4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/kcov.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);
177 #ifdef CONFIG_CPU_QUIET
178 extern u64 nr_running_integral(unsigned int cpu);
181 extern void calc_global_load(unsigned long ticks);
183 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
184 extern void update_cpu_load_nohz(void);
186 static inline void update_cpu_load_nohz(void) { }
189 extern unsigned long get_parent_ip(unsigned long addr);
191 extern void dump_cpu_task(int cpu);
196 #ifdef CONFIG_SCHED_DEBUG
197 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
198 extern void proc_sched_set_task(struct task_struct *p);
202 * Task state bitmask. NOTE! These bits are also
203 * encoded in fs/proc/array.c: get_task_state().
205 * We have two separate sets of flags: task->state
206 * is about runnability, while task->exit_state are
207 * about the task exiting. Confusing, but this way
208 * modifying one set can't modify the other one by
211 #define TASK_RUNNING 0
212 #define TASK_INTERRUPTIBLE 1
213 #define TASK_UNINTERRUPTIBLE 2
214 #define __TASK_STOPPED 4
215 #define __TASK_TRACED 8
216 /* in tsk->exit_state */
218 #define EXIT_ZOMBIE 32
219 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
220 /* in tsk->state again */
222 #define TASK_WAKEKILL 128
223 #define TASK_WAKING 256
224 #define TASK_PARKED 512
225 #define TASK_NOLOAD 1024
226 #define TASK_NEW 2048
227 #define TASK_STATE_MAX 4096
229 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
231 extern char ___assert_task_state[1 - 2*!!(
232 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
234 /* Convenience macros for the sake of set_task_state */
235 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
236 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
237 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
239 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
241 /* Convenience macros for the sake of wake_up */
242 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
243 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
245 /* get_task_state() */
246 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
247 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
248 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
250 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
251 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
252 #define task_is_stopped_or_traced(task) \
253 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
254 #define task_contributes_to_load(task) \
255 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
256 (task->flags & PF_FROZEN) == 0 && \
257 (task->state & TASK_NOLOAD) == 0)
259 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
261 #define __set_task_state(tsk, state_value) \
263 (tsk)->task_state_change = _THIS_IP_; \
264 (tsk)->state = (state_value); \
266 #define set_task_state(tsk, state_value) \
268 (tsk)->task_state_change = _THIS_IP_; \
269 smp_store_mb((tsk)->state, (state_value)); \
273 * set_current_state() includes a barrier so that the write of current->state
274 * is correctly serialised wrt the caller's subsequent test of whether to
277 * set_current_state(TASK_UNINTERRUPTIBLE);
278 * if (do_i_need_to_sleep())
281 * If the caller does not need such serialisation then use __set_current_state()
283 #define __set_current_state(state_value) \
285 current->task_state_change = _THIS_IP_; \
286 current->state = (state_value); \
288 #define set_current_state(state_value) \
290 current->task_state_change = _THIS_IP_; \
291 smp_store_mb(current->state, (state_value)); \
296 #define __set_task_state(tsk, state_value) \
297 do { (tsk)->state = (state_value); } while (0)
298 #define set_task_state(tsk, state_value) \
299 smp_store_mb((tsk)->state, (state_value))
302 * set_current_state() includes a barrier so that the write of current->state
303 * is correctly serialised wrt the caller's subsequent test of whether to
306 * set_current_state(TASK_UNINTERRUPTIBLE);
307 * if (do_i_need_to_sleep())
310 * If the caller does not need such serialisation then use __set_current_state()
312 #define __set_current_state(state_value) \
313 do { current->state = (state_value); } while (0)
314 #define set_current_state(state_value) \
315 smp_store_mb(current->state, (state_value))
319 /* Task command name length */
320 #define TASK_COMM_LEN 16
331 #include <linux/spinlock.h>
334 * This serializes "schedule()" and also protects
335 * the run-queue from deletions/modifications (but
336 * _adding_ to the beginning of the run-queue has
339 extern rwlock_t tasklist_lock;
340 extern spinlock_t mmlist_lock;
344 #ifdef CONFIG_PROVE_RCU
345 extern int lockdep_tasklist_lock_is_held(void);
346 #endif /* #ifdef CONFIG_PROVE_RCU */
348 extern void sched_init(void);
349 extern void sched_init_smp(void);
350 extern asmlinkage void schedule_tail(struct task_struct *prev);
351 extern void init_idle(struct task_struct *idle, int cpu);
352 extern void init_idle_bootup_task(struct task_struct *idle);
354 extern cpumask_var_t cpu_isolated_map;
356 extern int runqueue_is_locked(int cpu);
358 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
359 extern void nohz_balance_enter_idle(int cpu);
360 extern void set_cpu_sd_state_idle(void);
361 extern int get_nohz_timer_target(void);
363 static inline void nohz_balance_enter_idle(int cpu) { }
364 static inline void set_cpu_sd_state_idle(void) { }
368 * Only dump TASK_* tasks. (0 for all tasks)
370 extern void show_state_filter(unsigned long state_filter);
372 static inline void show_state(void)
374 show_state_filter(0);
377 extern void show_regs(struct pt_regs *);
380 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
381 * task), SP is the stack pointer of the first frame that should be shown in the back
382 * trace (or NULL if the entire call-chain of the task should be shown).
384 extern void show_stack(struct task_struct *task, unsigned long *sp);
386 extern void cpu_init (void);
387 extern void trap_init(void);
388 extern void update_process_times(int user);
389 extern void scheduler_tick(void);
391 extern void sched_show_task(struct task_struct *p);
393 #ifdef CONFIG_LOCKUP_DETECTOR
394 extern void touch_softlockup_watchdog(void);
395 extern void touch_softlockup_watchdog_sync(void);
396 extern void touch_all_softlockup_watchdogs(void);
397 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
399 size_t *lenp, loff_t *ppos);
400 extern unsigned int softlockup_panic;
401 extern unsigned int hardlockup_panic;
402 void lockup_detector_init(void);
404 static inline void touch_softlockup_watchdog(void)
407 static inline void touch_softlockup_watchdog_sync(void)
410 static inline void touch_all_softlockup_watchdogs(void)
413 static inline void lockup_detector_init(void)
418 #ifdef CONFIG_DETECT_HUNG_TASK
419 void reset_hung_task_detector(void);
421 static inline void reset_hung_task_detector(void)
426 /* Attach to any functions which should be ignored in wchan output. */
427 #define __sched __attribute__((__section__(".sched.text")))
429 /* Linker adds these: start and end of __sched functions */
430 extern char __sched_text_start[], __sched_text_end[];
432 /* Is this address in the __sched functions? */
433 extern int in_sched_functions(unsigned long addr);
435 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
436 extern signed long schedule_timeout(signed long timeout);
437 extern signed long schedule_timeout_interruptible(signed long timeout);
438 extern signed long schedule_timeout_killable(signed long timeout);
439 extern signed long schedule_timeout_uninterruptible(signed long timeout);
440 asmlinkage void schedule(void);
441 extern void schedule_preempt_disabled(void);
443 extern long io_schedule_timeout(long timeout);
445 static inline void io_schedule(void)
447 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
451 struct user_namespace;
454 extern void arch_pick_mmap_layout(struct mm_struct *mm);
456 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
457 unsigned long, unsigned long);
459 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
460 unsigned long len, unsigned long pgoff,
461 unsigned long flags);
463 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
466 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
467 #define SUID_DUMP_USER 1 /* Dump as user of process */
468 #define SUID_DUMP_ROOT 2 /* Dump as root */
472 /* for SUID_DUMP_* above */
473 #define MMF_DUMPABLE_BITS 2
474 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
476 extern void set_dumpable(struct mm_struct *mm, int value);
478 * This returns the actual value of the suid_dumpable flag. For things
479 * that are using this for checking for privilege transitions, it must
480 * test against SUID_DUMP_USER rather than treating it as a boolean
483 static inline int __get_dumpable(unsigned long mm_flags)
485 return mm_flags & MMF_DUMPABLE_MASK;
488 static inline int get_dumpable(struct mm_struct *mm)
490 return __get_dumpable(mm->flags);
493 /* coredump filter bits */
494 #define MMF_DUMP_ANON_PRIVATE 2
495 #define MMF_DUMP_ANON_SHARED 3
496 #define MMF_DUMP_MAPPED_PRIVATE 4
497 #define MMF_DUMP_MAPPED_SHARED 5
498 #define MMF_DUMP_ELF_HEADERS 6
499 #define MMF_DUMP_HUGETLB_PRIVATE 7
500 #define MMF_DUMP_HUGETLB_SHARED 8
501 #define MMF_DUMP_DAX_PRIVATE 9
502 #define MMF_DUMP_DAX_SHARED 10
504 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
505 #define MMF_DUMP_FILTER_BITS 9
506 #define MMF_DUMP_FILTER_MASK \
507 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
508 #define MMF_DUMP_FILTER_DEFAULT \
509 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
510 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
512 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
513 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
515 # define MMF_DUMP_MASK_DEFAULT_ELF 0
517 /* leave room for more dump flags */
518 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
519 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
520 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
522 #define MMF_HAS_UPROBES 19 /* has uprobes */
523 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
525 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
527 struct sighand_struct {
529 struct k_sigaction action[_NSIG];
531 wait_queue_head_t signalfd_wqh;
534 struct pacct_struct {
537 unsigned long ac_mem;
538 cputime_t ac_utime, ac_stime;
539 unsigned long ac_minflt, ac_majflt;
550 * struct prev_cputime - snaphsot of system and user cputime
551 * @utime: time spent in user mode
552 * @stime: time spent in system mode
553 * @lock: protects the above two fields
555 * Stores previous user/system time values such that we can guarantee
558 struct prev_cputime {
559 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
566 static inline void prev_cputime_init(struct prev_cputime *prev)
568 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
569 prev->utime = prev->stime = 0;
570 raw_spin_lock_init(&prev->lock);
575 * struct task_cputime - collected CPU time counts
576 * @utime: time spent in user mode, in &cputime_t units
577 * @stime: time spent in kernel mode, in &cputime_t units
578 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
580 * This structure groups together three kinds of CPU time that are tracked for
581 * threads and thread groups. Most things considering CPU time want to group
582 * these counts together and treat all three of them in parallel.
584 struct task_cputime {
587 unsigned long long sum_exec_runtime;
590 /* Alternate field names when used to cache expirations. */
591 #define virt_exp utime
592 #define prof_exp stime
593 #define sched_exp sum_exec_runtime
595 #define INIT_CPUTIME \
596 (struct task_cputime) { \
599 .sum_exec_runtime = 0, \
603 * This is the atomic variant of task_cputime, which can be used for
604 * storing and updating task_cputime statistics without locking.
606 struct task_cputime_atomic {
609 atomic64_t sum_exec_runtime;
612 #define INIT_CPUTIME_ATOMIC \
613 (struct task_cputime_atomic) { \
614 .utime = ATOMIC64_INIT(0), \
615 .stime = ATOMIC64_INIT(0), \
616 .sum_exec_runtime = ATOMIC64_INIT(0), \
619 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
622 * Disable preemption until the scheduler is running -- use an unconditional
623 * value so that it also works on !PREEMPT_COUNT kernels.
625 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
627 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
630 * Initial preempt_count value; reflects the preempt_count schedule invariant
631 * which states that during context switches:
633 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
635 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
636 * Note: See finish_task_switch().
638 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
641 * struct thread_group_cputimer - thread group interval timer counts
642 * @cputime_atomic: atomic thread group interval timers.
643 * @running: true when there are timers running and
644 * @cputime_atomic receives updates.
645 * @checking_timer: true when a thread in the group is in the
646 * process of checking for thread group timers.
648 * This structure contains the version of task_cputime, above, that is
649 * used for thread group CPU timer calculations.
651 struct thread_group_cputimer {
652 struct task_cputime_atomic cputime_atomic;
657 #include <linux/rwsem.h>
661 * NOTE! "signal_struct" does not have its own
662 * locking, because a shared signal_struct always
663 * implies a shared sighand_struct, so locking
664 * sighand_struct is always a proper superset of
665 * the locking of signal_struct.
667 struct signal_struct {
671 struct list_head thread_head;
673 wait_queue_head_t wait_chldexit; /* for wait4() */
675 /* current thread group signal load-balancing target: */
676 struct task_struct *curr_target;
678 /* shared signal handling: */
679 struct sigpending shared_pending;
681 /* thread group exit support */
684 * - notify group_exit_task when ->count is equal to notify_count
685 * - everyone except group_exit_task is stopped during signal delivery
686 * of fatal signals, group_exit_task processes the signal.
689 struct task_struct *group_exit_task;
691 /* thread group stop support, overloads group_exit_code too */
692 int group_stop_count;
693 unsigned int flags; /* see SIGNAL_* flags below */
696 * PR_SET_CHILD_SUBREAPER marks a process, like a service
697 * manager, to re-parent orphan (double-forking) child processes
698 * to this process instead of 'init'. The service manager is
699 * able to receive SIGCHLD signals and is able to investigate
700 * the process until it calls wait(). All children of this
701 * process will inherit a flag if they should look for a
702 * child_subreaper process at exit.
704 unsigned int is_child_subreaper:1;
705 unsigned int has_child_subreaper:1;
707 /* POSIX.1b Interval Timers */
709 struct list_head posix_timers;
711 /* ITIMER_REAL timer for the process */
712 struct hrtimer real_timer;
713 struct pid *leader_pid;
714 ktime_t it_real_incr;
717 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
718 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
719 * values are defined to 0 and 1 respectively
721 struct cpu_itimer it[2];
724 * Thread group totals for process CPU timers.
725 * See thread_group_cputimer(), et al, for details.
727 struct thread_group_cputimer cputimer;
729 /* Earliest-expiration cache. */
730 struct task_cputime cputime_expires;
732 struct list_head cpu_timers[3];
734 struct pid *tty_old_pgrp;
736 /* boolean value for session group leader */
739 struct tty_struct *tty; /* NULL if no tty */
741 #ifdef CONFIG_SCHED_AUTOGROUP
742 struct autogroup *autogroup;
745 * Cumulative resource counters for dead threads in the group,
746 * and for reaped dead child processes forked by this group.
747 * Live threads maintain their own counters and add to these
748 * in __exit_signal, except for the group leader.
750 seqlock_t stats_lock;
751 cputime_t utime, stime, cutime, cstime;
754 struct prev_cputime prev_cputime;
755 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
756 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
757 unsigned long inblock, oublock, cinblock, coublock;
758 unsigned long maxrss, cmaxrss;
759 struct task_io_accounting ioac;
762 * Cumulative ns of schedule CPU time fo dead threads in the
763 * group, not including a zombie group leader, (This only differs
764 * from jiffies_to_ns(utime + stime) if sched_clock uses something
765 * other than jiffies.)
767 unsigned long long sum_sched_runtime;
770 * We don't bother to synchronize most readers of this at all,
771 * because there is no reader checking a limit that actually needs
772 * to get both rlim_cur and rlim_max atomically, and either one
773 * alone is a single word that can safely be read normally.
774 * getrlimit/setrlimit use task_lock(current->group_leader) to
775 * protect this instead of the siglock, because they really
776 * have no need to disable irqs.
778 struct rlimit rlim[RLIM_NLIMITS];
780 #ifdef CONFIG_BSD_PROCESS_ACCT
781 struct pacct_struct pacct; /* per-process accounting information */
783 #ifdef CONFIG_TASKSTATS
784 struct taskstats *stats;
788 unsigned audit_tty_log_passwd;
789 struct tty_audit_buf *tty_audit_buf;
792 oom_flags_t oom_flags;
793 short oom_score_adj; /* OOM kill score adjustment */
794 short oom_score_adj_min; /* OOM kill score adjustment min value.
795 * Only settable by CAP_SYS_RESOURCE. */
797 struct mutex cred_guard_mutex; /* guard against foreign influences on
798 * credential calculations
799 * (notably. ptrace) */
803 * Bits in flags field of signal_struct.
805 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
806 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
807 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
808 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
810 * Pending notifications to parent.
812 #define SIGNAL_CLD_STOPPED 0x00000010
813 #define SIGNAL_CLD_CONTINUED 0x00000020
814 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
816 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
818 #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
819 SIGNAL_STOP_CONTINUED)
821 static inline void signal_set_stop_flags(struct signal_struct *sig,
824 WARN_ON(sig->flags & (SIGNAL_GROUP_EXIT|SIGNAL_GROUP_COREDUMP));
825 sig->flags = (sig->flags & ~SIGNAL_STOP_MASK) | flags;
828 /* If true, all threads except ->group_exit_task have pending SIGKILL */
829 static inline int signal_group_exit(const struct signal_struct *sig)
831 return (sig->flags & SIGNAL_GROUP_EXIT) ||
832 (sig->group_exit_task != NULL);
836 * Some day this will be a full-fledged user tracking system..
839 atomic_t __count; /* reference count */
840 atomic_t processes; /* How many processes does this user have? */
841 atomic_t sigpending; /* How many pending signals does this user have? */
842 #ifdef CONFIG_INOTIFY_USER
843 atomic_t inotify_watches; /* How many inotify watches does this user have? */
844 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
846 #ifdef CONFIG_FANOTIFY
847 atomic_t fanotify_listeners;
850 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
852 #ifdef CONFIG_POSIX_MQUEUE
853 /* protected by mq_lock */
854 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
856 unsigned long locked_shm; /* How many pages of mlocked shm ? */
857 unsigned long unix_inflight; /* How many files in flight in unix sockets */
858 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
861 struct key *uid_keyring; /* UID specific keyring */
862 struct key *session_keyring; /* UID's default session keyring */
865 /* Hash table maintenance information */
866 struct hlist_node uidhash_node;
869 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
870 atomic_long_t locked_vm;
874 extern int uids_sysfs_init(void);
876 extern struct user_struct *find_user(kuid_t);
878 extern struct user_struct root_user;
879 #define INIT_USER (&root_user)
882 struct backing_dev_info;
883 struct reclaim_state;
885 #ifdef CONFIG_SCHED_INFO
887 /* cumulative counters */
888 unsigned long pcount; /* # of times run on this cpu */
889 unsigned long long run_delay; /* time spent waiting on a runqueue */
892 unsigned long long last_arrival,/* when we last ran on a cpu */
893 last_queued; /* when we were last queued to run */
895 #endif /* CONFIG_SCHED_INFO */
897 #ifdef CONFIG_TASK_DELAY_ACCT
898 struct task_delay_info {
900 unsigned int flags; /* Private per-task flags */
902 /* For each stat XXX, add following, aligned appropriately
904 * struct timespec XXX_start, XXX_end;
908 * Atomicity of updates to XXX_delay, XXX_count protected by
909 * single lock above (split into XXX_lock if contention is an issue).
913 * XXX_count is incremented on every XXX operation, the delay
914 * associated with the operation is added to XXX_delay.
915 * XXX_delay contains the accumulated delay time in nanoseconds.
917 u64 blkio_start; /* Shared by blkio, swapin */
918 u64 blkio_delay; /* wait for sync block io completion */
919 u64 swapin_delay; /* wait for swapin block io completion */
920 u32 blkio_count; /* total count of the number of sync block */
921 /* io operations performed */
922 u32 swapin_count; /* total count of the number of swapin block */
923 /* io operations performed */
926 u64 freepages_delay; /* wait for memory reclaim */
927 u32 freepages_count; /* total count of memory reclaim */
929 #endif /* CONFIG_TASK_DELAY_ACCT */
931 static inline int sched_info_on(void)
933 #ifdef CONFIG_SCHEDSTATS
935 #elif defined(CONFIG_TASK_DELAY_ACCT)
936 extern int delayacct_on;
951 * Increase resolution of cpu_capacity calculations
953 #define SCHED_CAPACITY_SHIFT 10
954 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
956 struct sched_capacity_reqs {
965 * Wake-queues are lists of tasks with a pending wakeup, whose
966 * callers have already marked the task as woken internally,
967 * and can thus carry on. A common use case is being able to
968 * do the wakeups once the corresponding user lock as been
971 * We hold reference to each task in the list across the wakeup,
972 * thus guaranteeing that the memory is still valid by the time
973 * the actual wakeups are performed in wake_up_q().
975 * One per task suffices, because there's never a need for a task to be
976 * in two wake queues simultaneously; it is forbidden to abandon a task
977 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
978 * already in a wake queue, the wakeup will happen soon and the second
979 * waker can just skip it.
981 * The WAKE_Q macro declares and initializes the list head.
982 * wake_up_q() does NOT reinitialize the list; it's expected to be
983 * called near the end of a function, where the fact that the queue is
984 * not used again will be easy to see by inspection.
986 * Note that this can cause spurious wakeups. schedule() callers
987 * must ensure the call is done inside a loop, confirming that the
988 * wakeup condition has in fact occurred.
991 struct wake_q_node *next;
995 struct wake_q_node *first;
996 struct wake_q_node **lastp;
1000 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1002 #define WAKE_Q(name) \
1003 struct wake_q_head name = { WAKE_Q_TAIL, &name.first, 0 }
1005 extern void wake_q_add(struct wake_q_head *head,
1006 struct task_struct *task);
1007 extern void wake_up_q(struct wake_q_head *head);
1010 * sched-domains (multiprocessor balancing) declarations:
1013 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1014 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1015 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1016 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
1017 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
1018 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
1019 #define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
1020 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
1021 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1022 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1023 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1024 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1025 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1026 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1027 #define SD_NUMA 0x4000 /* cross-node balancing */
1028 #define SD_SHARE_CAP_STATES 0x8000 /* Domain members share capacity state */
1030 #ifdef CONFIG_SCHED_SMT
1031 static inline int cpu_smt_flags(void)
1033 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1037 #ifdef CONFIG_SCHED_MC
1038 static inline int cpu_core_flags(void)
1040 return SD_SHARE_PKG_RESOURCES;
1045 static inline int cpu_numa_flags(void)
1051 struct sched_domain_attr {
1052 int relax_domain_level;
1055 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1056 .relax_domain_level = -1, \
1059 extern int sched_domain_level_max;
1061 struct capacity_state {
1062 unsigned long cap; /* compute capacity */
1063 unsigned long power; /* power consumption at this compute capacity */
1067 unsigned long power; /* power consumption in this idle state */
1070 struct sched_group_energy {
1071 unsigned int nr_idle_states; /* number of idle states */
1072 struct idle_state *idle_states; /* ptr to idle state array */
1073 unsigned int nr_cap_states; /* number of capacity states */
1074 struct capacity_state *cap_states; /* ptr to capacity state array */
1077 unsigned long capacity_curr_of(int cpu);
1082 /* select_idle_sibling() stats */
1085 u64 sis_cache_affine;
1090 /* select_energy_cpu_brute() stats */
1094 u64 secb_insuff_cap;
1095 u64 secb_no_nrg_sav;
1099 /* find_best_target() stats */
1107 /* select_task_rq_fair() stats */
1112 struct sched_domain {
1113 /* These fields must be setup */
1114 struct sched_domain *parent; /* top domain must be null terminated */
1115 struct sched_domain *child; /* bottom domain must be null terminated */
1116 struct sched_group *groups; /* the balancing groups of the domain */
1117 unsigned long min_interval; /* Minimum balance interval ms */
1118 unsigned long max_interval; /* Maximum balance interval ms */
1119 unsigned int busy_factor; /* less balancing by factor if busy */
1120 unsigned int imbalance_pct; /* No balance until over watermark */
1121 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1122 unsigned int busy_idx;
1123 unsigned int idle_idx;
1124 unsigned int newidle_idx;
1125 unsigned int wake_idx;
1126 unsigned int forkexec_idx;
1127 unsigned int smt_gain;
1129 int nohz_idle; /* NOHZ IDLE status */
1130 int flags; /* See SD_* */
1133 /* Runtime fields. */
1134 unsigned long last_balance; /* init to jiffies. units in jiffies */
1135 unsigned int balance_interval; /* initialise to 1. units in ms. */
1136 unsigned int nr_balance_failed; /* initialise to 0 */
1138 /* idle_balance() stats */
1139 u64 max_newidle_lb_cost;
1140 unsigned long next_decay_max_lb_cost;
1142 #ifdef CONFIG_SCHEDSTATS
1143 /* load_balance() stats */
1144 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1145 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1146 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1147 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1148 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1149 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1150 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1151 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1153 /* Active load balancing */
1154 unsigned int alb_count;
1155 unsigned int alb_failed;
1156 unsigned int alb_pushed;
1158 /* SD_BALANCE_EXEC stats */
1159 unsigned int sbe_count;
1160 unsigned int sbe_balanced;
1161 unsigned int sbe_pushed;
1163 /* SD_BALANCE_FORK stats */
1164 unsigned int sbf_count;
1165 unsigned int sbf_balanced;
1166 unsigned int sbf_pushed;
1168 /* try_to_wake_up() stats */
1169 unsigned int ttwu_wake_remote;
1170 unsigned int ttwu_move_affine;
1171 unsigned int ttwu_move_balance;
1173 struct eas_stats eas_stats;
1175 #ifdef CONFIG_SCHED_DEBUG
1179 void *private; /* used during construction */
1180 struct rcu_head rcu; /* used during destruction */
1183 unsigned int span_weight;
1185 * Span of all CPUs in this domain.
1187 * NOTE: this field is variable length. (Allocated dynamically
1188 * by attaching extra space to the end of the structure,
1189 * depending on how many CPUs the kernel has booted up with)
1191 unsigned long span[0];
1194 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1196 return to_cpumask(sd->span);
1199 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1200 struct sched_domain_attr *dattr_new);
1202 /* Allocate an array of sched domains, for partition_sched_domains(). */
1203 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1204 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1206 bool cpus_share_cache(int this_cpu, int that_cpu);
1208 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1209 typedef int (*sched_domain_flags_f)(void);
1211 const struct sched_group_energy * const(*sched_domain_energy_f)(int cpu);
1213 #define SDTL_OVERLAP 0x01
1216 struct sched_domain **__percpu sd;
1217 struct sched_group **__percpu sg;
1218 struct sched_group_capacity **__percpu sgc;
1221 struct sched_domain_topology_level {
1222 sched_domain_mask_f mask;
1223 sched_domain_flags_f sd_flags;
1224 sched_domain_energy_f energy;
1227 struct sd_data data;
1228 #ifdef CONFIG_SCHED_DEBUG
1233 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1234 extern void wake_up_if_idle(int cpu);
1236 #ifdef CONFIG_SCHED_DEBUG
1237 # define SD_INIT_NAME(type) .name = #type
1239 # define SD_INIT_NAME(type)
1242 #else /* CONFIG_SMP */
1244 struct sched_domain_attr;
1247 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1248 struct sched_domain_attr *dattr_new)
1252 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1257 #endif /* !CONFIG_SMP */
1260 struct io_context; /* See blkdev.h */
1263 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1264 extern void prefetch_stack(struct task_struct *t);
1266 static inline void prefetch_stack(struct task_struct *t) { }
1269 struct audit_context; /* See audit.c */
1271 struct pipe_inode_info;
1272 struct uts_namespace;
1274 struct load_weight {
1275 unsigned long weight;
1280 * The load_avg/util_avg accumulates an infinite geometric series.
1281 * 1) load_avg factors frequency scaling into the amount of time that a
1282 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1283 * aggregated such weights of all runnable and blocked sched_entities.
1284 * 2) util_avg factors frequency and cpu scaling into the amount of time
1285 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1286 * For cfs_rq, it is the aggregated such times of all runnable and
1287 * blocked sched_entities.
1288 * The 64 bit load_sum can:
1289 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1290 * the highest weight (=88761) always runnable, we should not overflow
1291 * 2) for entity, support any load.weight always runnable
1294 u64 last_update_time, load_sum;
1295 u32 util_sum, period_contrib;
1296 unsigned long load_avg, util_avg;
1299 #ifdef CONFIG_SCHEDSTATS
1300 struct sched_statistics {
1310 s64 sum_sleep_runtime;
1317 u64 nr_migrations_cold;
1318 u64 nr_failed_migrations_affine;
1319 u64 nr_failed_migrations_running;
1320 u64 nr_failed_migrations_hot;
1321 u64 nr_forced_migrations;
1324 u64 nr_wakeups_sync;
1325 u64 nr_wakeups_migrate;
1326 u64 nr_wakeups_local;
1327 u64 nr_wakeups_remote;
1328 u64 nr_wakeups_affine;
1329 u64 nr_wakeups_affine_attempts;
1330 u64 nr_wakeups_passive;
1331 u64 nr_wakeups_idle;
1333 /* select_idle_sibling() */
1334 u64 nr_wakeups_sis_attempts;
1335 u64 nr_wakeups_sis_idle;
1336 u64 nr_wakeups_sis_cache_affine;
1337 u64 nr_wakeups_sis_suff_cap;
1338 u64 nr_wakeups_sis_idle_cpu;
1339 u64 nr_wakeups_sis_count;
1341 /* energy_aware_wake_cpu() */
1342 u64 nr_wakeups_secb_attempts;
1343 u64 nr_wakeups_secb_sync;
1344 u64 nr_wakeups_secb_idle_bt;
1345 u64 nr_wakeups_secb_insuff_cap;
1346 u64 nr_wakeups_secb_no_nrg_sav;
1347 u64 nr_wakeups_secb_nrg_sav;
1348 u64 nr_wakeups_secb_count;
1350 /* find_best_target() */
1351 u64 nr_wakeups_fbt_attempts;
1352 u64 nr_wakeups_fbt_no_cpu;
1353 u64 nr_wakeups_fbt_no_sd;
1354 u64 nr_wakeups_fbt_pref_idle;
1355 u64 nr_wakeups_fbt_count;
1358 /* select_task_rq_fair() */
1359 u64 nr_wakeups_cas_attempts;
1360 u64 nr_wakeups_cas_count;
1364 #ifdef CONFIG_SCHED_WALT
1365 #define RAVG_HIST_SIZE_MAX 5
1367 /* ravg represents frequency scaled cpu-demand of tasks */
1370 * 'mark_start' marks the beginning of an event (task waking up, task
1371 * starting to execute, task being preempted) within a window
1373 * 'sum' represents how runnable a task has been within current
1374 * window. It incorporates both running time and wait time and is
1377 * 'sum_history' keeps track of history of 'sum' seen over previous
1378 * RAVG_HIST_SIZE windows. Windows where task was entirely sleeping are
1381 * 'demand' represents maximum sum seen over previous
1382 * sysctl_sched_ravg_hist_size windows. 'demand' could drive frequency
1385 * 'curr_window' represents task's contribution to cpu busy time
1386 * statistics (rq->curr_runnable_sum) in current window
1388 * 'prev_window' represents task's contribution to cpu busy time
1389 * statistics (rq->prev_runnable_sum) in previous window
1393 u32 sum_history[RAVG_HIST_SIZE_MAX];
1394 u32 curr_window, prev_window;
1399 struct sched_entity {
1400 struct load_weight load; /* for load-balancing */
1401 struct rb_node run_node;
1402 struct list_head group_node;
1406 u64 sum_exec_runtime;
1408 u64 prev_sum_exec_runtime;
1412 #ifdef CONFIG_SCHEDSTATS
1413 struct sched_statistics statistics;
1416 #ifdef CONFIG_FAIR_GROUP_SCHED
1418 struct sched_entity *parent;
1419 /* rq on which this entity is (to be) queued: */
1420 struct cfs_rq *cfs_rq;
1421 /* rq "owned" by this entity/group: */
1422 struct cfs_rq *my_q;
1426 /* Per entity load average tracking */
1427 struct sched_avg avg;
1431 struct sched_rt_entity {
1432 struct list_head run_list;
1433 unsigned long timeout;
1434 unsigned long watchdog_stamp;
1435 unsigned int time_slice;
1437 /* Accesses for these must be guarded by rq->lock of the task's rq */
1438 bool schedtune_enqueued;
1439 struct hrtimer schedtune_timer;
1441 struct sched_rt_entity *back;
1442 #ifdef CONFIG_RT_GROUP_SCHED
1443 struct sched_rt_entity *parent;
1444 /* rq on which this entity is (to be) queued: */
1445 struct rt_rq *rt_rq;
1446 /* rq "owned" by this entity/group: */
1451 struct sched_dl_entity {
1452 struct rb_node rb_node;
1455 * Original scheduling parameters. Copied here from sched_attr
1456 * during sched_setattr(), they will remain the same until
1457 * the next sched_setattr().
1459 u64 dl_runtime; /* maximum runtime for each instance */
1460 u64 dl_deadline; /* relative deadline of each instance */
1461 u64 dl_period; /* separation of two instances (period) */
1462 u64 dl_bw; /* dl_runtime / dl_deadline */
1463 u64 dl_density; /* dl_runtime / dl_deadline */
1466 * Actual scheduling parameters. Initialized with the values above,
1467 * they are continously updated during task execution. Note that
1468 * the remaining runtime could be < 0 in case we are in overrun.
1470 s64 runtime; /* remaining runtime for this instance */
1471 u64 deadline; /* absolute deadline for this instance */
1472 unsigned int flags; /* specifying the scheduler behaviour */
1477 * @dl_throttled tells if we exhausted the runtime. If so, the
1478 * task has to wait for a replenishment to be performed at the
1479 * next firing of dl_timer.
1481 * @dl_new tells if a new instance arrived. If so we must
1482 * start executing it with full runtime and reset its absolute
1485 * @dl_boosted tells if we are boosted due to DI. If so we are
1486 * outside bandwidth enforcement mechanism (but only until we
1487 * exit the critical section);
1489 * @dl_yielded tells if task gave up the cpu before consuming
1490 * all its available runtime during the last job.
1492 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1495 * Bandwidth enforcement timer. Each -deadline task has its
1496 * own bandwidth to be enforced, thus we need one timer per task.
1498 struct hrtimer dl_timer;
1506 u8 pad; /* Otherwise the compiler can store garbage here. */
1508 u32 s; /* Set of bits. */
1512 enum perf_event_task_context {
1513 perf_invalid_context = -1,
1514 perf_hw_context = 0,
1516 perf_nr_task_contexts,
1519 /* Track pages that require TLB flushes */
1520 struct tlbflush_unmap_batch {
1522 * Each bit set is a CPU that potentially has a TLB entry for one of
1523 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1525 struct cpumask cpumask;
1527 /* True if any bit in cpumask is set */
1528 bool flush_required;
1531 * If true then the PTE was dirty when unmapped. The entry must be
1532 * flushed before IO is initiated or a stale TLB entry potentially
1533 * allows an update without redirtying the page.
1538 struct task_struct {
1539 #ifdef CONFIG_THREAD_INFO_IN_TASK
1541 * For reasons of header soup (see current_thread_info()), this
1542 * must be the first element of task_struct.
1544 struct thread_info thread_info;
1546 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1549 unsigned int flags; /* per process flags, defined below */
1550 unsigned int ptrace;
1553 struct llist_node wake_entry;
1555 #ifdef CONFIG_THREAD_INFO_IN_TASK
1556 unsigned int cpu; /* current CPU */
1558 unsigned int wakee_flips;
1559 unsigned long wakee_flip_decay_ts;
1560 struct task_struct *last_wakee;
1566 int prio, static_prio, normal_prio;
1567 unsigned int rt_priority;
1568 const struct sched_class *sched_class;
1569 struct sched_entity se;
1570 struct sched_rt_entity rt;
1571 #ifdef CONFIG_SCHED_WALT
1574 * 'init_load_pct' represents the initial task load assigned to children
1581 #ifdef CONFIG_CGROUP_SCHED
1582 struct task_group *sched_task_group;
1584 struct sched_dl_entity dl;
1586 #ifdef CONFIG_PREEMPT_NOTIFIERS
1587 /* list of struct preempt_notifier: */
1588 struct hlist_head preempt_notifiers;
1591 #ifdef CONFIG_BLK_DEV_IO_TRACE
1592 unsigned int btrace_seq;
1595 unsigned int policy;
1596 int nr_cpus_allowed;
1597 cpumask_t cpus_allowed;
1599 #ifdef CONFIG_PREEMPT_RCU
1600 int rcu_read_lock_nesting;
1601 union rcu_special rcu_read_unlock_special;
1602 struct list_head rcu_node_entry;
1603 struct rcu_node *rcu_blocked_node;
1604 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1605 #ifdef CONFIG_TASKS_RCU
1606 unsigned long rcu_tasks_nvcsw;
1607 bool rcu_tasks_holdout;
1608 struct list_head rcu_tasks_holdout_list;
1609 int rcu_tasks_idle_cpu;
1610 #endif /* #ifdef CONFIG_TASKS_RCU */
1612 #ifdef CONFIG_SCHED_INFO
1613 struct sched_info sched_info;
1616 struct list_head tasks;
1618 struct plist_node pushable_tasks;
1619 struct rb_node pushable_dl_tasks;
1622 struct mm_struct *mm, *active_mm;
1623 /* per-thread vma caching */
1624 u32 vmacache_seqnum;
1625 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1626 #if defined(SPLIT_RSS_COUNTING)
1627 struct task_rss_stat rss_stat;
1631 int exit_code, exit_signal;
1632 int pdeath_signal; /* The signal sent when the parent dies */
1633 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1635 /* Used for emulating ABI behavior of previous Linux versions */
1636 unsigned int personality;
1638 /* scheduler bits, serialized by scheduler locks */
1639 unsigned sched_reset_on_fork:1;
1640 unsigned sched_contributes_to_load:1;
1641 unsigned sched_migrated:1;
1642 unsigned :0; /* force alignment to the next boundary */
1644 /* unserialized, strictly 'current' */
1645 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1646 unsigned in_iowait:1;
1648 unsigned memcg_may_oom:1;
1650 #ifdef CONFIG_MEMCG_KMEM
1651 unsigned memcg_kmem_skip_account:1;
1653 #ifdef CONFIG_COMPAT_BRK
1654 unsigned brk_randomized:1;
1656 #ifdef CONFIG_CGROUPS
1657 /* disallow userland-initiated cgroup migration */
1658 unsigned no_cgroup_migration:1;
1661 unsigned long atomic_flags; /* Flags needing atomic access. */
1663 struct restart_block restart_block;
1668 #ifdef CONFIG_CC_STACKPROTECTOR
1669 /* Canary value for the -fstack-protector gcc feature */
1670 unsigned long stack_canary;
1673 * pointers to (original) parent process, youngest child, younger sibling,
1674 * older sibling, respectively. (p->father can be replaced with
1675 * p->real_parent->pid)
1677 struct task_struct __rcu *real_parent; /* real parent process */
1678 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1680 * children/sibling forms the list of my natural children
1682 struct list_head children; /* list of my children */
1683 struct list_head sibling; /* linkage in my parent's children list */
1684 struct task_struct *group_leader; /* threadgroup leader */
1687 * ptraced is the list of tasks this task is using ptrace on.
1688 * This includes both natural children and PTRACE_ATTACH targets.
1689 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1691 struct list_head ptraced;
1692 struct list_head ptrace_entry;
1694 /* PID/PID hash table linkage. */
1695 struct pid_link pids[PIDTYPE_MAX];
1696 struct list_head thread_group;
1697 struct list_head thread_node;
1699 struct completion *vfork_done; /* for vfork() */
1700 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1701 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1703 cputime_t utime, stime, utimescaled, stimescaled;
1705 #ifdef CONFIG_CPU_FREQ_TIMES
1707 unsigned int max_state;
1709 struct prev_cputime prev_cputime;
1710 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1711 seqlock_t vtime_seqlock;
1712 unsigned long long vtime_snap;
1717 } vtime_snap_whence;
1719 unsigned long nvcsw, nivcsw; /* context switch counts */
1720 u64 start_time; /* monotonic time in nsec */
1721 u64 real_start_time; /* boot based time in nsec */
1722 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1723 unsigned long min_flt, maj_flt;
1725 struct task_cputime cputime_expires;
1726 struct list_head cpu_timers[3];
1728 /* process credentials */
1729 const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
1730 const struct cred __rcu *real_cred; /* objective and real subjective task
1731 * credentials (COW) */
1732 const struct cred __rcu *cred; /* effective (overridable) subjective task
1733 * credentials (COW) */
1734 char comm[TASK_COMM_LEN]; /* executable name excluding path
1735 - access with [gs]et_task_comm (which lock
1736 it with task_lock())
1737 - initialized normally by setup_new_exec */
1738 /* file system info */
1739 struct nameidata *nameidata;
1740 #ifdef CONFIG_SYSVIPC
1742 struct sysv_sem sysvsem;
1743 struct sysv_shm sysvshm;
1745 #ifdef CONFIG_DETECT_HUNG_TASK
1746 /* hung task detection */
1747 unsigned long last_switch_count;
1749 /* filesystem information */
1750 struct fs_struct *fs;
1751 /* open file information */
1752 struct files_struct *files;
1754 struct nsproxy *nsproxy;
1755 /* signal handlers */
1756 struct signal_struct *signal;
1757 struct sighand_struct *sighand;
1759 sigset_t blocked, real_blocked;
1760 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1761 struct sigpending pending;
1763 unsigned long sas_ss_sp;
1766 struct callback_head *task_works;
1768 struct audit_context *audit_context;
1769 #ifdef CONFIG_AUDITSYSCALL
1771 unsigned int sessionid;
1773 struct seccomp seccomp;
1775 /* Thread group tracking */
1778 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1780 spinlock_t alloc_lock;
1782 /* Protection of the PI data structures: */
1783 raw_spinlock_t pi_lock;
1785 struct wake_q_node wake_q;
1787 #ifdef CONFIG_RT_MUTEXES
1788 /* PI waiters blocked on a rt_mutex held by this task */
1789 struct rb_root pi_waiters;
1790 struct rb_node *pi_waiters_leftmost;
1791 /* Deadlock detection and priority inheritance handling */
1792 struct rt_mutex_waiter *pi_blocked_on;
1795 #ifdef CONFIG_DEBUG_MUTEXES
1796 /* mutex deadlock detection */
1797 struct mutex_waiter *blocked_on;
1799 #ifdef CONFIG_TRACE_IRQFLAGS
1800 unsigned int irq_events;
1801 unsigned long hardirq_enable_ip;
1802 unsigned long hardirq_disable_ip;
1803 unsigned int hardirq_enable_event;
1804 unsigned int hardirq_disable_event;
1805 int hardirqs_enabled;
1806 int hardirq_context;
1807 unsigned long softirq_disable_ip;
1808 unsigned long softirq_enable_ip;
1809 unsigned int softirq_disable_event;
1810 unsigned int softirq_enable_event;
1811 int softirqs_enabled;
1812 int softirq_context;
1814 #ifdef CONFIG_LOCKDEP
1815 # define MAX_LOCK_DEPTH 48UL
1818 unsigned int lockdep_recursion;
1819 struct held_lock held_locks[MAX_LOCK_DEPTH];
1820 gfp_t lockdep_reclaim_gfp;
1823 /* journalling filesystem info */
1826 /* stacked block device info */
1827 struct bio_list *bio_list;
1830 /* stack plugging */
1831 struct blk_plug *plug;
1835 struct reclaim_state *reclaim_state;
1837 struct backing_dev_info *backing_dev_info;
1839 struct io_context *io_context;
1841 unsigned long ptrace_message;
1842 siginfo_t *last_siginfo; /* For ptrace use. */
1843 struct task_io_accounting ioac;
1844 #if defined(CONFIG_TASK_XACCT)
1845 u64 acct_rss_mem1; /* accumulated rss usage */
1846 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1847 cputime_t acct_timexpd; /* stime + utime since last update */
1849 #ifdef CONFIG_CPUSETS
1850 nodemask_t mems_allowed; /* Protected by alloc_lock */
1851 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1852 int cpuset_mem_spread_rotor;
1853 int cpuset_slab_spread_rotor;
1855 #ifdef CONFIG_CGROUPS
1856 /* Control Group info protected by css_set_lock */
1857 struct css_set __rcu *cgroups;
1858 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1859 struct list_head cg_list;
1862 struct robust_list_head __user *robust_list;
1863 #ifdef CONFIG_COMPAT
1864 struct compat_robust_list_head __user *compat_robust_list;
1866 struct list_head pi_state_list;
1867 struct futex_pi_state *pi_state_cache;
1869 #ifdef CONFIG_PERF_EVENTS
1870 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1871 struct mutex perf_event_mutex;
1872 struct list_head perf_event_list;
1874 #ifdef CONFIG_DEBUG_PREEMPT
1875 unsigned long preempt_disable_ip;
1878 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1880 short pref_node_fork;
1882 #ifdef CONFIG_NUMA_BALANCING
1884 unsigned int numa_scan_period;
1885 unsigned int numa_scan_period_max;
1886 int numa_preferred_nid;
1887 unsigned long numa_migrate_retry;
1888 u64 node_stamp; /* migration stamp */
1889 u64 last_task_numa_placement;
1890 u64 last_sum_exec_runtime;
1891 struct callback_head numa_work;
1893 struct list_head numa_entry;
1894 struct numa_group *numa_group;
1897 * numa_faults is an array split into four regions:
1898 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1899 * in this precise order.
1901 * faults_memory: Exponential decaying average of faults on a per-node
1902 * basis. Scheduling placement decisions are made based on these
1903 * counts. The values remain static for the duration of a PTE scan.
1904 * faults_cpu: Track the nodes the process was running on when a NUMA
1905 * hinting fault was incurred.
1906 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1907 * during the current scan window. When the scan completes, the counts
1908 * in faults_memory and faults_cpu decay and these values are copied.
1910 unsigned long *numa_faults;
1911 unsigned long total_numa_faults;
1914 * numa_faults_locality tracks if faults recorded during the last
1915 * scan window were remote/local or failed to migrate. The task scan
1916 * period is adapted based on the locality of the faults with different
1917 * weights depending on whether they were shared or private faults
1919 unsigned long numa_faults_locality[3];
1921 unsigned long numa_pages_migrated;
1922 #endif /* CONFIG_NUMA_BALANCING */
1924 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1925 struct tlbflush_unmap_batch tlb_ubc;
1928 struct rcu_head rcu;
1931 * cache last used pipe for splice
1933 struct pipe_inode_info *splice_pipe;
1935 struct page_frag task_frag;
1937 #ifdef CONFIG_TASK_DELAY_ACCT
1938 struct task_delay_info *delays;
1940 #ifdef CONFIG_FAULT_INJECTION
1944 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1945 * balance_dirty_pages() for some dirty throttling pause
1948 int nr_dirtied_pause;
1949 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1951 #ifdef CONFIG_LATENCYTOP
1952 int latency_record_count;
1953 struct latency_record latency_record[LT_SAVECOUNT];
1956 * time slack values; these are used to round up poll() and
1957 * select() etc timeout values. These are in nanoseconds.
1960 u64 default_timer_slack_ns;
1963 unsigned int kasan_depth;
1965 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1966 /* Index of current stored address in ret_stack */
1968 /* Stack of return addresses for return function tracing */
1969 struct ftrace_ret_stack *ret_stack;
1970 /* time stamp for last schedule */
1971 unsigned long long ftrace_timestamp;
1973 * Number of functions that haven't been traced
1974 * because of depth overrun.
1976 atomic_t trace_overrun;
1977 /* Pause for the tracing */
1978 atomic_t tracing_graph_pause;
1980 #ifdef CONFIG_TRACING
1981 /* state flags for use by tracers */
1982 unsigned long trace;
1983 /* bitmask and counter of trace recursion */
1984 unsigned long trace_recursion;
1985 #endif /* CONFIG_TRACING */
1987 /* Coverage collection mode enabled for this task (0 if disabled). */
1988 enum kcov_mode kcov_mode;
1989 /* Size of the kcov_area. */
1991 /* Buffer for coverage collection. */
1993 /* kcov desciptor wired with this task or NULL. */
1997 struct mem_cgroup *memcg_in_oom;
1998 gfp_t memcg_oom_gfp_mask;
1999 int memcg_oom_order;
2001 /* number of pages to reclaim on returning to userland */
2002 unsigned int memcg_nr_pages_over_high;
2004 #ifdef CONFIG_UPROBES
2005 struct uprobe_task *utask;
2007 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
2008 unsigned int sequential_io;
2009 unsigned int sequential_io_avg;
2011 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
2012 unsigned long task_state_change;
2014 int pagefault_disabled;
2015 /* CPU-specific state of this task */
2016 struct thread_struct thread;
2018 * WARNING: on x86, 'thread_struct' contains a variable-sized
2019 * structure. It *MUST* be at the end of 'task_struct'.
2021 * Do not put anything below here!
2025 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
2026 extern int arch_task_struct_size __read_mostly;
2028 # define arch_task_struct_size (sizeof(struct task_struct))
2031 /* Future-safe accessor for struct task_struct's cpus_allowed. */
2032 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
2034 #define TNF_MIGRATED 0x01
2035 #define TNF_NO_GROUP 0x02
2036 #define TNF_SHARED 0x04
2037 #define TNF_FAULT_LOCAL 0x08
2038 #define TNF_MIGRATE_FAIL 0x10
2040 #ifdef CONFIG_NUMA_BALANCING
2041 extern void task_numa_fault(int last_node, int node, int pages, int flags);
2042 extern pid_t task_numa_group_id(struct task_struct *p);
2043 extern void set_numabalancing_state(bool enabled);
2044 extern void task_numa_free(struct task_struct *p);
2045 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
2046 int src_nid, int dst_cpu);
2048 static inline void task_numa_fault(int last_node, int node, int pages,
2052 static inline pid_t task_numa_group_id(struct task_struct *p)
2056 static inline void set_numabalancing_state(bool enabled)
2059 static inline void task_numa_free(struct task_struct *p)
2062 static inline bool should_numa_migrate_memory(struct task_struct *p,
2063 struct page *page, int src_nid, int dst_cpu)
2069 static inline struct pid *task_pid(struct task_struct *task)
2071 return task->pids[PIDTYPE_PID].pid;
2074 static inline struct pid *task_tgid(struct task_struct *task)
2076 return task->group_leader->pids[PIDTYPE_PID].pid;
2080 * Without tasklist or rcu lock it is not safe to dereference
2081 * the result of task_pgrp/task_session even if task == current,
2082 * we can race with another thread doing sys_setsid/sys_setpgid.
2084 static inline struct pid *task_pgrp(struct task_struct *task)
2086 return task->group_leader->pids[PIDTYPE_PGID].pid;
2089 static inline struct pid *task_session(struct task_struct *task)
2091 return task->group_leader->pids[PIDTYPE_SID].pid;
2094 struct pid_namespace;
2097 * the helpers to get the task's different pids as they are seen
2098 * from various namespaces
2100 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
2101 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2103 * task_xid_nr_ns() : id seen from the ns specified;
2105 * set_task_vxid() : assigns a virtual id to a task;
2107 * see also pid_nr() etc in include/linux/pid.h
2109 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
2110 struct pid_namespace *ns);
2112 static inline pid_t task_pid_nr(struct task_struct *tsk)
2117 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
2118 struct pid_namespace *ns)
2120 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
2123 static inline pid_t task_pid_vnr(struct task_struct *tsk)
2125 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
2129 static inline pid_t task_tgid_nr(struct task_struct *tsk)
2135 static inline int pid_alive(const struct task_struct *p);
2137 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2138 struct pid_namespace *ns)
2140 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2143 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2145 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2149 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2150 struct pid_namespace *ns)
2152 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2155 static inline pid_t task_session_vnr(struct task_struct *tsk)
2157 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2160 static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns)
2162 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, ns);
2165 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
2167 return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, NULL);
2170 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
2176 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
2182 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
2184 return task_ppid_nr_ns(tsk, &init_pid_ns);
2187 /* obsolete, do not use */
2188 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2190 return task_pgrp_nr_ns(tsk, &init_pid_ns);
2194 * pid_alive - check that a task structure is not stale
2195 * @p: Task structure to be checked.
2197 * Test if a process is not yet dead (at most zombie state)
2198 * If pid_alive fails, then pointers within the task structure
2199 * can be stale and must not be dereferenced.
2201 * Return: 1 if the process is alive. 0 otherwise.
2203 static inline int pid_alive(const struct task_struct *p)
2205 return p->pids[PIDTYPE_PID].pid != NULL;
2209 * is_global_init - check if a task structure is init. Since init
2210 * is free to have sub-threads we need to check tgid.
2211 * @tsk: Task structure to be checked.
2213 * Check if a task structure is the first user space task the kernel created.
2215 * Return: 1 if the task structure is init. 0 otherwise.
2217 static inline int is_global_init(struct task_struct *tsk)
2219 return task_tgid_nr(tsk) == 1;
2222 extern struct pid *cad_pid;
2224 extern void free_task(struct task_struct *tsk);
2225 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2227 extern void __put_task_struct(struct task_struct *t);
2229 static inline void put_task_struct(struct task_struct *t)
2231 if (atomic_dec_and_test(&t->usage))
2232 __put_task_struct(t);
2235 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2236 extern void task_cputime(struct task_struct *t,
2237 cputime_t *utime, cputime_t *stime);
2238 extern void task_cputime_scaled(struct task_struct *t,
2239 cputime_t *utimescaled, cputime_t *stimescaled);
2240 extern cputime_t task_gtime(struct task_struct *t);
2242 static inline void task_cputime(struct task_struct *t,
2243 cputime_t *utime, cputime_t *stime)
2251 static inline void task_cputime_scaled(struct task_struct *t,
2252 cputime_t *utimescaled,
2253 cputime_t *stimescaled)
2256 *utimescaled = t->utimescaled;
2258 *stimescaled = t->stimescaled;
2261 static inline cputime_t task_gtime(struct task_struct *t)
2266 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2267 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2272 #define PF_EXITING 0x00000004 /* getting shut down */
2273 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2274 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2275 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2276 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2277 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2278 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2279 #define PF_DUMPCORE 0x00000200 /* dumped core */
2280 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2281 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2282 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2283 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2284 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2285 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2286 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2287 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2288 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2289 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2290 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2291 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2292 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2293 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2294 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2295 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2296 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2297 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2298 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2301 * Only the _current_ task can read/write to tsk->flags, but other
2302 * tasks can access tsk->flags in readonly mode for example
2303 * with tsk_used_math (like during threaded core dumping).
2304 * There is however an exception to this rule during ptrace
2305 * or during fork: the ptracer task is allowed to write to the
2306 * child->flags of its traced child (same goes for fork, the parent
2307 * can write to the child->flags), because we're guaranteed the
2308 * child is not running and in turn not changing child->flags
2309 * at the same time the parent does it.
2311 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2312 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2313 #define clear_used_math() clear_stopped_child_used_math(current)
2314 #define set_used_math() set_stopped_child_used_math(current)
2315 #define conditional_stopped_child_used_math(condition, child) \
2316 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2317 #define conditional_used_math(condition) \
2318 conditional_stopped_child_used_math(condition, current)
2319 #define copy_to_stopped_child_used_math(child) \
2320 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2321 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2322 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2323 #define used_math() tsk_used_math(current)
2325 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2326 * __GFP_FS is also cleared as it implies __GFP_IO.
2328 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2330 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2331 flags &= ~(__GFP_IO | __GFP_FS);
2335 static inline unsigned int memalloc_noio_save(void)
2337 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2338 current->flags |= PF_MEMALLOC_NOIO;
2342 static inline void memalloc_noio_restore(unsigned int flags)
2344 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2347 /* Per-process atomic flags. */
2348 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2349 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2350 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2353 #define TASK_PFA_TEST(name, func) \
2354 static inline bool task_##func(struct task_struct *p) \
2355 { return test_bit(PFA_##name, &p->atomic_flags); }
2356 #define TASK_PFA_SET(name, func) \
2357 static inline void task_set_##func(struct task_struct *p) \
2358 { set_bit(PFA_##name, &p->atomic_flags); }
2359 #define TASK_PFA_CLEAR(name, func) \
2360 static inline void task_clear_##func(struct task_struct *p) \
2361 { clear_bit(PFA_##name, &p->atomic_flags); }
2363 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2364 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2366 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2367 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2368 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2370 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2371 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2372 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2375 * task->jobctl flags
2377 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2379 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2380 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2381 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2382 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2383 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2384 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2385 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2387 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2388 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2389 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2390 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2391 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2392 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2393 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2395 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2396 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2398 extern bool task_set_jobctl_pending(struct task_struct *task,
2399 unsigned long mask);
2400 extern void task_clear_jobctl_trapping(struct task_struct *task);
2401 extern void task_clear_jobctl_pending(struct task_struct *task,
2402 unsigned long mask);
2404 static inline void rcu_copy_process(struct task_struct *p)
2406 #ifdef CONFIG_PREEMPT_RCU
2407 p->rcu_read_lock_nesting = 0;
2408 p->rcu_read_unlock_special.s = 0;
2409 p->rcu_blocked_node = NULL;
2410 INIT_LIST_HEAD(&p->rcu_node_entry);
2411 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2412 #ifdef CONFIG_TASKS_RCU
2413 p->rcu_tasks_holdout = false;
2414 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2415 p->rcu_tasks_idle_cpu = -1;
2416 #endif /* #ifdef CONFIG_TASKS_RCU */
2419 static inline void tsk_restore_flags(struct task_struct *task,
2420 unsigned long orig_flags, unsigned long flags)
2422 task->flags &= ~flags;
2423 task->flags |= orig_flags & flags;
2426 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2427 const struct cpumask *trial);
2428 extern int task_can_attach(struct task_struct *p,
2429 const struct cpumask *cs_cpus_allowed);
2431 extern void do_set_cpus_allowed(struct task_struct *p,
2432 const struct cpumask *new_mask);
2434 extern int set_cpus_allowed_ptr(struct task_struct *p,
2435 const struct cpumask *new_mask);
2437 static inline void do_set_cpus_allowed(struct task_struct *p,
2438 const struct cpumask *new_mask)
2441 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2442 const struct cpumask *new_mask)
2444 if (!cpumask_test_cpu(0, new_mask))
2450 #ifdef CONFIG_NO_HZ_COMMON
2451 void calc_load_enter_idle(void);
2452 void calc_load_exit_idle(void);
2454 static inline void calc_load_enter_idle(void) { }
2455 static inline void calc_load_exit_idle(void) { }
2456 #endif /* CONFIG_NO_HZ_COMMON */
2459 * Do not use outside of architecture code which knows its limitations.
2461 * sched_clock() has no promise of monotonicity or bounded drift between
2462 * CPUs, use (which you should not) requires disabling IRQs.
2464 * Please use one of the three interfaces below.
2466 extern unsigned long long notrace sched_clock(void);
2468 * See the comment in kernel/sched/clock.c
2470 extern u64 cpu_clock(int cpu);
2471 extern u64 local_clock(void);
2472 extern u64 running_clock(void);
2473 extern u64 sched_clock_cpu(int cpu);
2476 extern void sched_clock_init(void);
2478 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2479 static inline void sched_clock_tick(void)
2483 static inline void sched_clock_idle_sleep_event(void)
2487 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2492 * Architectures can set this to 1 if they have specified
2493 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2494 * but then during bootup it turns out that sched_clock()
2495 * is reliable after all:
2497 extern int sched_clock_stable(void);
2498 extern void set_sched_clock_stable(void);
2499 extern void clear_sched_clock_stable(void);
2501 extern void sched_clock_tick(void);
2502 extern void sched_clock_idle_sleep_event(void);
2503 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2506 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2508 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2509 * The reason for this explicit opt-in is not to have perf penalty with
2510 * slow sched_clocks.
2512 extern void enable_sched_clock_irqtime(void);
2513 extern void disable_sched_clock_irqtime(void);
2515 static inline void enable_sched_clock_irqtime(void) {}
2516 static inline void disable_sched_clock_irqtime(void) {}
2519 extern unsigned long long
2520 task_sched_runtime(struct task_struct *task);
2522 /* sched_exec is called by processes performing an exec */
2524 extern void sched_exec(void);
2526 #define sched_exec() {}
2529 extern void sched_clock_idle_sleep_event(void);
2530 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2532 #ifdef CONFIG_HOTPLUG_CPU
2533 extern void idle_task_exit(void);
2535 static inline void idle_task_exit(void) {}
2538 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2539 extern void wake_up_nohz_cpu(int cpu);
2541 static inline void wake_up_nohz_cpu(int cpu) { }
2544 #ifdef CONFIG_NO_HZ_FULL
2545 extern bool sched_can_stop_tick(void);
2546 extern u64 scheduler_tick_max_deferment(void);
2548 static inline bool sched_can_stop_tick(void) { return false; }
2551 #ifdef CONFIG_SCHED_AUTOGROUP
2552 extern void sched_autogroup_create_attach(struct task_struct *p);
2553 extern void sched_autogroup_detach(struct task_struct *p);
2554 extern void sched_autogroup_fork(struct signal_struct *sig);
2555 extern void sched_autogroup_exit(struct signal_struct *sig);
2556 #ifdef CONFIG_PROC_FS
2557 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2558 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2561 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2562 static inline void sched_autogroup_detach(struct task_struct *p) { }
2563 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2564 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2567 extern int yield_to(struct task_struct *p, bool preempt);
2568 extern void set_user_nice(struct task_struct *p, long nice);
2569 extern int task_prio(const struct task_struct *p);
2571 * task_nice - return the nice value of a given task.
2572 * @p: the task in question.
2574 * Return: The nice value [ -20 ... 0 ... 19 ].
2576 static inline int task_nice(const struct task_struct *p)
2578 return PRIO_TO_NICE((p)->static_prio);
2580 extern int can_nice(const struct task_struct *p, const int nice);
2581 extern int task_curr(const struct task_struct *p);
2582 extern int idle_cpu(int cpu);
2583 extern int sched_setscheduler(struct task_struct *, int,
2584 const struct sched_param *);
2585 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2586 const struct sched_param *);
2587 extern int sched_setattr(struct task_struct *,
2588 const struct sched_attr *);
2589 extern struct task_struct *idle_task(int cpu);
2591 * is_idle_task - is the specified task an idle task?
2592 * @p: the task in question.
2594 * Return: 1 if @p is an idle task. 0 otherwise.
2596 static inline bool is_idle_task(const struct task_struct *p)
2600 extern struct task_struct *curr_task(int cpu);
2601 extern void set_curr_task(int cpu, struct task_struct *p);
2605 union thread_union {
2606 #ifndef CONFIG_THREAD_INFO_IN_TASK
2607 struct thread_info thread_info;
2609 unsigned long stack[THREAD_SIZE/sizeof(long)];
2612 #ifndef __HAVE_ARCH_KSTACK_END
2613 static inline int kstack_end(void *addr)
2615 /* Reliable end of stack detection:
2616 * Some APM bios versions misalign the stack
2618 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2622 extern union thread_union init_thread_union;
2623 extern struct task_struct init_task;
2625 extern struct mm_struct init_mm;
2627 extern struct pid_namespace init_pid_ns;
2630 * find a task by one of its numerical ids
2632 * find_task_by_pid_ns():
2633 * finds a task by its pid in the specified namespace
2634 * find_task_by_vpid():
2635 * finds a task by its virtual pid
2637 * see also find_vpid() etc in include/linux/pid.h
2640 extern struct task_struct *find_task_by_vpid(pid_t nr);
2641 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2642 struct pid_namespace *ns);
2644 /* per-UID process charging. */
2645 extern struct user_struct * alloc_uid(kuid_t);
2646 static inline struct user_struct *get_uid(struct user_struct *u)
2648 atomic_inc(&u->__count);
2651 extern void free_uid(struct user_struct *);
2653 #include <asm/current.h>
2655 extern void xtime_update(unsigned long ticks);
2657 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2658 extern int wake_up_process(struct task_struct *tsk);
2659 extern void wake_up_new_task(struct task_struct *tsk);
2661 extern void kick_process(struct task_struct *tsk);
2663 static inline void kick_process(struct task_struct *tsk) { }
2665 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2666 extern void sched_dead(struct task_struct *p);
2668 extern void proc_caches_init(void);
2669 extern void flush_signals(struct task_struct *);
2670 extern void ignore_signals(struct task_struct *);
2671 extern void flush_signal_handlers(struct task_struct *, int force_default);
2672 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2674 static inline int kernel_dequeue_signal(siginfo_t *info)
2676 struct task_struct *tsk = current;
2680 spin_lock_irq(&tsk->sighand->siglock);
2681 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2682 spin_unlock_irq(&tsk->sighand->siglock);
2687 static inline void kernel_signal_stop(void)
2689 spin_lock_irq(¤t->sighand->siglock);
2690 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2691 __set_current_state(TASK_STOPPED);
2692 spin_unlock_irq(¤t->sighand->siglock);
2697 extern void release_task(struct task_struct * p);
2698 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2699 extern int force_sigsegv(int, struct task_struct *);
2700 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2701 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2702 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2703 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2704 const struct cred *, u32);
2705 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2706 extern int kill_pid(struct pid *pid, int sig, int priv);
2707 extern int kill_proc_info(int, struct siginfo *, pid_t);
2708 extern __must_check bool do_notify_parent(struct task_struct *, int);
2709 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2710 extern void force_sig(int, struct task_struct *);
2711 extern int send_sig(int, struct task_struct *, int);
2712 extern int zap_other_threads(struct task_struct *p);
2713 extern struct sigqueue *sigqueue_alloc(void);
2714 extern void sigqueue_free(struct sigqueue *);
2715 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2716 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2718 static inline void restore_saved_sigmask(void)
2720 if (test_and_clear_restore_sigmask())
2721 __set_current_blocked(¤t->saved_sigmask);
2724 static inline sigset_t *sigmask_to_save(void)
2726 sigset_t *res = ¤t->blocked;
2727 if (unlikely(test_restore_sigmask()))
2728 res = ¤t->saved_sigmask;
2732 static inline int kill_cad_pid(int sig, int priv)
2734 return kill_pid(cad_pid, sig, priv);
2737 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2738 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2739 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2740 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2743 * True if we are on the alternate signal stack.
2745 static inline int on_sig_stack(unsigned long sp)
2747 #ifdef CONFIG_STACK_GROWSUP
2748 return sp >= current->sas_ss_sp &&
2749 sp - current->sas_ss_sp < current->sas_ss_size;
2751 return sp > current->sas_ss_sp &&
2752 sp - current->sas_ss_sp <= current->sas_ss_size;
2756 static inline int sas_ss_flags(unsigned long sp)
2758 if (!current->sas_ss_size)
2761 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2764 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2766 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2767 #ifdef CONFIG_STACK_GROWSUP
2768 return current->sas_ss_sp;
2770 return current->sas_ss_sp + current->sas_ss_size;
2776 * Routines for handling mm_structs
2778 extern struct mm_struct * mm_alloc(void);
2780 /* mmdrop drops the mm and the page tables */
2781 extern void __mmdrop(struct mm_struct *);
2782 static inline void mmdrop(struct mm_struct * mm)
2784 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2788 /* mmput gets rid of the mappings and all user-space */
2789 extern void mmput(struct mm_struct *);
2790 /* same as above but performs the slow path from the async kontext. Can
2791 * be called from the atomic context as well
2793 extern void mmput_async(struct mm_struct *);
2795 /* Grab a reference to a task's mm, if it is not already going away */
2796 extern struct mm_struct *get_task_mm(struct task_struct *task);
2798 * Grab a reference to a task's mm, if it is not already going away
2799 * and ptrace_may_access with the mode parameter passed to it
2802 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2803 /* Remove the current tasks stale references to the old mm_struct */
2804 extern void mm_release(struct task_struct *, struct mm_struct *);
2806 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2807 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2808 struct task_struct *, unsigned long);
2810 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2811 struct task_struct *);
2813 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2814 * via pt_regs, so ignore the tls argument passed via C. */
2815 static inline int copy_thread_tls(
2816 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2817 struct task_struct *p, unsigned long tls)
2819 return copy_thread(clone_flags, sp, arg, p);
2822 extern void flush_thread(void);
2824 #ifdef CONFIG_HAVE_EXIT_THREAD
2825 extern void exit_thread(struct task_struct *tsk);
2827 static inline void exit_thread(struct task_struct *tsk)
2832 extern void exit_files(struct task_struct *);
2833 extern void __cleanup_sighand(struct sighand_struct *);
2835 extern void exit_itimers(struct signal_struct *);
2836 extern void flush_itimer_signals(void);
2838 extern void do_group_exit(int);
2840 extern int do_execve(struct filename *,
2841 const char __user * const __user *,
2842 const char __user * const __user *);
2843 extern int do_execveat(int, struct filename *,
2844 const char __user * const __user *,
2845 const char __user * const __user *,
2847 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2848 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2849 struct task_struct *fork_idle(int);
2850 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2852 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2853 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2855 __set_task_comm(tsk, from, false);
2857 extern char *get_task_comm(char *to, struct task_struct *tsk);
2860 void scheduler_ipi(void);
2861 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2863 static inline void scheduler_ipi(void) { }
2864 static inline unsigned long wait_task_inactive(struct task_struct *p,
2871 #define tasklist_empty() \
2872 list_empty(&init_task.tasks)
2874 #define next_task(p) \
2875 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2877 #define for_each_process(p) \
2878 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2880 extern bool current_is_single_threaded(void);
2883 * Careful: do_each_thread/while_each_thread is a double loop so
2884 * 'break' will not work as expected - use goto instead.
2886 #define do_each_thread(g, t) \
2887 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2889 #define while_each_thread(g, t) \
2890 while ((t = next_thread(t)) != g)
2892 #define __for_each_thread(signal, t) \
2893 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2895 #define for_each_thread(p, t) \
2896 __for_each_thread((p)->signal, t)
2898 /* Careful: this is a double loop, 'break' won't work as expected. */
2899 #define for_each_process_thread(p, t) \
2900 for_each_process(p) for_each_thread(p, t)
2902 static inline int get_nr_threads(struct task_struct *tsk)
2904 return tsk->signal->nr_threads;
2907 static inline bool thread_group_leader(struct task_struct *p)
2909 return p->exit_signal >= 0;
2912 /* Do to the insanities of de_thread it is possible for a process
2913 * to have the pid of the thread group leader without actually being
2914 * the thread group leader. For iteration through the pids in proc
2915 * all we care about is that we have a task with the appropriate
2916 * pid, we don't actually care if we have the right task.
2918 static inline bool has_group_leader_pid(struct task_struct *p)
2920 return task_pid(p) == p->signal->leader_pid;
2924 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2926 return p1->signal == p2->signal;
2929 static inline struct task_struct *next_thread(const struct task_struct *p)
2931 return list_entry_rcu(p->thread_group.next,
2932 struct task_struct, thread_group);
2935 static inline int thread_group_empty(struct task_struct *p)
2937 return list_empty(&p->thread_group);
2940 #define delay_group_leader(p) \
2941 (thread_group_leader(p) && !thread_group_empty(p))
2944 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2945 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2946 * pins the final release of task.io_context. Also protects ->cpuset and
2947 * ->cgroup.subsys[]. And ->vfork_done.
2949 * Nests both inside and outside of read_lock(&tasklist_lock).
2950 * It must not be nested with write_lock_irq(&tasklist_lock),
2951 * neither inside nor outside.
2953 static inline void task_lock(struct task_struct *p)
2955 spin_lock(&p->alloc_lock);
2958 static inline void task_unlock(struct task_struct *p)
2960 spin_unlock(&p->alloc_lock);
2963 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2964 unsigned long *flags);
2966 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2967 unsigned long *flags)
2969 struct sighand_struct *ret;
2971 ret = __lock_task_sighand(tsk, flags);
2972 (void)__cond_lock(&tsk->sighand->siglock, ret);
2976 static inline void unlock_task_sighand(struct task_struct *tsk,
2977 unsigned long *flags)
2979 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2983 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2984 * @tsk: task causing the changes
2986 * All operations which modify a threadgroup - a new thread joining the
2987 * group, death of a member thread (the assertion of PF_EXITING) and
2988 * exec(2) dethreading the process and replacing the leader - are wrapped
2989 * by threadgroup_change_{begin|end}(). This is to provide a place which
2990 * subsystems needing threadgroup stability can hook into for
2993 static inline void threadgroup_change_begin(struct task_struct *tsk)
2996 cgroup_threadgroup_change_begin(tsk);
3000 * threadgroup_change_end - mark the end of changes to a threadgroup
3001 * @tsk: task causing the changes
3003 * See threadgroup_change_begin().
3005 static inline void threadgroup_change_end(struct task_struct *tsk)
3007 cgroup_threadgroup_change_end(tsk);
3010 #ifdef CONFIG_THREAD_INFO_IN_TASK
3012 static inline struct thread_info *task_thread_info(struct task_struct *task)
3014 return &task->thread_info;
3018 * When accessing the stack of a non-current task that might exit, use
3019 * try_get_task_stack() instead. task_stack_page will return a pointer
3020 * that could get freed out from under you.
3022 static inline void *task_stack_page(const struct task_struct *task)
3027 #define setup_thread_stack(new,old) do { } while(0)
3029 static inline unsigned long *end_of_stack(const struct task_struct *task)
3034 #elif !defined(__HAVE_THREAD_FUNCTIONS)
3036 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
3037 #define task_stack_page(task) ((void *)(task)->stack)
3039 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
3041 *task_thread_info(p) = *task_thread_info(org);
3042 task_thread_info(p)->task = p;
3046 * Return the address of the last usable long on the stack.
3048 * When the stack grows down, this is just above the thread
3049 * info struct. Going any lower will corrupt the threadinfo.
3051 * When the stack grows up, this is the highest address.
3052 * Beyond that position, we corrupt data on the next page.
3054 static inline unsigned long *end_of_stack(struct task_struct *p)
3056 #ifdef CONFIG_STACK_GROWSUP
3057 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
3059 return (unsigned long *)(task_thread_info(p) + 1);
3065 static inline void *try_get_task_stack(struct task_struct *tsk)
3067 return task_stack_page(tsk);
3070 static inline void put_task_stack(struct task_struct *tsk) {}
3072 #define task_stack_end_corrupted(task) \
3073 (*(end_of_stack(task)) != STACK_END_MAGIC)
3075 static inline int object_is_on_stack(void *obj)
3077 void *stack = task_stack_page(current);
3079 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
3082 extern void thread_stack_cache_init(void);
3084 #ifdef CONFIG_DEBUG_STACK_USAGE
3085 static inline unsigned long stack_not_used(struct task_struct *p)
3087 unsigned long *n = end_of_stack(p);
3089 do { /* Skip over canary */
3093 return (unsigned long)n - (unsigned long)end_of_stack(p);
3096 extern void set_task_stack_end_magic(struct task_struct *tsk);
3098 /* set thread flags in other task's structures
3099 * - see asm/thread_info.h for TIF_xxxx flags available
3101 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
3103 set_ti_thread_flag(task_thread_info(tsk), flag);
3106 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3108 clear_ti_thread_flag(task_thread_info(tsk), flag);
3111 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
3113 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
3116 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3118 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
3121 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
3123 return test_ti_thread_flag(task_thread_info(tsk), flag);
3126 static inline void set_tsk_need_resched(struct task_struct *tsk)
3128 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3131 static inline void clear_tsk_need_resched(struct task_struct *tsk)
3133 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3136 static inline int test_tsk_need_resched(struct task_struct *tsk)
3138 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
3141 static inline int restart_syscall(void)
3143 set_tsk_thread_flag(current, TIF_SIGPENDING);
3144 return -ERESTARTNOINTR;
3147 static inline int signal_pending(struct task_struct *p)
3149 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
3152 static inline int __fatal_signal_pending(struct task_struct *p)
3154 return unlikely(sigismember(&p->pending.signal, SIGKILL));
3157 static inline int fatal_signal_pending(struct task_struct *p)
3159 return signal_pending(p) && __fatal_signal_pending(p);
3162 static inline int signal_pending_state(long state, struct task_struct *p)
3164 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
3166 if (!signal_pending(p))
3169 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
3173 * cond_resched() and cond_resched_lock(): latency reduction via
3174 * explicit rescheduling in places that are safe. The return
3175 * value indicates whether a reschedule was done in fact.
3176 * cond_resched_lock() will drop the spinlock before scheduling,
3177 * cond_resched_softirq() will enable bhs before scheduling.
3179 extern int _cond_resched(void);
3181 #define cond_resched() ({ \
3182 ___might_sleep(__FILE__, __LINE__, 0); \
3186 extern int __cond_resched_lock(spinlock_t *lock);
3188 #define cond_resched_lock(lock) ({ \
3189 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3190 __cond_resched_lock(lock); \
3193 extern int __cond_resched_softirq(void);
3195 #define cond_resched_softirq() ({ \
3196 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3197 __cond_resched_softirq(); \
3200 static inline void cond_resched_rcu(void)
3202 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3210 * Does a critical section need to be broken due to another
3211 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3212 * but a general need for low latency)
3214 static inline int spin_needbreak(spinlock_t *lock)
3216 #ifdef CONFIG_PREEMPT
3217 return spin_is_contended(lock);
3224 * Idle thread specific functions to determine the need_resched
3227 #ifdef TIF_POLLING_NRFLAG
3228 static inline int tsk_is_polling(struct task_struct *p)
3230 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3233 static inline void __current_set_polling(void)
3235 set_thread_flag(TIF_POLLING_NRFLAG);
3238 static inline bool __must_check current_set_polling_and_test(void)
3240 __current_set_polling();
3243 * Polling state must be visible before we test NEED_RESCHED,
3244 * paired by resched_curr()
3246 smp_mb__after_atomic();
3248 return unlikely(tif_need_resched());
3251 static inline void __current_clr_polling(void)
3253 clear_thread_flag(TIF_POLLING_NRFLAG);
3256 static inline bool __must_check current_clr_polling_and_test(void)
3258 __current_clr_polling();
3261 * Polling state must be visible before we test NEED_RESCHED,
3262 * paired by resched_curr()
3264 smp_mb__after_atomic();
3266 return unlikely(tif_need_resched());
3270 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3271 static inline void __current_set_polling(void) { }
3272 static inline void __current_clr_polling(void) { }
3274 static inline bool __must_check current_set_polling_and_test(void)
3276 return unlikely(tif_need_resched());
3278 static inline bool __must_check current_clr_polling_and_test(void)
3280 return unlikely(tif_need_resched());
3284 static inline void current_clr_polling(void)
3286 __current_clr_polling();
3289 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3290 * Once the bit is cleared, we'll get IPIs with every new
3291 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3294 smp_mb(); /* paired with resched_curr() */
3296 preempt_fold_need_resched();
3299 static __always_inline bool need_resched(void)
3301 return unlikely(tif_need_resched());
3305 * Thread group CPU time accounting.
3307 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3308 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3311 * Reevaluate whether the task has signals pending delivery.
3312 * Wake the task if so.
3313 * This is required every time the blocked sigset_t changes.
3314 * callers must hold sighand->siglock.
3316 extern void recalc_sigpending_and_wake(struct task_struct *t);
3317 extern void recalc_sigpending(void);
3319 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3321 static inline void signal_wake_up(struct task_struct *t, bool resume)
3323 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3325 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3327 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3331 * Wrappers for p->thread_info->cpu access. No-op on UP.
3335 static inline unsigned int task_cpu(const struct task_struct *p)
3337 #ifdef CONFIG_THREAD_INFO_IN_TASK
3340 return task_thread_info(p)->cpu;
3344 static inline int task_node(const struct task_struct *p)
3346 return cpu_to_node(task_cpu(p));
3349 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3353 static inline unsigned int task_cpu(const struct task_struct *p)
3358 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3362 #endif /* CONFIG_SMP */
3364 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3365 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3367 #ifdef CONFIG_CGROUP_SCHED
3368 extern struct task_group root_task_group;
3369 #endif /* CONFIG_CGROUP_SCHED */
3371 extern int task_can_switch_user(struct user_struct *up,
3372 struct task_struct *tsk);
3374 #ifdef CONFIG_TASK_XACCT
3375 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3377 tsk->ioac.rchar += amt;
3380 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3382 tsk->ioac.wchar += amt;
3385 static inline void inc_syscr(struct task_struct *tsk)
3390 static inline void inc_syscw(struct task_struct *tsk)
3395 static inline void inc_syscfs(struct task_struct *tsk)
3400 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3404 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3408 static inline void inc_syscr(struct task_struct *tsk)
3412 static inline void inc_syscw(struct task_struct *tsk)
3415 static inline void inc_syscfs(struct task_struct *tsk)
3420 #ifndef TASK_SIZE_OF
3421 #define TASK_SIZE_OF(tsk) TASK_SIZE
3425 extern void mm_update_next_owner(struct mm_struct *mm);
3427 static inline void mm_update_next_owner(struct mm_struct *mm)
3430 #endif /* CONFIG_MEMCG */
3432 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3435 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3438 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3441 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3444 static inline unsigned long rlimit(unsigned int limit)
3446 return task_rlimit(current, limit);
3449 static inline unsigned long rlimit_max(unsigned int limit)
3451 return task_rlimit_max(current, limit);
3454 #define SCHED_CPUFREQ_RT (1U << 0)
3455 #define SCHED_CPUFREQ_DL (1U << 1)
3456 #define SCHED_CPUFREQ_IOWAIT (1U << 2)
3458 #ifdef CONFIG_CPU_FREQ
3459 struct update_util_data {
3460 void (*func)(struct update_util_data *data, u64 time, unsigned int flags);
3463 void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
3464 void (*func)(struct update_util_data *data, u64 time,
3465 unsigned int flags));
3466 void cpufreq_remove_update_util_hook(int cpu);
3467 #endif /* CONFIG_CPU_FREQ */