
335 extern void sched_init(void);
336 extern void sched_init_smp(void);
337 extern asmlinkage void schedule_tail(struct task_struct *prev);
338 extern void init_idle(struct task_struct *idle, int cpu);
339 extern void init_idle_bootup_task(struct task_struct *idle);
341 extern cpumask_var_t cpu_isolated_map;
343 extern int runqueue_is_locked(int cpu);
345 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
346 extern void nohz_balance_enter_idle(int cpu);
347 extern void set_cpu_sd_state_idle(void);
348 extern int get_nohz_timer_target(void);
350 static inline void nohz_balance_enter_idle(int cpu) { }
351 static inline void set_cpu_sd_state_idle(void) { }
355 * Only dump TASK_* tasks. (0 for all tasks)
357 extern void show_state_filter(unsigned long state_filter);
359 static inline void show_state(void)
361 show_state_filter(0);
364 extern void show_regs(struct pt_regs *);
367 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
368 * task), SP is the stack pointer of the first frame that should be shown in the back
369 * trace (or NULL if the entire call-chain of the task should be shown).
371 extern void show_stack(struct task_struct *task, unsigned long *sp);
373 extern void cpu_init (void);
374 extern void trap_init(void);
375 extern void update_process_times(int user);
376 extern void scheduler_tick(void);
378 extern void sched_show_task(struct task_struct *p);
380 #ifdef CONFIG_LOCKUP_DETECTOR
381 extern void touch_softlockup_watchdog(void);
382 extern void touch_softlockup_watchdog_sync(void);
383 extern void touch_all_softlockup_watchdogs(void);
384 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
386 size_t *lenp, loff_t *ppos);
387 extern unsigned int softlockup_panic;
388 extern unsigned int hardlockup_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
488 #define MMF_DUMP_DAX_PRIVATE 9
489 #define MMF_DUMP_DAX_SHARED 10
491 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
492 #define MMF_DUMP_FILTER_BITS 9
493 #define MMF_DUMP_FILTER_MASK \
494 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
495 #define MMF_DUMP_FILTER_DEFAULT \
496 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
497 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
499 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
500 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
502 # define MMF_DUMP_MASK_DEFAULT_ELF 0
504 /* leave room for more dump flags */
505 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
506 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
507 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
509 #define MMF_HAS_UPROBES 19 /* has uprobes */
510 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
512 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
514 struct sighand_struct {
516 struct k_sigaction action[_NSIG];
518 wait_queue_head_t signalfd_wqh;
521 struct pacct_struct {
524 unsigned long ac_mem;
525 cputime_t ac_utime, ac_stime;
526 unsigned long ac_minflt, ac_majflt;
537 * struct prev_cputime - snaphsot of system and user cputime
538 * @utime: time spent in user mode
539 * @stime: time spent in system mode
540 * @lock: protects the above two fields
542 * Stores previous user/system time values such that we can guarantee
545 struct prev_cputime {
546 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
553 static inline void prev_cputime_init(struct prev_cputime *prev)
555 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
556 prev->utime = prev->stime = 0;
557 raw_spin_lock_init(&prev->lock);
562 * struct task_cputime - collected CPU time counts
563 * @utime: time spent in user mode, in &cputime_t units
564 * @stime: time spent in kernel mode, in &cputime_t units
565 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
567 * This structure groups together three kinds of CPU time that are tracked for
568 * threads and thread groups. Most things considering CPU time want to group
569 * these counts together and treat all three of them in parallel.
571 struct task_cputime {
574 unsigned long long sum_exec_runtime;
577 /* Alternate field names when used to cache expirations. */
578 #define virt_exp utime
579 #define prof_exp stime
580 #define sched_exp sum_exec_runtime
582 #define INIT_CPUTIME \
583 (struct task_cputime) { \
586 .sum_exec_runtime = 0, \
590 * This is the atomic variant of task_cputime, which can be used for
591 * storing and updating task_cputime statistics without locking.
593 struct task_cputime_atomic {
596 atomic64_t sum_exec_runtime;
599 #define INIT_CPUTIME_ATOMIC \
600 (struct task_cputime_atomic) { \
601 .utime = ATOMIC64_INIT(0), \
602 .stime = ATOMIC64_INIT(0), \
603 .sum_exec_runtime = ATOMIC64_INIT(0), \
606 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
609 * Disable preemption until the scheduler is running -- use an unconditional
610 * value so that it also works on !PREEMPT_COUNT kernels.
612 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
614 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
617 * Initial preempt_count value; reflects the preempt_count schedule invariant
618 * which states that during context switches:
620 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
622 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
623 * Note: See finish_task_switch().
625 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
628 * struct thread_group_cputimer - thread group interval timer counts
629 * @cputime_atomic: atomic thread group interval timers.
630 * @running: true when there are timers running and
631 * @cputime_atomic receives updates.
632 * @checking_timer: true when a thread in the group is in the
633 * process of checking for thread group timers.
635 * This structure contains the version of task_cputime, above, that is
636 * used for thread group CPU timer calculations.
638 struct thread_group_cputimer {
639 struct task_cputime_atomic cputime_atomic;
644 #include <linux/rwsem.h>
648 * NOTE! "signal_struct" does not have its own
649 * locking, because a shared signal_struct always
650 * implies a shared sighand_struct, so locking
651 * sighand_struct is always a proper superset of
652 * the locking of signal_struct.
654 struct signal_struct {
658 struct list_head thread_head;
660 wait_queue_head_t wait_chldexit; /* for wait4() */
662 /* current thread group signal load-balancing target: */
663 struct task_struct *curr_target;
665 /* shared signal handling: */
666 struct sigpending shared_pending;
668 /* thread group exit support */
671 * - notify group_exit_task when ->count is equal to notify_count
672 * - everyone except group_exit_task is stopped during signal delivery
673 * of fatal signals, group_exit_task processes the signal.
676 struct task_struct *group_exit_task;
678 /* thread group stop support, overloads group_exit_code too */
679 int group_stop_count;
680 unsigned int flags; /* see SIGNAL_* flags below */
683 * PR_SET_CHILD_SUBREAPER marks a process, like a service
684 * manager, to re-parent orphan (double-forking) child processes
685 * to this process instead of 'init'. The service manager is
686 * able to receive SIGCHLD signals and is able to investigate
687 * the process until it calls wait(). All children of this
688 * process will inherit a flag if they should look for a
689 * child_subreaper process at exit.
691 unsigned int is_child_subreaper:1;
692 unsigned int has_child_subreaper:1;
694 /* POSIX.1b Interval Timers */
696 struct list_head posix_timers;
698 /* ITIMER_REAL timer for the process */
699 struct hrtimer real_timer;
700 struct pid *leader_pid;
701 ktime_t it_real_incr;
704 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
705 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
706 * values are defined to 0 and 1 respectively
708 struct cpu_itimer it[2];
711 * Thread group totals for process CPU timers.
712 * See thread_group_cputimer(), et al, for details.
714 struct thread_group_cputimer cputimer;
716 /* Earliest-expiration cache. */
717 struct task_cputime cputime_expires;
719 struct list_head cpu_timers[3];
721 struct pid *tty_old_pgrp;
723 /* boolean value for session group leader */
726 struct tty_struct *tty; /* NULL if no tty */
728 #ifdef CONFIG_SCHED_AUTOGROUP
729 struct autogroup *autogroup;
732 * Cumulative resource counters for dead threads in the group,
733 * and for reaped dead child processes forked by this group.
734 * Live threads maintain their own counters and add to these
735 * in __exit_signal, except for the group leader.
737 seqlock_t stats_lock;
738 cputime_t utime, stime, cutime, cstime;
741 struct prev_cputime prev_cputime;
742 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
743 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
744 unsigned long inblock, oublock, cinblock, coublock;
745 unsigned long maxrss, cmaxrss;
746 struct task_io_accounting ioac;
749 * Cumulative ns of schedule CPU time fo dead threads in the
750 * group, not including a zombie group leader, (This only differs
751 * from jiffies_to_ns(utime + stime) if sched_clock uses something
752 * other than jiffies.)
754 unsigned long long sum_sched_runtime;
757 * We don't bother to synchronize most readers of this at all,
758 * because there is no reader checking a limit that actually needs
759 * to get both rlim_cur and rlim_max atomically, and either one
760 * alone is a single word that can safely be read normally.
761 * getrlimit/setrlimit use task_lock(current->group_leader) to
762 * protect this instead of the siglock, because they really
763 * have no need to disable irqs.
765 struct rlimit rlim[RLIM_NLIMITS];
767 #ifdef CONFIG_BSD_PROCESS_ACCT
768 struct pacct_struct pacct; /* per-process accounting information */
770 #ifdef CONFIG_TASKSTATS
771 struct taskstats *stats;
775 unsigned audit_tty_log_passwd;
776 struct tty_audit_buf *tty_audit_buf;
779 oom_flags_t oom_flags;
780 short oom_score_adj; /* OOM kill score adjustment */
781 short oom_score_adj_min; /* OOM kill score adjustment min value.
782 * Only settable by CAP_SYS_RESOURCE. */
784 struct mutex cred_guard_mutex; /* guard against foreign influences on
785 * credential calculations
786 * (notably. ptrace) */
790 * Bits in flags field of signal_struct.
792 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
793 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
794 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
795 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
797 * Pending notifications to parent.
799 #define SIGNAL_CLD_STOPPED 0x00000010
800 #define SIGNAL_CLD_CONTINUED 0x00000020
801 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
803 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
805 /* If true, all threads except ->group_exit_task have pending SIGKILL */
806 static inline int signal_group_exit(const struct signal_struct *sig)
808 return (sig->flags & SIGNAL_GROUP_EXIT) ||
809 (sig->group_exit_task != NULL);
813 * Some day this will be a full-fledged user tracking system..
816 atomic_t __count; /* reference count */
817 atomic_t processes; /* How many processes does this user have? */
818 atomic_t sigpending; /* How many pending signals does this user have? */
819 #ifdef CONFIG_INOTIFY_USER
820 atomic_t inotify_watches; /* How many inotify watches does this user have? */
821 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
823 #ifdef CONFIG_FANOTIFY
824 atomic_t fanotify_listeners;
827 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
829 #ifdef CONFIG_POSIX_MQUEUE
830 /* protected by mq_lock */
831 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
833 unsigned long locked_shm; /* How many pages of mlocked shm ? */
834 unsigned long unix_inflight; /* How many files in flight in unix sockets */
837 struct key *uid_keyring; /* UID specific keyring */
838 struct key *session_keyring; /* UID's default session keyring */
841 /* Hash table maintenance information */
842 struct hlist_node uidhash_node;
845 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
846 atomic_long_t locked_vm;
850 extern int uids_sysfs_init(void);
852 extern struct user_struct *find_user(kuid_t);
854 extern struct user_struct root_user;
855 #define INIT_USER (&root_user)
858 struct backing_dev_info;
859 struct reclaim_state;
861 #ifdef CONFIG_SCHED_INFO
863 /* cumulative counters */
864 unsigned long pcount; /* # of times run on this cpu */
865 unsigned long long run_delay; /* time spent waiting on a runqueue */
868 unsigned long long last_arrival,/* when we last ran on a cpu */
869 last_queued; /* when we were last queued to run */
871 #endif /* CONFIG_SCHED_INFO */
873 #ifdef CONFIG_TASK_DELAY_ACCT
874 struct task_delay_info {
876 unsigned int flags; /* Private per-task flags */
878 /* For each stat XXX, add following, aligned appropriately
880 * struct timespec XXX_start, XXX_end;
884 * Atomicity of updates to XXX_delay, XXX_count protected by
885 * single lock above (split into XXX_lock if contention is an issue).
889 * XXX_count is incremented on every XXX operation, the delay
890 * associated with the operation is added to XXX_delay.
891 * XXX_delay contains the accumulated delay time in nanoseconds.
893 u64 blkio_start; /* Shared by blkio, swapin */
894 u64 blkio_delay; /* wait for sync block io completion */
895 u64 swapin_delay; /* wait for swapin block io completion */
896 u32 blkio_count; /* total count of the number of sync block */
897 /* io operations performed */
898 u32 swapin_count; /* total count of the number of swapin block */
899 /* io operations performed */
902 u64 freepages_delay; /* wait for memory reclaim */
903 u32 freepages_count; /* total count of memory reclaim */
905 #endif /* CONFIG_TASK_DELAY_ACCT */
907 static inline int sched_info_on(void)
909 #ifdef CONFIG_SCHEDSTATS
911 #elif defined(CONFIG_TASK_DELAY_ACCT)
912 extern int delayacct_on;
927 * Increase resolution of cpu_capacity calculations
929 #define SCHED_CAPACITY_SHIFT 10
930 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
933 * Wake-queues are lists of tasks with a pending wakeup, whose
934 * callers have already marked the task as woken internally,
935 * and can thus carry on. A common use case is being able to
936 * do the wakeups once the corresponding user lock as been
939 * We hold reference to each task in the list across the wakeup,
940 * thus guaranteeing that the memory is still valid by the time
941 * the actual wakeups are performed in wake_up_q().
943 * One per task suffices, because there's never a need for a task to be
944 * in two wake queues simultaneously; it is forbidden to abandon a task
945 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
946 * already in a wake queue, the wakeup will happen soon and the second
947 * waker can just skip it.
949 * The WAKE_Q macro declares and initializes the list head.
950 * wake_up_q() does NOT reinitialize the list; it's expected to be
951 * called near the end of a function, where the fact that the queue is
952 * not used again will be easy to see by inspection.
954 * Note that this can cause spurious wakeups. schedule() callers
955 * must ensure the call is done inside a loop, confirming that the
956 * wakeup condition has in fact occurred.
959 struct wake_q_node *next;
963 struct wake_q_node *first;
964 struct wake_q_node **lastp;
967 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
969 #define WAKE_Q(name) \
970 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
972 extern void wake_q_add(struct wake_q_head *head,
973 struct task_struct *task);
974 extern void __wake_up_q(struct wake_q_head *head, bool sleeper);
976 static inline void wake_up_q(struct wake_q_head *head)
978 __wake_up_q(head, false);
981 static inline void wake_up_q_sleeper(struct wake_q_head *head)
983 __wake_up_q(head, true);
987 * sched-domains (multiprocessor balancing) declarations:
990 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
991 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
992 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
993 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
994 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
995 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
996 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
997 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
998 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
999 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1000 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1001 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1002 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1003 #define SD_NUMA 0x4000 /* cross-node balancing */
1005 #ifdef CONFIG_SCHED_SMT
1006 static inline int cpu_smt_flags(void)
1008 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1012 #ifdef CONFIG_SCHED_MC
1013 static inline int cpu_core_flags(void)
1015 return SD_SHARE_PKG_RESOURCES;
1020 static inline int cpu_numa_flags(void)
1026 struct sched_domain_attr {
1027 int relax_domain_level;
1030 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1031 .relax_domain_level = -1, \
1034 extern int sched_domain_level_max;
1038 struct sched_domain {
1039 /* These fields must be setup */
1040 struct sched_domain *parent; /* top domain must be null terminated */
1041 struct sched_domain *child; /* bottom domain must be null terminated */
1042 struct sched_group *groups; /* the balancing groups of the domain */
1043 unsigned long min_interval; /* Minimum balance interval ms */
1044 unsigned long max_interval; /* Maximum balance interval ms */
1045 unsigned int busy_factor; /* less balancing by factor if busy */
1046 unsigned int imbalance_pct; /* No balance until over watermark */
1047 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1048 unsigned int busy_idx;
1049 unsigned int idle_idx;
1050 unsigned int newidle_idx;
1051 unsigned int wake_idx;
1052 unsigned int forkexec_idx;
1053 unsigned int smt_gain;
1055 int nohz_idle; /* NOHZ IDLE status */
1056 int flags; /* See SD_* */
1059 /* Runtime fields. */
1060 unsigned long last_balance; /* init to jiffies. units in jiffies */
1061 unsigned int balance_interval; /* initialise to 1. units in ms. */
1062 unsigned int nr_balance_failed; /* initialise to 0 */
1064 /* idle_balance() stats */
1065 u64 max_newidle_lb_cost;
1066 unsigned long next_decay_max_lb_cost;
1068 #ifdef CONFIG_SCHEDSTATS
1069 /* load_balance() stats */
1070 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1071 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1072 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1073 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1074 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1075 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1076 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1077 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1079 /* Active load balancing */
1080 unsigned int alb_count;
1081 unsigned int alb_failed;
1082 unsigned int alb_pushed;
1084 /* SD_BALANCE_EXEC stats */
1085 unsigned int sbe_count;
1086 unsigned int sbe_balanced;
1087 unsigned int sbe_pushed;
1089 /* SD_BALANCE_FORK stats */
1090 unsigned int sbf_count;
1091 unsigned int sbf_balanced;
1092 unsigned int sbf_pushed;
1094 /* try_to_wake_up() stats */
1095 unsigned int ttwu_wake_remote;
1096 unsigned int ttwu_move_affine;
1097 unsigned int ttwu_move_balance;
1099 #ifdef CONFIG_SCHED_DEBUG
1103 void *private; /* used during construction */
1104 struct rcu_head rcu; /* used during destruction */
1107 unsigned int span_weight;
1109 * Span of all CPUs in this domain.
1111 * NOTE: this field is variable length. (Allocated dynamically
1112 * by attaching extra space to the end of the structure,
1113 * depending on how many CPUs the kernel has booted up with)
1115 unsigned long span[0];
1118 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1120 return to_cpumask(sd->span);
1123 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1124 struct sched_domain_attr *dattr_new);
1126 /* Allocate an array of sched domains, for partition_sched_domains(). */
1127 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1128 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1130 bool cpus_share_cache(int this_cpu, int that_cpu);
1132 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1133 typedef int (*sched_domain_flags_f)(void);
1135 #define SDTL_OVERLAP 0x01
1138 struct sched_domain **__percpu sd;
1139 struct sched_group **__percpu sg;
1140 struct sched_group_capacity **__percpu sgc;
1143 struct sched_domain_topology_level {
1144 sched_domain_mask_f mask;
1145 sched_domain_flags_f sd_flags;
1148 struct sd_data data;
1149 #ifdef CONFIG_SCHED_DEBUG
1154 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1155 extern void wake_up_if_idle(int cpu);
1157 #ifdef CONFIG_SCHED_DEBUG
1158 # define SD_INIT_NAME(type) .name = #type
1160 # define SD_INIT_NAME(type)
1163 #else /* CONFIG_SMP */
1165 struct sched_domain_attr;
1168 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1169 struct sched_domain_attr *dattr_new)
1173 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1178 #endif /* !CONFIG_SMP */
1181 struct io_context; /* See blkdev.h */
1184 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1185 extern void prefetch_stack(struct task_struct *t);
1187 static inline void prefetch_stack(struct task_struct *t) { }
1190 struct audit_context; /* See audit.c */
1192 struct pipe_inode_info;
1193 struct uts_namespace;
1195 struct load_weight {
1196 unsigned long weight;
1201 * The load_avg/util_avg accumulates an infinite geometric series.
1202 * 1) load_avg factors frequency scaling into the amount of time that a
1203 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1204 * aggregated such weights of all runnable and blocked sched_entities.
1205 * 2) util_avg factors frequency and cpu scaling into the amount of time
1206 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1207 * For cfs_rq, it is the aggregated such times of all runnable and
1208 * blocked sched_entities.
1209 * The 64 bit load_sum can:
1210 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1211 * the highest weight (=88761) always runnable, we should not overflow
1212 * 2) for entity, support any load.weight always runnable
1215 u64 last_update_time, load_sum;
1216 u32 util_sum, period_contrib;
1217 unsigned long load_avg, util_avg;
1220 #ifdef CONFIG_SCHEDSTATS
1221 struct sched_statistics {
1231 s64 sum_sleep_runtime;
1238 u64 nr_migrations_cold;
1239 u64 nr_failed_migrations_affine;
1240 u64 nr_failed_migrations_running;
1241 u64 nr_failed_migrations_hot;
1242 u64 nr_forced_migrations;
1245 u64 nr_wakeups_sync;
1246 u64 nr_wakeups_migrate;
1247 u64 nr_wakeups_local;
1248 u64 nr_wakeups_remote;
1249 u64 nr_wakeups_affine;
1250 u64 nr_wakeups_affine_attempts;
1251 u64 nr_wakeups_passive;
1252 u64 nr_wakeups_idle;
1256 struct sched_entity {
1257 struct load_weight load; /* for load-balancing */
1258 struct rb_node run_node;
1259 struct list_head group_node;
1263 u64 sum_exec_runtime;
1265 u64 prev_sum_exec_runtime;
1269 #ifdef CONFIG_SCHEDSTATS
1270 struct sched_statistics statistics;
1273 #ifdef CONFIG_FAIR_GROUP_SCHED
1275 struct sched_entity *parent;
1276 /* rq on which this entity is (to be) queued: */
1277 struct cfs_rq *cfs_rq;
1278 /* rq "owned" by this entity/group: */
1279 struct cfs_rq *my_q;
1283 /* Per entity load average tracking */
1284 struct sched_avg avg;
1288 struct sched_rt_entity {
1289 struct list_head run_list;
1290 unsigned long timeout;
1291 unsigned long watchdog_stamp;
1292 unsigned int time_slice;
1294 struct sched_rt_entity *back;
1295 #ifdef CONFIG_RT_GROUP_SCHED
1296 struct sched_rt_entity *parent;
1297 /* rq on which this entity is (to be) queued: */
1298 struct rt_rq *rt_rq;
1299 /* rq "owned" by this entity/group: */
1304 struct sched_dl_entity {
1305 struct rb_node rb_node;
1308 * Original scheduling parameters. Copied here from sched_attr
1309 * during sched_setattr(), they will remain the same until
1310 * the next sched_setattr().
1312 u64 dl_runtime; /* maximum runtime for each instance */
1313 u64 dl_deadline; /* relative deadline of each instance */
1314 u64 dl_period; /* separation of two instances (period) */
1315 u64 dl_bw; /* dl_runtime / dl_deadline */
1318 * Actual scheduling parameters. Initialized with the values above,
1319 * they are continously updated during task execution. Note that
1320 * the remaining runtime could be < 0 in case we are in overrun.
1322 s64 runtime; /* remaining runtime for this instance */
1323 u64 deadline; /* absolute deadline for this instance */
1324 unsigned int flags; /* specifying the scheduler behaviour */
1329 * @dl_throttled tells if we exhausted the runtime. If so, the
1330 * task has to wait for a replenishment to be performed at the
1331 * next firing of dl_timer.
1333 * @dl_new tells if a new instance arrived. If so we must
1334 * start executing it with full runtime and reset its absolute
1337 * @dl_boosted tells if we are boosted due to DI. If so we are
1338 * outside bandwidth enforcement mechanism (but only until we
1339 * exit the critical section);
1341 * @dl_yielded tells if task gave up the cpu before consuming
1342 * all its available runtime during the last job.
1344 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1347 * Bandwidth enforcement timer. Each -deadline task has its
1348 * own bandwidth to be enforced, thus we need one timer per task.
1350 struct hrtimer dl_timer;
1358 u8 pad; /* Otherwise the compiler can store garbage here. */
1360 u32 s; /* Set of bits. */
1364 enum perf_event_task_context {
1365 perf_invalid_context = -1,
1366 perf_hw_context = 0,
1368 perf_nr_task_contexts,
1371 /* Track pages that require TLB flushes */
1372 struct tlbflush_unmap_batch {
1374 * Each bit set is a CPU that potentially has a TLB entry for one of
1375 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1377 struct cpumask cpumask;
1379 /* True if any bit in cpumask is set */
1380 bool flush_required;
1383 * If true then the PTE was dirty when unmapped. The entry must be
1384 * flushed before IO is initiated or a stale TLB entry potentially
1385 * allows an update without redirtying the page.
1390 struct task_struct {
1391 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1392 volatile long saved_state; /* saved state for "spinlock sleepers" */
1395 unsigned int flags; /* per process flags, defined below */
1396 unsigned int ptrace;
1399 struct llist_node wake_entry;
1401 unsigned int wakee_flips;
1402 unsigned long wakee_flip_decay_ts;
1403 struct task_struct *last_wakee;
1409 int prio, static_prio, normal_prio;
1410 unsigned int rt_priority;
1411 const struct sched_class *sched_class;
1412 struct sched_entity se;
1413 struct sched_rt_entity rt;
1414 #ifdef CONFIG_CGROUP_SCHED
1415 struct task_group *sched_task_group;
1417 struct sched_dl_entity dl;
1419 #ifdef CONFIG_PREEMPT_NOTIFIERS
1420 /* list of struct preempt_notifier: */
1421 struct hlist_head preempt_notifiers;
1424 #ifdef CONFIG_BLK_DEV_IO_TRACE
1425 unsigned int btrace_seq;
1428 unsigned int policy;
1429 #ifdef CONFIG_PREEMPT_RT_FULL
1430 int migrate_disable;
1431 # ifdef CONFIG_SCHED_DEBUG
1432 int migrate_disable_atomic;
1435 int nr_cpus_allowed;
1436 cpumask_t cpus_allowed;
1438 #ifdef CONFIG_PREEMPT_RCU
1439 int rcu_read_lock_nesting;
1440 union rcu_special rcu_read_unlock_special;
1441 struct list_head rcu_node_entry;
1442 struct rcu_node *rcu_blocked_node;
1443 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1444 #ifdef CONFIG_TASKS_RCU
1445 unsigned long rcu_tasks_nvcsw;
1446 bool rcu_tasks_holdout;
1447 struct list_head rcu_tasks_holdout_list;
1448 int rcu_tasks_idle_cpu;
1449 #endif /* #ifdef CONFIG_TASKS_RCU */
1451 #ifdef CONFIG_SCHED_INFO
1452 struct sched_info sched_info;
1455 struct list_head tasks;
1457 struct plist_node pushable_tasks;
1458 struct rb_node pushable_dl_tasks;
1461 struct mm_struct *mm, *active_mm;
1462 /* per-thread vma caching */
1463 u32 vmacache_seqnum;
1464 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1465 #if defined(SPLIT_RSS_COUNTING)
1466 struct task_rss_stat rss_stat;
1470 int exit_code, exit_signal;
1471 int pdeath_signal; /* The signal sent when the parent dies */
1472 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1474 /* Used for emulating ABI behavior of previous Linux versions */
1475 unsigned int personality;
1477 /* scheduler bits, serialized by scheduler locks */
1478 unsigned sched_reset_on_fork:1;
1479 unsigned sched_contributes_to_load:1;
1480 unsigned sched_migrated:1;
1481 unsigned :0; /* force alignment to the next boundary */
1483 /* unserialized, strictly 'current' */
1484 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1485 unsigned in_iowait:1;
1487 unsigned memcg_may_oom:1;
1489 #ifdef CONFIG_MEMCG_KMEM
1490 unsigned memcg_kmem_skip_account:1;
1492 #ifdef CONFIG_COMPAT_BRK
1493 unsigned brk_randomized:1;
1496 unsigned long atomic_flags; /* Flags needing atomic access. */
1498 struct restart_block restart_block;
1503 #ifdef CONFIG_CC_STACKPROTECTOR
1504 /* Canary value for the -fstack-protector gcc feature */
1505 unsigned long stack_canary;
1508 * pointers to (original) parent process, youngest child, younger sibling,
1509 * older sibling, respectively. (p->father can be replaced with
1510 * p->real_parent->pid)
1512 struct task_struct __rcu *real_parent; /* real parent process */
1513 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1515 * children/sibling forms the list of my natural children
1517 struct list_head children; /* list of my children */
1518 struct list_head sibling; /* linkage in my parent's children list */
1519 struct task_struct *group_leader; /* threadgroup leader */
1522 * ptraced is the list of tasks this task is using ptrace on.
1523 * This includes both natural children and PTRACE_ATTACH targets.
1524 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1526 struct list_head ptraced;
1527 struct list_head ptrace_entry;
1529 /* PID/PID hash table linkage. */
1530 struct pid_link pids[PIDTYPE_MAX];
1531 struct list_head thread_group;
1532 struct list_head thread_node;
1534 struct completion *vfork_done; /* for vfork() */
1535 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1536 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1538 cputime_t utime, stime, utimescaled, stimescaled;
1540 struct prev_cputime prev_cputime;
1541 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1542 raw_spinlock_t vtime_lock;
1543 seqcount_t vtime_seq;
1544 unsigned long long vtime_snap;
1549 } vtime_snap_whence;
1551 unsigned long nvcsw, nivcsw; /* context switch counts */
1552 u64 start_time; /* monotonic time in nsec */
1553 u64 real_start_time; /* boot based time in nsec */
1554 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1555 unsigned long min_flt, maj_flt;
1557 struct task_cputime cputime_expires;
1558 struct list_head cpu_timers[3];
1559 #ifdef CONFIG_PREEMPT_RT_BASE
1560 struct task_struct *posix_timer_list;
1563 /* process credentials */
1564 const struct cred __rcu *real_cred; /* objective and real subjective task
1565 * credentials (COW) */
1566 const struct cred __rcu *cred; /* effective (overridable) subjective task
1567 * credentials (COW) */
1568 char comm[TASK_COMM_LEN]; /* executable name excluding path
1569 - access with [gs]et_task_comm (which lock
1570 it with task_lock())
1571 - initialized normally by setup_new_exec */
1572 /* file system info */
1573 struct nameidata *nameidata;
1574 #ifdef CONFIG_SYSVIPC
1576 struct sysv_sem sysvsem;
1577 struct sysv_shm sysvshm;
1579 #ifdef CONFIG_DETECT_HUNG_TASK
1580 /* hung task detection */
1581 unsigned long last_switch_count;
1583 /* filesystem information */
1584 struct fs_struct *fs;
1585 /* open file information */
1586 struct files_struct *files;
1588 struct nsproxy *nsproxy;
1589 /* signal handlers */
1590 struct signal_struct *signal;
1591 struct sighand_struct *sighand;
1592 struct sigqueue *sigqueue_cache;
1594 sigset_t blocked, real_blocked;
1595 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1596 struct sigpending pending;
1597 #ifdef CONFIG_PREEMPT_RT_FULL
1598 /* TODO: move me into ->restart_block ? */
1599 struct siginfo forced_info;
1602 unsigned long sas_ss_sp;
1605 struct callback_head *task_works;
1607 struct audit_context *audit_context;
1608 #ifdef CONFIG_AUDITSYSCALL
1610 unsigned int sessionid;
1612 struct seccomp seccomp;
1614 /* Thread group tracking */
1617 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1619 spinlock_t alloc_lock;
1621 /* Protection of the PI data structures: */
1622 raw_spinlock_t pi_lock;
1624 struct wake_q_node wake_q;
1626 #ifdef CONFIG_RT_MUTEXES
1627 /* PI waiters blocked on a rt_mutex held by this task */
1628 struct rb_root pi_waiters;
1629 struct rb_node *pi_waiters_leftmost;
1630 /* Deadlock detection and priority inheritance handling */
1631 struct rt_mutex_waiter *pi_blocked_on;
1634 #ifdef CONFIG_DEBUG_MUTEXES
1635 /* mutex deadlock detection */
1636 struct mutex_waiter *blocked_on;
1638 #ifdef CONFIG_TRACE_IRQFLAGS
1639 unsigned int irq_events;
1640 unsigned long hardirq_enable_ip;
1641 unsigned long hardirq_disable_ip;
1642 unsigned int hardirq_enable_event;
1643 unsigned int hardirq_disable_event;
1644 int hardirqs_enabled;
1645 int hardirq_context;
1646 unsigned long softirq_disable_ip;
1647 unsigned long softirq_enable_ip;
1648 unsigned int softirq_disable_event;
1649 unsigned int softirq_enable_event;
1650 int softirqs_enabled;
1651 int softirq_context;
1653 #ifdef CONFIG_LOCKDEP
1654 # define MAX_LOCK_DEPTH 48UL
1657 unsigned int lockdep_recursion;
1658 struct held_lock held_locks[MAX_LOCK_DEPTH];
1659 gfp_t lockdep_reclaim_gfp;
1662 /* journalling filesystem info */
1665 /* stacked block device info */
1666 struct bio_list *bio_list;
1669 /* stack plugging */
1670 struct blk_plug *plug;
1674 struct reclaim_state *reclaim_state;
1676 struct backing_dev_info *backing_dev_info;
1678 struct io_context *io_context;
1680 unsigned long ptrace_message;
1681 siginfo_t *last_siginfo; /* For ptrace use. */
1682 struct task_io_accounting ioac;
1683 #if defined(CONFIG_TASK_XACCT)
1684 u64 acct_rss_mem1; /* accumulated rss usage */
1685 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1686 cputime_t acct_timexpd; /* stime + utime since last update */
1688 #ifdef CONFIG_CPUSETS
1689 nodemask_t mems_allowed; /* Protected by alloc_lock */
1690 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1691 int cpuset_mem_spread_rotor;
1692 int cpuset_slab_spread_rotor;
1694 #ifdef CONFIG_CGROUPS
1695 /* Control Group info protected by css_set_lock */
1696 struct css_set __rcu *cgroups;
1697 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1698 struct list_head cg_list;
1701 struct robust_list_head __user *robust_list;
1702 #ifdef CONFIG_COMPAT
1703 struct compat_robust_list_head __user *compat_robust_list;
1705 struct list_head pi_state_list;
1706 struct futex_pi_state *pi_state_cache;
1708 #ifdef CONFIG_PERF_EVENTS
1709 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1710 struct mutex perf_event_mutex;
1711 struct list_head perf_event_list;
1713 #ifdef CONFIG_DEBUG_PREEMPT
1714 unsigned long preempt_disable_ip;
1717 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1719 short pref_node_fork;
1721 #ifdef CONFIG_NUMA_BALANCING
1723 unsigned int numa_scan_period;
1724 unsigned int numa_scan_period_max;
1725 int numa_preferred_nid;
1726 unsigned long numa_migrate_retry;
1727 u64 node_stamp; /* migration stamp */
1728 u64 last_task_numa_placement;
1729 u64 last_sum_exec_runtime;
1730 struct callback_head numa_work;
1732 struct list_head numa_entry;
1733 struct numa_group *numa_group;
1736 * numa_faults is an array split into four regions:
1737 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1738 * in this precise order.
1740 * faults_memory: Exponential decaying average of faults on a per-node
1741 * basis. Scheduling placement decisions are made based on these
1742 * counts. The values remain static for the duration of a PTE scan.
1743 * faults_cpu: Track the nodes the process was running on when a NUMA
1744 * hinting fault was incurred.
1745 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1746 * during the current scan window. When the scan completes, the counts
1747 * in faults_memory and faults_cpu decay and these values are copied.
1749 unsigned long *numa_faults;
1750 unsigned long total_numa_faults;
1753 * numa_faults_locality tracks if faults recorded during the last
1754 * scan window were remote/local or failed to migrate. The task scan
1755 * period is adapted based on the locality of the faults with different
1756 * weights depending on whether they were shared or private faults
1758 unsigned long numa_faults_locality[3];
1760 unsigned long numa_pages_migrated;
1761 #endif /* CONFIG_NUMA_BALANCING */
1763 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1764 struct tlbflush_unmap_batch tlb_ubc;
1767 struct rcu_head rcu;
1770 * cache last used pipe for splice
1772 struct pipe_inode_info *splice_pipe;
1774 struct page_frag task_frag;
1776 #ifdef CONFIG_TASK_DELAY_ACCT
1777 struct task_delay_info *delays;
1779 #ifdef CONFIG_FAULT_INJECTION
1783 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1784 * balance_dirty_pages() for some dirty throttling pause
1787 int nr_dirtied_pause;
1788 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1790 #ifdef CONFIG_LATENCYTOP
1791 int latency_record_count;
1792 struct latency_record latency_record[LT_SAVECOUNT];
1795 * time slack values; these are used to round up poll() and
1796 * select() etc timeout values. These are in nanoseconds.
1798 unsigned long timer_slack_ns;
1799 unsigned long default_timer_slack_ns;
1802 unsigned int kasan_depth;
1804 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1805 /* Index of current stored address in ret_stack */
1807 /* Stack of return addresses for return function tracing */
1808 struct ftrace_ret_stack *ret_stack;
1809 /* time stamp for last schedule */
1810 unsigned long long ftrace_timestamp;
1812 * Number of functions that haven't been traced
1813 * because of depth overrun.
1815 atomic_t trace_overrun;
1816 /* Pause for the tracing */
1817 atomic_t tracing_graph_pause;
1819 #ifdef CONFIG_TRACING
1820 /* state flags for use by tracers */
1821 unsigned long trace;
1822 /* bitmask and counter of trace recursion */
1823 unsigned long trace_recursion;
1824 #ifdef CONFIG_WAKEUP_LATENCY_HIST
1825 u64 preempt_timestamp_hist;
1826 #ifdef CONFIG_MISSED_TIMER_OFFSETS_HIST
1830 #endif /* CONFIG_TRACING */
1832 struct mem_cgroup *memcg_in_oom;
1833 gfp_t memcg_oom_gfp_mask;
1834 int memcg_oom_order;
1836 /* number of pages to reclaim on returning to userland */
1837 unsigned int memcg_nr_pages_over_high;
1839 #ifdef CONFIG_UPROBES
1840 struct uprobe_task *utask;
1842 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1843 unsigned int sequential_io;
1844 unsigned int sequential_io_avg;
1846 #ifdef CONFIG_PREEMPT_RT_BASE
1847 struct rcu_head put_rcu;
1848 int softirq_nestcnt;
1849 unsigned int softirqs_raised;
1851 #ifdef CONFIG_PREEMPT_RT_FULL
1852 # if defined CONFIG_HIGHMEM || defined CONFIG_X86_32
1854 pte_t kmap_pte[KM_TYPE_NR];
1857 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1858 unsigned long task_state_change;
1860 #ifdef CONFIG_PREEMPT_RT_FULL
1863 int pagefault_disabled;
1864 /* CPU-specific state of this task */
1865 struct thread_struct thread;
1867 * WARNING: on x86, 'thread_struct' contains a variable-sized
1868 * structure. It *MUST* be at the end of 'task_struct'.
1870 * Do not put anything below here!
1874 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1875 extern int arch_task_struct_size __read_mostly;
1877 # define arch_task_struct_size (sizeof(struct task_struct))
1880 #define TNF_MIGRATED 0x01
1881 #define TNF_NO_GROUP 0x02
1882 #define TNF_SHARED 0x04
1883 #define TNF_FAULT_LOCAL 0x08
1884 #define TNF_MIGRATE_FAIL 0x10
1886 #ifdef CONFIG_NUMA_BALANCING
1887 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1888 extern pid_t task_numa_group_id(struct task_struct *p);
1889 extern void set_numabalancing_state(bool enabled);
1890 extern void task_numa_free(struct task_struct *p);
1891 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1892 int src_nid, int dst_cpu);
1894 static inline void task_numa_fault(int last_node, int node, int pages,
1898 static inline pid_t task_numa_group_id(struct task_struct *p)
1902 static inline void set_numabalancing_state(bool enabled)
1905 static inline void task_numa_free(struct task_struct *p)
1908 static inline bool should_numa_migrate_memory(struct task_struct *p,
1909 struct page *page, int src_nid, int dst_cpu)
1915 static inline struct pid *task_pid(struct task_struct *task)
1917 return task->pids[PIDTYPE_PID].pid;
1920 static inline struct pid *task_tgid(struct task_struct *task)
1922 return task->group_leader->pids[PIDTYPE_PID].pid;
1926 * Without tasklist or rcu lock it is not safe to dereference
1927 * the result of task_pgrp/task_session even if task == current,
1928 * we can race with another thread doing sys_setsid/sys_setpgid.
1930 static inline struct pid *task_pgrp(struct task_struct *task)
1932 return task->group_leader->pids[PIDTYPE_PGID].pid;
1935 static inline struct pid *task_session(struct task_struct *task)
1937 return task->group_leader->pids[PIDTYPE_SID].pid;
1940 struct pid_namespace;
1943 * the helpers to get the task's different pids as they are seen
1944 * from various namespaces
1946 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1947 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1949 * task_xid_nr_ns() : id seen from the ns specified;
1951 * set_task_vxid() : assigns a virtual id to a task;
1953 * see also pid_nr() etc in include/linux/pid.h
1955 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1956 struct pid_namespace *ns);
1958 static inline pid_t task_pid_nr(struct task_struct *tsk)
1963 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1964 struct pid_namespace *ns)
1966 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1969 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1971 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1975 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1980 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1982 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1984 return pid_vnr(task_tgid(tsk));
1988 static inline int pid_alive(const struct task_struct *p);
1989 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1995 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
2001 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
2003 return task_ppid_nr_ns(tsk, &init_pid_ns);
2006 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2007 struct pid_namespace *ns)
2009 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2012 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2014 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2018 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2019 struct pid_namespace *ns)
2021 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2024 static inline pid_t task_session_vnr(struct task_struct *tsk)
2026 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2029 /* obsolete, do not use */
2030 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2032 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2036 * pid_alive - check that a task structure is not stale
2037 * @p: Task structure to be checked.
2039 * Test if a process is not yet dead (at most zombie state)
2040 * If pid_alive fails, then pointers within the task structure
2041 * can be stale and must not be dereferenced.
2043 * Return: 1 if the process is alive. 0 otherwise.
2045 static inline int pid_alive(const struct task_struct *p)
2047 return p->pids[PIDTYPE_PID].pid != NULL;
2051 * is_global_init - check if a task structure is init. Since init
2052 * is free to have sub-threads we need to check tgid.
2053 * @tsk: Task structure to be checked.
2055 * Check if a task structure is the first user space task the kernel created.
2057 * Return: 1 if the task structure is init. 0 otherwise.
2059 static inline int is_global_init(struct task_struct *tsk)
2061 return task_tgid_nr(tsk) == 1;
2064 extern struct pid *cad_pid;
2066 extern void free_task(struct task_struct *tsk);
2067 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2069 #ifdef CONFIG_PREEMPT_RT_BASE
2070 extern void __put_task_struct_cb(struct rcu_head *rhp);
2072 static inline void put_task_struct(struct task_struct *t)
2074 if (atomic_dec_and_test(&t->usage))
2075 call_rcu(&t->put_rcu, __put_task_struct_cb);
2078 extern void __put_task_struct(struct task_struct *t);
2080 static inline void put_task_struct(struct task_struct *t)
2082 if (atomic_dec_and_test(&t->usage))
2083 __put_task_struct(t);
2087 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2088 extern void task_cputime(struct task_struct *t,
2089 cputime_t *utime, cputime_t *stime);
2090 extern void task_cputime_scaled(struct task_struct *t,
2091 cputime_t *utimescaled, cputime_t *stimescaled);
2092 extern cputime_t task_gtime(struct task_struct *t);
2094 static inline void task_cputime(struct task_struct *t,
2095 cputime_t *utime, cputime_t *stime)
2103 static inline void task_cputime_scaled(struct task_struct *t,
2104 cputime_t *utimescaled,
2105 cputime_t *stimescaled)
2108 *utimescaled = t->utimescaled;
2110 *stimescaled = t->stimescaled;
2113 static inline cputime_t task_gtime(struct task_struct *t)
2118 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2119 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2124 #define PF_IN_SOFTIRQ 0x00000001 /* Task is serving softirq */
2125 #define PF_EXITING 0x00000004 /* getting shut down */
2126 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2127 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2128 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2129 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2130 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2131 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2132 #define PF_DUMPCORE 0x00000200 /* dumped core */
2133 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2134 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2135 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2136 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2137 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2138 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2139 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2140 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2141 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2142 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2143 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2144 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2145 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2146 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2147 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2148 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2149 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2150 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2151 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2154 * Only the _current_ task can read/write to tsk->flags, but other
2155 * tasks can access tsk->flags in readonly mode for example
2156 * with tsk_used_math (like during threaded core dumping).
2157 * There is however an exception to this rule during ptrace
2158 * or during fork: the ptracer task is allowed to write to the
2159 * child->flags of its traced child (same goes for fork, the parent
2160 * can write to the child->flags), because we're guaranteed the
2161 * child is not running and in turn not changing child->flags
2162 * at the same time the parent does it.
2164 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2165 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2166 #define clear_used_math() clear_stopped_child_used_math(current)
2167 #define set_used_math() set_stopped_child_used_math(current)
2168 #define conditional_stopped_child_used_math(condition, child) \
2169 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2170 #define conditional_used_math(condition) \
2171 conditional_stopped_child_used_math(condition, current)
2172 #define copy_to_stopped_child_used_math(child) \
2173 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2174 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2175 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2176 #define used_math() tsk_used_math(current)
2178 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2179 * __GFP_FS is also cleared as it implies __GFP_IO.
2181 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2183 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2184 flags &= ~(__GFP_IO | __GFP_FS);
2188 static inline unsigned int memalloc_noio_save(void)
2190 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2191 current->flags |= PF_MEMALLOC_NOIO;
2195 static inline void memalloc_noio_restore(unsigned int flags)
2197 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2200 /* Per-process atomic flags. */
2201 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2202 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2203 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2206 #define TASK_PFA_TEST(name, func) \
2207 static inline bool task_##func(struct task_struct *p) \
2208 { return test_bit(PFA_##name, &p->atomic_flags); }
2209 #define TASK_PFA_SET(name, func) \
2210 static inline void task_set_##func(struct task_struct *p) \
2211 { set_bit(PFA_##name, &p->atomic_flags); }
2212 #define TASK_PFA_CLEAR(name, func) \
2213 static inline void task_clear_##func(struct task_struct *p) \
2214 { clear_bit(PFA_##name, &p->atomic_flags); }
2216 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2217 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2219 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2220 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2221 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2223 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2224 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2225 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2228 * task->jobctl flags
2230 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2232 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2233 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2234 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2235 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2236 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2237 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2238 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2240 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2241 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2242 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2243 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2244 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2245 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2246 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2248 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2249 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2251 extern bool task_set_jobctl_pending(struct task_struct *task,
2252 unsigned long mask);
2253 extern void task_clear_jobctl_trapping(struct task_struct *task);
2254 extern void task_clear_jobctl_pending(struct task_struct *task,
2255 unsigned long mask);
2257 static inline void rcu_copy_process(struct task_struct *p)
2259 #ifdef CONFIG_PREEMPT_RCU
2260 p->rcu_read_lock_nesting = 0;
2261 p->rcu_read_unlock_special.s = 0;
2262 p->rcu_blocked_node = NULL;
2263 INIT_LIST_HEAD(&p->rcu_node_entry);
2264 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2265 #ifdef CONFIG_TASKS_RCU
2266 p->rcu_tasks_holdout = false;
2267 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2268 p->rcu_tasks_idle_cpu = -1;
2269 #endif /* #ifdef CONFIG_TASKS_RCU */
2272 static inline void tsk_restore_flags(struct task_struct *task,
2273 unsigned long orig_flags, unsigned long flags)
2275 task->flags &= ~flags;
2276 task->flags |= orig_flags & flags;
2279 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2280 const struct cpumask *trial);
2281 extern int task_can_attach(struct task_struct *p,
2282 const struct cpumask *cs_cpus_allowed);
2284 extern void do_set_cpus_allowed(struct task_struct *p,
2285 const struct cpumask *new_mask);
2287 extern int set_cpus_allowed_ptr(struct task_struct *p,
2288 const struct cpumask *new_mask);
2289 int migrate_me(void);
2290 void tell_sched_cpu_down_begin(int cpu);
2291 void tell_sched_cpu_down_done(int cpu);
2294 static inline void do_set_cpus_allowed(struct task_struct *p,
2295 const struct cpumask *new_mask)
2298 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2299 const struct cpumask *new_mask)
2301 if (!cpumask_test_cpu(0, new_mask))
2305 static inline int migrate_me(void) { return 0; }
2306 static inline void tell_sched_cpu_down_begin(int cpu) { }
2307 static inline void tell_sched_cpu_down_done(int cpu) { }
2310 #ifdef CONFIG_NO_HZ_COMMON
2311 void calc_load_enter_idle(void);
2312 void calc_load_exit_idle(void);
2314 static inline void calc_load_enter_idle(void) { }
2315 static inline void calc_load_exit_idle(void) { }
2316 #endif /* CONFIG_NO_HZ_COMMON */
2319 * Do not use outside of architecture code which knows its limitations.
2321 * sched_clock() has no promise of monotonicity or bounded drift between
2322 * CPUs, use (which you should not) requires disabling IRQs.
2324 * Please use one of the three interfaces below.
2326 extern unsigned long long notrace sched_clock(void);
2328 * See the comment in kernel/sched/clock.c
2330 extern u64 cpu_clock(int cpu);
2331 extern u64 local_clock(void);
2332 extern u64 running_clock(void);
2333 extern u64 sched_clock_cpu(int cpu);
2336 extern void sched_clock_init(void);
2338 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2339 static inline void sched_clock_tick(void)
2343 static inline void sched_clock_idle_sleep_event(void)
2347 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2352 * Architectures can set this to 1 if they have specified
2353 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2354 * but then during bootup it turns out that sched_clock()
2355 * is reliable after all:
2357 extern int sched_clock_stable(void);
2358 extern void set_sched_clock_stable(void);
2359 extern void clear_sched_clock_stable(void);
2361 extern void sched_clock_tick(void);
2362 extern void sched_clock_idle_sleep_event(void);
2363 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2366 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2368 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2369 * The reason for this explicit opt-in is not to have perf penalty with
2370 * slow sched_clocks.
2372 extern void enable_sched_clock_irqtime(void);
2373 extern void disable_sched_clock_irqtime(void);
2375 static inline void enable_sched_clock_irqtime(void) {}
2376 static inline void disable_sched_clock_irqtime(void) {}
2379 extern unsigned long long
2380 task_sched_runtime(struct task_struct *task);
2382 /* sched_exec is called by processes performing an exec */
2384 extern void sched_exec(void);
2386 #define sched_exec() {}
2389 extern void sched_clock_idle_sleep_event(void);
2390 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2392 #ifdef CONFIG_HOTPLUG_CPU
2393 extern void idle_task_exit(void);
2395 static inline void idle_task_exit(void) {}
2398 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2399 extern void wake_up_nohz_cpu(int cpu);
2401 static inline void wake_up_nohz_cpu(int cpu) { }
2404 #ifdef CONFIG_NO_HZ_FULL
2405 extern bool sched_can_stop_tick(void);
2406 extern u64 scheduler_tick_max_deferment(void);
2408 static inline bool sched_can_stop_tick(void) { return false; }
2411 #ifdef CONFIG_SCHED_AUTOGROUP
2412 extern void sched_autogroup_create_attach(struct task_struct *p);
2413 extern void sched_autogroup_detach(struct task_struct *p);
2414 extern void sched_autogroup_fork(struct signal_struct *sig);
2415 extern void sched_autogroup_exit(struct signal_struct *sig);
2416 #ifdef CONFIG_PROC_FS
2417 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2418 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2421 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2422 static inline void sched_autogroup_detach(struct task_struct *p) { }
2423 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2424 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2427 extern int yield_to(struct task_struct *p, bool preempt);
2428 extern void set_user_nice(struct task_struct *p, long nice);
2429 extern int task_prio(const struct task_struct *p);
2431 * task_nice - return the nice value of a given task.
2432 * @p: the task in question.
2434 * Return: The nice value [ -20 ... 0 ... 19 ].
2436 static inline int task_nice(const struct task_struct *p)
2438 return PRIO_TO_NICE((p)->static_prio);
2440 extern int can_nice(const struct task_struct *p, const int nice);
2441 extern int task_curr(const struct task_struct *p);
2442 extern int idle_cpu(int cpu);
2443 extern int sched_setscheduler(struct task_struct *, int,
2444 const struct sched_param *);
2445 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2446 const struct sched_param *);
2447 extern int sched_setattr(struct task_struct *,
2448 const struct sched_attr *);
2449 extern struct task_struct *idle_task(int cpu);
2451 * is_idle_task - is the specified task an idle task?
2452 * @p: the task in question.
2454 * Return: 1 if @p is an idle task. 0 otherwise.
2456 static inline bool is_idle_task(const struct task_struct *p)
2460 extern struct task_struct *curr_task(int cpu);
2461 extern void set_curr_task(int cpu, struct task_struct *p);
2465 union thread_union {
2466 struct thread_info thread_info;
2467 unsigned long stack[THREAD_SIZE/sizeof(long)];
2470 #ifndef __HAVE_ARCH_KSTACK_END
2471 static inline int kstack_end(void *addr)
2473 /* Reliable end of stack detection:
2474 * Some APM bios versions misalign the stack
2476 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2480 extern union thread_union init_thread_union;
2481 extern struct task_struct init_task;
2483 extern struct mm_struct init_mm;
2485 extern struct pid_namespace init_pid_ns;
2488 * find a task by one of its numerical ids
2490 * find_task_by_pid_ns():
2491 * finds a task by its pid in the specified namespace
2492 * find_task_by_vpid():
2493 * finds a task by its virtual pid
2495 * see also find_vpid() etc in include/linux/pid.h
2498 extern struct task_struct *find_task_by_vpid(pid_t nr);
2499 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2500 struct pid_namespace *ns);
2502 /* per-UID process charging. */
2503 extern struct user_struct * alloc_uid(kuid_t);
2504 static inline struct user_struct *get_uid(struct user_struct *u)
2506 atomic_inc(&u->__count);
2509 extern void free_uid(struct user_struct *);
2511 #include <asm/current.h>
2513 extern void xtime_update(unsigned long ticks);
2515 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2516 extern int wake_up_process(struct task_struct *tsk);
2517 extern int wake_up_lock_sleeper(struct task_struct * tsk);
2518 extern void wake_up_new_task(struct task_struct *tsk);
2520 extern void kick_process(struct task_struct *tsk);
2522 static inline void kick_process(struct task_struct *tsk) { }
2524 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2525 extern void sched_dead(struct task_struct *p);
2527 extern void proc_caches_init(void);
2528 extern void flush_signals(struct task_struct *);
2529 extern void ignore_signals(struct task_struct *);
2530 extern void flush_signal_handlers(struct task_struct *, int force_default);
2531 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2533 static inline int kernel_dequeue_signal(siginfo_t *info)
2535 struct task_struct *tsk = current;
2539 spin_lock_irq(&tsk->sighand->siglock);
2540 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2541 spin_unlock_irq(&tsk->sighand->siglock);
2546 static inline void kernel_signal_stop(void)
2548 spin_lock_irq(¤t->sighand->siglock);
2549 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2550 __set_current_state(TASK_STOPPED);
2551 spin_unlock_irq(¤t->sighand->siglock);
2556 extern void release_task(struct task_struct * p);
2557 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2558 extern int force_sigsegv(int, struct task_struct *);
2559 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2560 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2561 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2562 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2563 const struct cred *, u32);
2564 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2565 extern int kill_pid(struct pid *pid, int sig, int priv);
2566 extern int kill_proc_info(int, struct siginfo *, pid_t);
2567 extern __must_check bool do_notify_parent(struct task_struct *, int);
2568 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2569 extern void force_sig(int, struct task_struct *);
2570 extern int send_sig(int, struct task_struct *, int);
2571 extern int zap_other_threads(struct task_struct *p);
2572 extern struct sigqueue *sigqueue_alloc(void);
2573 extern void sigqueue_free(struct sigqueue *);
2574 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2575 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2577 static inline void restore_saved_sigmask(void)
2579 if (test_and_clear_restore_sigmask())
2580 __set_current_blocked(¤t->saved_sigmask);
2583 static inline sigset_t *sigmask_to_save(void)
2585 sigset_t *res = ¤t->blocked;
2586 if (unlikely(test_restore_sigmask()))
2587 res = ¤t->saved_sigmask;
2591 static inline int kill_cad_pid(int sig, int priv)
2593 return kill_pid(cad_pid, sig, priv);
2596 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2597 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2598 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2599 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2602 * True if we are on the alternate signal stack.
2604 static inline int on_sig_stack(unsigned long sp)
2606 #ifdef CONFIG_STACK_GROWSUP
2607 return sp >= current->sas_ss_sp &&
2608 sp - current->sas_ss_sp < current->sas_ss_size;
2610 return sp > current->sas_ss_sp &&
2611 sp - current->sas_ss_sp <= current->sas_ss_size;
2615 static inline int sas_ss_flags(unsigned long sp)
2617 if (!current->sas_ss_size)
2620 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2623 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2625 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2626 #ifdef CONFIG_STACK_GROWSUP
2627 return current->sas_ss_sp;
2629 return current->sas_ss_sp + current->sas_ss_size;
2635 * Routines for handling mm_structs
2637 extern struct mm_struct * mm_alloc(void);
2639 /* mmdrop drops the mm and the page tables */
2640 extern void __mmdrop(struct mm_struct *);
2642 static inline void mmdrop(struct mm_struct * mm)
2644 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2648 #ifdef CONFIG_PREEMPT_RT_BASE
2649 extern void __mmdrop_delayed(struct rcu_head *rhp);
2650 static inline void mmdrop_delayed(struct mm_struct *mm)
2652 if (atomic_dec_and_test(&mm->mm_count))
2653 call_rcu(&mm->delayed_drop, __mmdrop_delayed);
2656 # define mmdrop_delayed(mm) mmdrop(mm)
2659 /* mmput gets rid of the mappings and all user-space */
2660 extern void mmput(struct mm_struct *);
2661 /* Grab a reference to a task's mm, if it is not already going away */
2662 extern struct mm_struct *get_task_mm(struct task_struct *task);
2664 * Grab a reference to a task's mm, if it is not already going away
2665 * and ptrace_may_access with the mode parameter passed to it
2668 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2669 /* Remove the current tasks stale references to the old mm_struct */
2670 extern void mm_release(struct task_struct *, struct mm_struct *);
2672 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2673 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2674 struct task_struct *, unsigned long);
2676 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2677 struct task_struct *);
2679 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2680 * via pt_regs, so ignore the tls argument passed via C. */
2681 static inline int copy_thread_tls(
2682 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2683 struct task_struct *p, unsigned long tls)
2685 return copy_thread(clone_flags, sp, arg, p);
2688 extern void flush_thread(void);
2689 extern void exit_thread(void);
2691 extern void exit_files(struct task_struct *);
2692 extern void __cleanup_sighand(struct sighand_struct *);
2694 extern void exit_itimers(struct signal_struct *);
2695 extern void flush_itimer_signals(void);
2697 extern void do_group_exit(int);
2699 extern int do_execve(struct filename *,
2700 const char __user * const __user *,
2701 const char __user * const __user *);
2702 extern int do_execveat(int, struct filename *,
2703 const char __user * const __user *,
2704 const char __user * const __user *,
2706 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2707 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2708 struct task_struct *fork_idle(int);
2709 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2711 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2712 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2714 __set_task_comm(tsk, from, false);
2716 extern char *get_task_comm(char *to, struct task_struct *tsk);
2719 void scheduler_ipi(void);
2720 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2722 static inline void scheduler_ipi(void) { }
2723 static inline unsigned long wait_task_inactive(struct task_struct *p,
2730 #define tasklist_empty() \
2731 list_empty(&init_task.tasks)
2733 #define next_task(p) \
2734 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2736 #define for_each_process(p) \
2737 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2739 extern bool current_is_single_threaded(void);
2742 * Careful: do_each_thread/while_each_thread is a double loop so
2743 * 'break' will not work as expected - use goto instead.
2745 #define do_each_thread(g, t) \
2746 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2748 #define while_each_thread(g, t) \
2749 while ((t = next_thread(t)) != g)
2751 #define __for_each_thread(signal, t) \
2752 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2754 #define for_each_thread(p, t) \
2755 __for_each_thread((p)->signal, t)
2757 /* Careful: this is a double loop, 'break' won't work as expected. */
2758 #define for_each_process_thread(p, t) \
2759 for_each_process(p) for_each_thread(p, t)
2761 static inline int get_nr_threads(struct task_struct *tsk)
2763 return tsk->signal->nr_threads;
2766 static inline bool thread_group_leader(struct task_struct *p)
2768 return p->exit_signal >= 0;
2771 /* Do to the insanities of de_thread it is possible for a process
2772 * to have the pid of the thread group leader without actually being
2773 * the thread group leader. For iteration through the pids in proc
2774 * all we care about is that we have a task with the appropriate
2775 * pid, we don't actually care if we have the right task.
2777 static inline bool has_group_leader_pid(struct task_struct *p)
2779 return task_pid(p) == p->signal->leader_pid;
2783 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2785 return p1->signal == p2->signal;
2788 static inline struct task_struct *next_thread(const struct task_struct *p)
2790 return list_entry_rcu(p->thread_group.next,
2791 struct task_struct, thread_group);
2794 static inline int thread_group_empty(struct task_struct *p)
2796 return list_empty(&p->thread_group);
2799 #define delay_group_leader(p) \
2800 (thread_group_leader(p) && !thread_group_empty(p))
2803 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2804 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2805 * pins the final release of task.io_context. Also protects ->cpuset and
2806 * ->cgroup.subsys[]. And ->vfork_done.
2808 * Nests both inside and outside of read_lock(&tasklist_lock).
2809 * It must not be nested with write_lock_irq(&tasklist_lock),
2810 * neither inside nor outside.
2812 static inline void task_lock(struct task_struct *p)
2814 spin_lock(&p->alloc_lock);
2817 static inline void task_unlock(struct task_struct *p)
2819 spin_unlock(&p->alloc_lock);
2822 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2823 unsigned long *flags);
2825 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2826 unsigned long *flags)
2828 struct sighand_struct *ret;
2830 ret = __lock_task_sighand(tsk, flags);
2831 (void)__cond_lock(&tsk->sighand->siglock, ret);
2835 static inline void unlock_task_sighand(struct task_struct *tsk,
2836 unsigned long *flags)
2838 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2842 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2843 * @tsk: task causing the changes
2845 * All operations which modify a threadgroup - a new thread joining the
2846 * group, death of a member thread (the assertion of PF_EXITING) and
2847 * exec(2) dethreading the process and replacing the leader - are wrapped
2848 * by threadgroup_change_{begin|end}(). This is to provide a place which
2849 * subsystems needing threadgroup stability can hook into for
2852 static inline void threadgroup_change_begin(struct task_struct *tsk)
2855 cgroup_threadgroup_change_begin(tsk);
2859 * threadgroup_change_end - mark the end of changes to a threadgroup
2860 * @tsk: task causing the changes
2862 * See threadgroup_change_begin().
2864 static inline void threadgroup_change_end(struct task_struct *tsk)
2866 cgroup_threadgroup_change_end(tsk);
2869 #ifndef __HAVE_THREAD_FUNCTIONS
2871 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2872 #define task_stack_page(task) ((task)->stack)
2874 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2876 *task_thread_info(p) = *task_thread_info(org);
2877 task_thread_info(p)->task = p;
2881 * Return the address of the last usable long on the stack.
2883 * When the stack grows down, this is just above the thread
2884 * info struct. Going any lower will corrupt the threadinfo.
2886 * When the stack grows up, this is the highest address.
2887 * Beyond that position, we corrupt data on the next page.
2889 static inline unsigned long *end_of_stack(struct task_struct *p)
2891 #ifdef CONFIG_STACK_GROWSUP
2892 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2894 return (unsigned long *)(task_thread_info(p) + 1);
2899 #define task_stack_end_corrupted(task) \
2900 (*(end_of_stack(task)) != STACK_END_MAGIC)
2902 static inline int object_is_on_stack(void *obj)
2904 void *stack = task_stack_page(current);
2906 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2909 extern void thread_info_cache_init(void);
2911 #ifdef CONFIG_DEBUG_STACK_USAGE
2912 static inline unsigned long stack_not_used(struct task_struct *p)
2914 unsigned long *n = end_of_stack(p);
2916 do { /* Skip over canary */
2920 return (unsigned long)n - (unsigned long)end_of_stack(p);
2923 extern void set_task_stack_end_magic(struct task_struct *tsk);
2925 /* set thread flags in other task's structures
2926 * - see asm/thread_info.h for TIF_xxxx flags available
2928 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2930 set_ti_thread_flag(task_thread_info(tsk), flag);
2933 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2935 clear_ti_thread_flag(task_thread_info(tsk), flag);
2938 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2940 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2943 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2945 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2948 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2950 return test_ti_thread_flag(task_thread_info(tsk), flag);
2953 static inline void set_tsk_need_resched(struct task_struct *tsk)
2955 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2958 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2960 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2963 static inline int test_tsk_need_resched(struct task_struct *tsk)
2965 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2968 #ifdef CONFIG_PREEMPT_LAZY
2969 static inline void set_tsk_need_resched_lazy(struct task_struct *tsk)
2971 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2974 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk)
2976 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY);
2979 static inline int test_tsk_need_resched_lazy(struct task_struct *tsk)
2981 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED_LAZY));
2984 static inline int need_resched_lazy(void)
2986 return test_thread_flag(TIF_NEED_RESCHED_LAZY);
2989 static inline int need_resched_now(void)
2991 return test_thread_flag(TIF_NEED_RESCHED);
2995 static inline void clear_tsk_need_resched_lazy(struct task_struct *tsk) { }
2996 static inline int need_resched_lazy(void) { return 0; }
2998 static inline int need_resched_now(void)
3000 return test_thread_flag(TIF_NEED_RESCHED);
3005 static inline int restart_syscall(void)
3007 set_tsk_thread_flag(current, TIF_SIGPENDING);
3008 return -ERESTARTNOINTR;
3011 static inline int signal_pending(struct task_struct *p)
3013 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
3016 static inline int __fatal_signal_pending(struct task_struct *p)
3018 return unlikely(sigismember(&p->pending.signal, SIGKILL));
3021 static inline int fatal_signal_pending(struct task_struct *p)
3023 return signal_pending(p) && __fatal_signal_pending(p);
3026 static inline int signal_pending_state(long state, struct task_struct *p)
3028 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
3030 if (!signal_pending(p))
3033 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
3036 static inline bool __task_is_stopped_or_traced(struct task_struct *task)
3038 if (task->state & (__TASK_STOPPED | __TASK_TRACED))
3040 #ifdef CONFIG_PREEMPT_RT_FULL
3041 if (task->saved_state & (__TASK_STOPPED | __TASK_TRACED))
3047 static inline bool task_is_stopped_or_traced(struct task_struct *task)
3049 bool traced_stopped;
3051 #ifdef CONFIG_PREEMPT_RT_FULL
3052 unsigned long flags;
3054 raw_spin_lock_irqsave(&task->pi_lock, flags);
3055 traced_stopped = __task_is_stopped_or_traced(task);
3056 raw_spin_unlock_irqrestore(&task->pi_lock, flags);
3058 traced_stopped = __task_is_stopped_or_traced(task);
3060 return traced_stopped;
3063 static inline bool task_is_traced(struct task_struct *task)
3065 bool traced = false;
3067 if (task->state & __TASK_TRACED)
3069 #ifdef CONFIG_PREEMPT_RT_FULL
3070 /* in case the task is sleeping on tasklist_lock */
3071 raw_spin_lock_irq(&task->pi_lock);
3072 if (task->state & __TASK_TRACED)
3074 else if (task->saved_state & __TASK_TRACED)
3076 raw_spin_unlock_irq(&task->pi_lock);
3082 * cond_resched() and cond_resched_lock(): latency reduction via
3083 * explicit rescheduling in places that are safe. The return
3084 * value indicates whether a reschedule was done in fact.
3085 * cond_resched_lock() will drop the spinlock before scheduling,
3086 * cond_resched_softirq() will enable bhs before scheduling.
3088 extern int _cond_resched(void);
3090 #define cond_resched() ({ \
3091 ___might_sleep(__FILE__, __LINE__, 0); \
3095 extern int __cond_resched_lock(spinlock_t *lock);
3097 #define cond_resched_lock(lock) ({ \
3098 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3099 __cond_resched_lock(lock); \
3102 #ifndef CONFIG_PREEMPT_RT_FULL
3103 extern int __cond_resched_softirq(void);
3105 #define cond_resched_softirq() ({ \
3106 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3107 __cond_resched_softirq(); \
3110 # define cond_resched_softirq() cond_resched()
3113 static inline void cond_resched_rcu(void)
3115 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3123 * Does a critical section need to be broken due to another
3124 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3125 * but a general need for low latency)
3127 static inline int spin_needbreak(spinlock_t *lock)
3129 #ifdef CONFIG_PREEMPT
3130 return spin_is_contended(lock);
3137 * Idle thread specific functions to determine the need_resched
3140 #ifdef TIF_POLLING_NRFLAG
3141 static inline int tsk_is_polling(struct task_struct *p)
3143 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3146 static inline void __current_set_polling(void)
3148 set_thread_flag(TIF_POLLING_NRFLAG);
3151 static inline bool __must_check current_set_polling_and_test(void)
3153 __current_set_polling();
3156 * Polling state must be visible before we test NEED_RESCHED,
3157 * paired by resched_curr()
3159 smp_mb__after_atomic();
3161 return unlikely(tif_need_resched());
3164 static inline void __current_clr_polling(void)
3166 clear_thread_flag(TIF_POLLING_NRFLAG);
3169 static inline bool __must_check current_clr_polling_and_test(void)
3171 __current_clr_polling();
3174 * Polling state must be visible before we test NEED_RESCHED,
3175 * paired by resched_curr()
3177 smp_mb__after_atomic();
3179 return unlikely(tif_need_resched());
3183 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3184 static inline void __current_set_polling(void) { }
3185 static inline void __current_clr_polling(void) { }
3187 static inline bool __must_check current_set_polling_and_test(void)
3189 return unlikely(tif_need_resched());
3191 static inline bool __must_check current_clr_polling_and_test(void)
3193 return unlikely(tif_need_resched());
3197 static inline void current_clr_polling(void)
3199 __current_clr_polling();
3202 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3203 * Once the bit is cleared, we'll get IPIs with every new
3204 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3207 smp_mb(); /* paired with resched_curr() */
3209 preempt_fold_need_resched();
3212 static __always_inline bool need_resched(void)
3214 return unlikely(tif_need_resched());
3218 * Thread group CPU time accounting.
3220 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3221 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3224 * Reevaluate whether the task has signals pending delivery.
3225 * Wake the task if so.
3226 * This is required every time the blocked sigset_t changes.
3227 * callers must hold sighand->siglock.
3229 extern void recalc_sigpending_and_wake(struct task_struct *t);
3230 extern void recalc_sigpending(void);
3232 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3234 static inline void signal_wake_up(struct task_struct *t, bool resume)
3236 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3238 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3240 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3244 * Wrappers for p->thread_info->cpu access. No-op on UP.
3248 static inline unsigned int task_cpu(const struct task_struct *p)
3250 return task_thread_info(p)->cpu;
3253 static inline int task_node(const struct task_struct *p)
3255 return cpu_to_node(task_cpu(p));
3258 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3262 static inline unsigned int task_cpu(const struct task_struct *p)
3267 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3271 #endif /* CONFIG_SMP */
3273 static inline int __migrate_disabled(struct task_struct *p)
3275 #ifdef CONFIG_PREEMPT_RT_FULL
3276 return p->migrate_disable;
3282 /* Future-safe accessor for struct task_struct's cpus_allowed. */
3283 static inline const struct cpumask *tsk_cpus_allowed(struct task_struct *p)
3285 if (__migrate_disabled(p))
3286 return cpumask_of(task_cpu(p));
3288 return &p->cpus_allowed;
3291 static inline int tsk_nr_cpus_allowed(struct task_struct *p)
3293 if (__migrate_disabled(p))
3295 return p->nr_cpus_allowed;
3298 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3299 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3301 #ifdef CONFIG_CGROUP_SCHED
3302 extern struct task_group root_task_group;
3303 #endif /* CONFIG_CGROUP_SCHED */
3305 extern int task_can_switch_user(struct user_struct *up,
3306 struct task_struct *tsk);
3308 #ifdef CONFIG_TASK_XACCT
3309 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3311 tsk->ioac.rchar += amt;
3314 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3316 tsk->ioac.wchar += amt;
3319 static inline void inc_syscr(struct task_struct *tsk)
3324 static inline void inc_syscw(struct task_struct *tsk)
3329 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3333 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3337 static inline void inc_syscr(struct task_struct *tsk)
3341 static inline void inc_syscw(struct task_struct *tsk)
3346 #ifndef TASK_SIZE_OF
3347 #define TASK_SIZE_OF(tsk) TASK_SIZE
3351 extern void mm_update_next_owner(struct mm_struct *mm);
3353 static inline void mm_update_next_owner(struct mm_struct *mm)
3356 #endif /* CONFIG_MEMCG */
3358 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3361 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3364 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3367 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3370 static inline unsigned long rlimit(unsigned int limit)
3372 return task_rlimit(current, limit);
3375 static inline unsigned long rlimit_max(unsigned int limit)
3377 return task_rlimit_max(current, limit);