4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
61 #include <linux/cgroup-defs.h>
63 #include <asm/processor.h>
65 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
68 * Extended scheduling parameters data structure.
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
91 * This is reflected by the actual fields of the sched_attr structure:
93 * @size size of the structure, for fwd/bwd compat.
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
117 /* SCHED_NORMAL, SCHED_BATCH */
120 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state;
130 struct robust_list_head;
133 struct perf_event_context;
138 #define VMACACHE_BITS 2
139 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
140 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
152 extern unsigned long avenrun[]; /* Load averages */
153 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
155 #define FSHIFT 11 /* nr of bits of precision */
156 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
157 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
158 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159 #define EXP_5 2014 /* 1/exp(5sec/5min) */
160 #define EXP_15 2037 /* 1/exp(5sec/15min) */
162 #define CALC_LOAD(load,exp,n) \
164 load += n*(FIXED_1-exp); \
167 extern unsigned long total_forks;
168 extern int nr_threads;
169 DECLARE_PER_CPU(unsigned long, process_counts);
170 extern int nr_processes(void);
171 extern unsigned long nr_running(void);
172 extern bool single_task_running(void);
173 extern unsigned long nr_iowait(void);
174 extern unsigned long nr_iowait_cpu(int cpu);
175 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
177 extern void calc_global_load(unsigned long ticks);
179 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
180 extern void update_cpu_load_nohz(void);
182 static inline void update_cpu_load_nohz(void) { }
185 extern unsigned long get_parent_ip(unsigned long addr);
187 extern void dump_cpu_task(int cpu);
192 #ifdef CONFIG_SCHED_DEBUG
193 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
194 extern void proc_sched_set_task(struct task_struct *p);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_NOLOAD 1024
222 #define TASK_STATE_MAX 2048
224 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
226 extern char ___assert_task_state[1 - 2*!!(
227 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
229 /* Convenience macros for the sake of set_task_state */
230 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
231 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
232 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
234 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
236 /* Convenience macros for the sake of wake_up */
237 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
238 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
240 /* get_task_state() */
241 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
242 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
243 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
245 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
246 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
247 #define task_is_stopped_or_traced(task) \
248 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
249 #define task_contributes_to_load(task) \
250 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
251 (task->flags & PF_FROZEN) == 0 && \
252 (task->state & TASK_NOLOAD) == 0)
254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256 #define __set_task_state(tsk, state_value) \
258 (tsk)->task_state_change = _THIS_IP_; \
259 (tsk)->state = (state_value); \
261 #define set_task_state(tsk, state_value) \
263 (tsk)->task_state_change = _THIS_IP_; \
264 smp_store_mb((tsk)->state, (state_value)); \
268 * set_current_state() includes a barrier so that the write of current->state
269 * is correctly serialised wrt the caller's subsequent test of whether to
272 * set_current_state(TASK_UNINTERRUPTIBLE);
273 * if (do_i_need_to_sleep())
276 * If the caller does not need such serialisation then use __set_current_state()
278 #define __set_current_state(state_value) \
280 current->task_state_change = _THIS_IP_; \
281 current->state = (state_value); \
283 #define set_current_state(state_value) \
285 current->task_state_change = _THIS_IP_; \
286 smp_store_mb(current->state, (state_value)); \
291 #define __set_task_state(tsk, state_value) \
292 do { (tsk)->state = (state_value); } while (0)
293 #define set_task_state(tsk, state_value) \
294 smp_store_mb((tsk)->state, (state_value))
297 * set_current_state() includes a barrier so that the write of current->state
298 * is correctly serialised wrt the caller's subsequent test of whether to
301 * set_current_state(TASK_UNINTERRUPTIBLE);
302 * if (do_i_need_to_sleep())
305 * If the caller does not need such serialisation then use __set_current_state()
307 #define __set_current_state(state_value) \
308 do { current->state = (state_value); } while (0)
309 #define set_current_state(state_value) \
310 smp_store_mb(current->state, (state_value))
314 /* Task command name length */
315 #define TASK_COMM_LEN 16
317 #include <linux/spinlock.h>
320 * This serializes "schedule()" and also protects
321 * the run-queue from deletions/modifications (but
322 * _adding_ to the beginning of the run-queue has
325 extern rwlock_t tasklist_lock;
326 extern spinlock_t mmlist_lock;
330 #ifdef CONFIG_PROVE_RCU
331 extern int lockdep_tasklist_lock_is_held(void);
332 #endif /* #ifdef CONFIG_PROVE_RCU */
334 extern void sched_init(void);
335 extern void sched_init_smp(void);
336 extern asmlinkage void schedule_tail(struct task_struct *prev);
337 extern void init_idle(struct task_struct *idle, int cpu);
338 extern void init_idle_bootup_task(struct task_struct *idle);
340 extern cpumask_var_t cpu_isolated_map;
342 extern int runqueue_is_locked(int cpu);
344 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
345 extern void nohz_balance_enter_idle(int cpu);
346 extern void set_cpu_sd_state_idle(void);
347 extern int get_nohz_timer_target(void);
349 static inline void nohz_balance_enter_idle(int cpu) { }
350 static inline void set_cpu_sd_state_idle(void) { }
354 * Only dump TASK_* tasks. (0 for all tasks)
356 extern void show_state_filter(unsigned long state_filter);
358 static inline void show_state(void)
360 show_state_filter(0);
363 extern void show_regs(struct pt_regs *);
366 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
367 * task), SP is the stack pointer of the first frame that should be shown in the back
368 * trace (or NULL if the entire call-chain of the task should be shown).
370 extern void show_stack(struct task_struct *task, unsigned long *sp);
372 extern void cpu_init (void);
373 extern void trap_init(void);
374 extern void update_process_times(int user);
375 extern void scheduler_tick(void);
377 extern void sched_show_task(struct task_struct *p);
379 #ifdef CONFIG_LOCKUP_DETECTOR
380 extern void touch_softlockup_watchdog(void);
381 extern void touch_softlockup_watchdog_sync(void);
382 extern void touch_all_softlockup_watchdogs(void);
383 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
385 size_t *lenp, loff_t *ppos);
386 extern unsigned int softlockup_panic;
387 extern unsigned int hardlockup_panic;
388 void lockup_detector_init(void);
390 static inline void touch_softlockup_watchdog(void)
393 static inline void touch_softlockup_watchdog_sync(void)
396 static inline void touch_all_softlockup_watchdogs(void)
399 static inline void lockup_detector_init(void)
404 #ifdef CONFIG_DETECT_HUNG_TASK
405 void reset_hung_task_detector(void);
407 static inline void reset_hung_task_detector(void)
412 /* Attach to any functions which should be ignored in wchan output. */
413 #define __sched __attribute__((__section__(".sched.text")))
415 /* Linker adds these: start and end of __sched functions */
416 extern char __sched_text_start[], __sched_text_end[];
418 /* Is this address in the __sched functions? */
419 extern int in_sched_functions(unsigned long addr);
421 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
422 extern signed long schedule_timeout(signed long timeout);
423 extern signed long schedule_timeout_interruptible(signed long timeout);
424 extern signed long schedule_timeout_killable(signed long timeout);
425 extern signed long schedule_timeout_uninterruptible(signed long timeout);
426 asmlinkage void schedule(void);
427 extern void schedule_preempt_disabled(void);
429 extern long io_schedule_timeout(long timeout);
431 static inline void io_schedule(void)
433 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
437 struct user_namespace;
440 extern void arch_pick_mmap_layout(struct mm_struct *mm);
442 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
443 unsigned long, unsigned long);
445 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
446 unsigned long len, unsigned long pgoff,
447 unsigned long flags);
449 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
452 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
453 #define SUID_DUMP_USER 1 /* Dump as user of process */
454 #define SUID_DUMP_ROOT 2 /* Dump as root */
458 /* for SUID_DUMP_* above */
459 #define MMF_DUMPABLE_BITS 2
460 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
462 extern void set_dumpable(struct mm_struct *mm, int value);
464 * This returns the actual value of the suid_dumpable flag. For things
465 * that are using this for checking for privilege transitions, it must
466 * test against SUID_DUMP_USER rather than treating it as a boolean
469 static inline int __get_dumpable(unsigned long mm_flags)
471 return mm_flags & MMF_DUMPABLE_MASK;
474 static inline int get_dumpable(struct mm_struct *mm)
476 return __get_dumpable(mm->flags);
479 /* coredump filter bits */
480 #define MMF_DUMP_ANON_PRIVATE 2
481 #define MMF_DUMP_ANON_SHARED 3
482 #define MMF_DUMP_MAPPED_PRIVATE 4
483 #define MMF_DUMP_MAPPED_SHARED 5
484 #define MMF_DUMP_ELF_HEADERS 6
485 #define MMF_DUMP_HUGETLB_PRIVATE 7
486 #define MMF_DUMP_HUGETLB_SHARED 8
487 #define MMF_DUMP_DAX_PRIVATE 9
488 #define MMF_DUMP_DAX_SHARED 10
490 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
491 #define MMF_DUMP_FILTER_BITS 9
492 #define MMF_DUMP_FILTER_MASK \
493 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
494 #define MMF_DUMP_FILTER_DEFAULT \
495 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
496 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
498 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
499 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
501 # define MMF_DUMP_MASK_DEFAULT_ELF 0
503 /* leave room for more dump flags */
504 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
505 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
506 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
508 #define MMF_HAS_UPROBES 19 /* has uprobes */
509 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
511 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
513 struct sighand_struct {
515 struct k_sigaction action[_NSIG];
517 wait_queue_head_t signalfd_wqh;
520 struct pacct_struct {
523 unsigned long ac_mem;
524 cputime_t ac_utime, ac_stime;
525 unsigned long ac_minflt, ac_majflt;
536 * struct prev_cputime - snaphsot of system and user cputime
537 * @utime: time spent in user mode
538 * @stime: time spent in system mode
539 * @lock: protects the above two fields
541 * Stores previous user/system time values such that we can guarantee
544 struct prev_cputime {
545 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
552 static inline void prev_cputime_init(struct prev_cputime *prev)
554 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
555 prev->utime = prev->stime = 0;
556 raw_spin_lock_init(&prev->lock);
561 * struct task_cputime - collected CPU time counts
562 * @utime: time spent in user mode, in &cputime_t units
563 * @stime: time spent in kernel mode, in &cputime_t units
564 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
566 * This structure groups together three kinds of CPU time that are tracked for
567 * threads and thread groups. Most things considering CPU time want to group
568 * these counts together and treat all three of them in parallel.
570 struct task_cputime {
573 unsigned long long sum_exec_runtime;
576 /* Alternate field names when used to cache expirations. */
577 #define virt_exp utime
578 #define prof_exp stime
579 #define sched_exp sum_exec_runtime
581 #define INIT_CPUTIME \
582 (struct task_cputime) { \
585 .sum_exec_runtime = 0, \
589 * This is the atomic variant of task_cputime, which can be used for
590 * storing and updating task_cputime statistics without locking.
592 struct task_cputime_atomic {
595 atomic64_t sum_exec_runtime;
598 #define INIT_CPUTIME_ATOMIC \
599 (struct task_cputime_atomic) { \
600 .utime = ATOMIC64_INIT(0), \
601 .stime = ATOMIC64_INIT(0), \
602 .sum_exec_runtime = ATOMIC64_INIT(0), \
605 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
608 * Disable preemption until the scheduler is running -- use an unconditional
609 * value so that it also works on !PREEMPT_COUNT kernels.
611 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
613 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
616 * Initial preempt_count value; reflects the preempt_count schedule invariant
617 * which states that during context switches:
619 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
621 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
622 * Note: See finish_task_switch().
624 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
627 * struct thread_group_cputimer - thread group interval timer counts
628 * @cputime_atomic: atomic thread group interval timers.
629 * @running: true when there are timers running and
630 * @cputime_atomic receives updates.
631 * @checking_timer: true when a thread in the group is in the
632 * process of checking for thread group timers.
634 * This structure contains the version of task_cputime, above, that is
635 * used for thread group CPU timer calculations.
637 struct thread_group_cputimer {
638 struct task_cputime_atomic cputime_atomic;
643 #include <linux/rwsem.h>
647 * NOTE! "signal_struct" does not have its own
648 * locking, because a shared signal_struct always
649 * implies a shared sighand_struct, so locking
650 * sighand_struct is always a proper superset of
651 * the locking of signal_struct.
653 struct signal_struct {
657 struct list_head thread_head;
659 wait_queue_head_t wait_chldexit; /* for wait4() */
661 /* current thread group signal load-balancing target: */
662 struct task_struct *curr_target;
664 /* shared signal handling: */
665 struct sigpending shared_pending;
667 /* thread group exit support */
670 * - notify group_exit_task when ->count is equal to notify_count
671 * - everyone except group_exit_task is stopped during signal delivery
672 * of fatal signals, group_exit_task processes the signal.
675 struct task_struct *group_exit_task;
677 /* thread group stop support, overloads group_exit_code too */
678 int group_stop_count;
679 unsigned int flags; /* see SIGNAL_* flags below */
682 * PR_SET_CHILD_SUBREAPER marks a process, like a service
683 * manager, to re-parent orphan (double-forking) child processes
684 * to this process instead of 'init'. The service manager is
685 * able to receive SIGCHLD signals and is able to investigate
686 * the process until it calls wait(). All children of this
687 * process will inherit a flag if they should look for a
688 * child_subreaper process at exit.
690 unsigned int is_child_subreaper:1;
691 unsigned int has_child_subreaper:1;
693 /* POSIX.1b Interval Timers */
695 struct list_head posix_timers;
697 /* ITIMER_REAL timer for the process */
698 struct hrtimer real_timer;
699 struct pid *leader_pid;
700 ktime_t it_real_incr;
703 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
704 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
705 * values are defined to 0 and 1 respectively
707 struct cpu_itimer it[2];
710 * Thread group totals for process CPU timers.
711 * See thread_group_cputimer(), et al, for details.
713 struct thread_group_cputimer cputimer;
715 /* Earliest-expiration cache. */
716 struct task_cputime cputime_expires;
718 struct list_head cpu_timers[3];
720 struct pid *tty_old_pgrp;
722 /* boolean value for session group leader */
725 struct tty_struct *tty; /* NULL if no tty */
727 #ifdef CONFIG_SCHED_AUTOGROUP
728 struct autogroup *autogroup;
731 * Cumulative resource counters for dead threads in the group,
732 * and for reaped dead child processes forked by this group.
733 * Live threads maintain their own counters and add to these
734 * in __exit_signal, except for the group leader.
736 seqlock_t stats_lock;
737 cputime_t utime, stime, cutime, cstime;
740 struct prev_cputime prev_cputime;
741 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
742 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
743 unsigned long inblock, oublock, cinblock, coublock;
744 unsigned long maxrss, cmaxrss;
745 struct task_io_accounting ioac;
748 * Cumulative ns of schedule CPU time fo dead threads in the
749 * group, not including a zombie group leader, (This only differs
750 * from jiffies_to_ns(utime + stime) if sched_clock uses something
751 * other than jiffies.)
753 unsigned long long sum_sched_runtime;
756 * We don't bother to synchronize most readers of this at all,
757 * because there is no reader checking a limit that actually needs
758 * to get both rlim_cur and rlim_max atomically, and either one
759 * alone is a single word that can safely be read normally.
760 * getrlimit/setrlimit use task_lock(current->group_leader) to
761 * protect this instead of the siglock, because they really
762 * have no need to disable irqs.
764 struct rlimit rlim[RLIM_NLIMITS];
766 #ifdef CONFIG_BSD_PROCESS_ACCT
767 struct pacct_struct pacct; /* per-process accounting information */
769 #ifdef CONFIG_TASKSTATS
770 struct taskstats *stats;
774 unsigned audit_tty_log_passwd;
775 struct tty_audit_buf *tty_audit_buf;
778 oom_flags_t oom_flags;
779 short oom_score_adj; /* OOM kill score adjustment */
780 short oom_score_adj_min; /* OOM kill score adjustment min value.
781 * Only settable by CAP_SYS_RESOURCE. */
783 struct mutex cred_guard_mutex; /* guard against foreign influences on
784 * credential calculations
785 * (notably. ptrace) */
789 * Bits in flags field of signal_struct.
791 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
792 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
793 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
794 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
796 * Pending notifications to parent.
798 #define SIGNAL_CLD_STOPPED 0x00000010
799 #define SIGNAL_CLD_CONTINUED 0x00000020
800 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
802 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
804 /* If true, all threads except ->group_exit_task have pending SIGKILL */
805 static inline int signal_group_exit(const struct signal_struct *sig)
807 return (sig->flags & SIGNAL_GROUP_EXIT) ||
808 (sig->group_exit_task != NULL);
812 * Some day this will be a full-fledged user tracking system..
815 atomic_t __count; /* reference count */
816 atomic_t processes; /* How many processes does this user have? */
817 atomic_t sigpending; /* How many pending signals does this user have? */
818 #ifdef CONFIG_INOTIFY_USER
819 atomic_t inotify_watches; /* How many inotify watches does this user have? */
820 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
822 #ifdef CONFIG_FANOTIFY
823 atomic_t fanotify_listeners;
826 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
828 #ifdef CONFIG_POSIX_MQUEUE
829 /* protected by mq_lock */
830 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
832 unsigned long locked_shm; /* How many pages of mlocked shm ? */
833 unsigned long unix_inflight; /* How many files in flight in unix sockets */
834 atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */
837 struct key *uid_keyring; /* UID specific keyring */
838 struct key *session_keyring; /* UID's default session keyring */
841 /* Hash table maintenance information */
842 struct hlist_node uidhash_node;
845 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
846 atomic_long_t locked_vm;
850 extern int uids_sysfs_init(void);
852 extern struct user_struct *find_user(kuid_t);
854 extern struct user_struct root_user;
855 #define INIT_USER (&root_user)
858 struct backing_dev_info;
859 struct reclaim_state;
861 #ifdef CONFIG_SCHED_INFO
863 /* cumulative counters */
864 unsigned long pcount; /* # of times run on this cpu */
865 unsigned long long run_delay; /* time spent waiting on a runqueue */
868 unsigned long long last_arrival,/* when we last ran on a cpu */
869 last_queued; /* when we were last queued to run */
871 #endif /* CONFIG_SCHED_INFO */
873 #ifdef CONFIG_TASK_DELAY_ACCT
874 struct task_delay_info {
876 unsigned int flags; /* Private per-task flags */
878 /* For each stat XXX, add following, aligned appropriately
880 * struct timespec XXX_start, XXX_end;
884 * Atomicity of updates to XXX_delay, XXX_count protected by
885 * single lock above (split into XXX_lock if contention is an issue).
889 * XXX_count is incremented on every XXX operation, the delay
890 * associated with the operation is added to XXX_delay.
891 * XXX_delay contains the accumulated delay time in nanoseconds.
893 u64 blkio_start; /* Shared by blkio, swapin */
894 u64 blkio_delay; /* wait for sync block io completion */
895 u64 swapin_delay; /* wait for swapin block io completion */
896 u32 blkio_count; /* total count of the number of sync block */
897 /* io operations performed */
898 u32 swapin_count; /* total count of the number of swapin block */
899 /* io operations performed */
902 u64 freepages_delay; /* wait for memory reclaim */
903 u32 freepages_count; /* total count of memory reclaim */
905 #endif /* CONFIG_TASK_DELAY_ACCT */
907 static inline int sched_info_on(void)
909 #ifdef CONFIG_SCHEDSTATS
911 #elif defined(CONFIG_TASK_DELAY_ACCT)
912 extern int delayacct_on;
927 * Increase resolution of cpu_capacity calculations
929 #define SCHED_CAPACITY_SHIFT 10
930 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
933 * Wake-queues are lists of tasks with a pending wakeup, whose
934 * callers have already marked the task as woken internally,
935 * and can thus carry on. A common use case is being able to
936 * do the wakeups once the corresponding user lock as been
939 * We hold reference to each task in the list across the wakeup,
940 * thus guaranteeing that the memory is still valid by the time
941 * the actual wakeups are performed in wake_up_q().
943 * One per task suffices, because there's never a need for a task to be
944 * in two wake queues simultaneously; it is forbidden to abandon a task
945 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
946 * already in a wake queue, the wakeup will happen soon and the second
947 * waker can just skip it.
949 * The WAKE_Q macro declares and initializes the list head.
950 * wake_up_q() does NOT reinitialize the list; it's expected to be
951 * called near the end of a function, where the fact that the queue is
952 * not used again will be easy to see by inspection.
954 * Note that this can cause spurious wakeups. schedule() callers
955 * must ensure the call is done inside a loop, confirming that the
956 * wakeup condition has in fact occurred.
959 struct wake_q_node *next;
963 struct wake_q_node *first;
964 struct wake_q_node **lastp;
967 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
969 #define WAKE_Q(name) \
970 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
972 extern void wake_q_add(struct wake_q_head *head,
973 struct task_struct *task);
974 extern void wake_up_q(struct wake_q_head *head);
977 * sched-domains (multiprocessor balancing) declarations:
980 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
981 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
982 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
983 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
984 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
985 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
986 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
987 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
988 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
989 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
990 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
991 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
992 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
993 #define SD_NUMA 0x4000 /* cross-node balancing */
995 #ifdef CONFIG_SCHED_SMT
996 static inline int cpu_smt_flags(void)
998 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1002 #ifdef CONFIG_SCHED_MC
1003 static inline int cpu_core_flags(void)
1005 return SD_SHARE_PKG_RESOURCES;
1010 static inline int cpu_numa_flags(void)
1016 struct sched_domain_attr {
1017 int relax_domain_level;
1020 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1021 .relax_domain_level = -1, \
1024 extern int sched_domain_level_max;
1028 struct sched_domain {
1029 /* These fields must be setup */
1030 struct sched_domain *parent; /* top domain must be null terminated */
1031 struct sched_domain *child; /* bottom domain must be null terminated */
1032 struct sched_group *groups; /* the balancing groups of the domain */
1033 unsigned long min_interval; /* Minimum balance interval ms */
1034 unsigned long max_interval; /* Maximum balance interval ms */
1035 unsigned int busy_factor; /* less balancing by factor if busy */
1036 unsigned int imbalance_pct; /* No balance until over watermark */
1037 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
1038 unsigned int busy_idx;
1039 unsigned int idle_idx;
1040 unsigned int newidle_idx;
1041 unsigned int wake_idx;
1042 unsigned int forkexec_idx;
1043 unsigned int smt_gain;
1045 int nohz_idle; /* NOHZ IDLE status */
1046 int flags; /* See SD_* */
1049 /* Runtime fields. */
1050 unsigned long last_balance; /* init to jiffies. units in jiffies */
1051 unsigned int balance_interval; /* initialise to 1. units in ms. */
1052 unsigned int nr_balance_failed; /* initialise to 0 */
1054 /* idle_balance() stats */
1055 u64 max_newidle_lb_cost;
1056 unsigned long next_decay_max_lb_cost;
1058 #ifdef CONFIG_SCHEDSTATS
1059 /* load_balance() stats */
1060 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1061 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1062 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1063 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1064 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1065 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1066 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1067 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1069 /* Active load balancing */
1070 unsigned int alb_count;
1071 unsigned int alb_failed;
1072 unsigned int alb_pushed;
1074 /* SD_BALANCE_EXEC stats */
1075 unsigned int sbe_count;
1076 unsigned int sbe_balanced;
1077 unsigned int sbe_pushed;
1079 /* SD_BALANCE_FORK stats */
1080 unsigned int sbf_count;
1081 unsigned int sbf_balanced;
1082 unsigned int sbf_pushed;
1084 /* try_to_wake_up() stats */
1085 unsigned int ttwu_wake_remote;
1086 unsigned int ttwu_move_affine;
1087 unsigned int ttwu_move_balance;
1089 #ifdef CONFIG_SCHED_DEBUG
1093 void *private; /* used during construction */
1094 struct rcu_head rcu; /* used during destruction */
1097 unsigned int span_weight;
1099 * Span of all CPUs in this domain.
1101 * NOTE: this field is variable length. (Allocated dynamically
1102 * by attaching extra space to the end of the structure,
1103 * depending on how many CPUs the kernel has booted up with)
1105 unsigned long span[0];
1108 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1110 return to_cpumask(sd->span);
1113 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1114 struct sched_domain_attr *dattr_new);
1116 /* Allocate an array of sched domains, for partition_sched_domains(). */
1117 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1118 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1120 bool cpus_share_cache(int this_cpu, int that_cpu);
1122 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1123 typedef int (*sched_domain_flags_f)(void);
1125 #define SDTL_OVERLAP 0x01
1128 struct sched_domain **__percpu sd;
1129 struct sched_group **__percpu sg;
1130 struct sched_group_capacity **__percpu sgc;
1133 struct sched_domain_topology_level {
1134 sched_domain_mask_f mask;
1135 sched_domain_flags_f sd_flags;
1138 struct sd_data data;
1139 #ifdef CONFIG_SCHED_DEBUG
1144 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1145 extern void wake_up_if_idle(int cpu);
1147 #ifdef CONFIG_SCHED_DEBUG
1148 # define SD_INIT_NAME(type) .name = #type
1150 # define SD_INIT_NAME(type)
1153 #else /* CONFIG_SMP */
1155 struct sched_domain_attr;
1158 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1159 struct sched_domain_attr *dattr_new)
1163 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1168 #endif /* !CONFIG_SMP */
1171 struct io_context; /* See blkdev.h */
1174 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1175 extern void prefetch_stack(struct task_struct *t);
1177 static inline void prefetch_stack(struct task_struct *t) { }
1180 struct audit_context; /* See audit.c */
1182 struct pipe_inode_info;
1183 struct uts_namespace;
1185 struct load_weight {
1186 unsigned long weight;
1191 * The load_avg/util_avg accumulates an infinite geometric series.
1192 * 1) load_avg factors frequency scaling into the amount of time that a
1193 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1194 * aggregated such weights of all runnable and blocked sched_entities.
1195 * 2) util_avg factors frequency and cpu scaling into the amount of time
1196 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1197 * For cfs_rq, it is the aggregated such times of all runnable and
1198 * blocked sched_entities.
1199 * The 64 bit load_sum can:
1200 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1201 * the highest weight (=88761) always runnable, we should not overflow
1202 * 2) for entity, support any load.weight always runnable
1205 u64 last_update_time, load_sum;
1206 u32 util_sum, period_contrib;
1207 unsigned long load_avg, util_avg;
1210 #ifdef CONFIG_SCHEDSTATS
1211 struct sched_statistics {
1221 s64 sum_sleep_runtime;
1228 u64 nr_migrations_cold;
1229 u64 nr_failed_migrations_affine;
1230 u64 nr_failed_migrations_running;
1231 u64 nr_failed_migrations_hot;
1232 u64 nr_forced_migrations;
1235 u64 nr_wakeups_sync;
1236 u64 nr_wakeups_migrate;
1237 u64 nr_wakeups_local;
1238 u64 nr_wakeups_remote;
1239 u64 nr_wakeups_affine;
1240 u64 nr_wakeups_affine_attempts;
1241 u64 nr_wakeups_passive;
1242 u64 nr_wakeups_idle;
1246 struct sched_entity {
1247 struct load_weight load; /* for load-balancing */
1248 struct rb_node run_node;
1249 struct list_head group_node;
1253 u64 sum_exec_runtime;
1255 u64 prev_sum_exec_runtime;
1259 #ifdef CONFIG_SCHEDSTATS
1260 struct sched_statistics statistics;
1263 #ifdef CONFIG_FAIR_GROUP_SCHED
1265 struct sched_entity *parent;
1266 /* rq on which this entity is (to be) queued: */
1267 struct cfs_rq *cfs_rq;
1268 /* rq "owned" by this entity/group: */
1269 struct cfs_rq *my_q;
1273 /* Per entity load average tracking */
1274 struct sched_avg avg;
1278 struct sched_rt_entity {
1279 struct list_head run_list;
1280 unsigned long timeout;
1281 unsigned long watchdog_stamp;
1282 unsigned int time_slice;
1284 struct sched_rt_entity *back;
1285 #ifdef CONFIG_RT_GROUP_SCHED
1286 struct sched_rt_entity *parent;
1287 /* rq on which this entity is (to be) queued: */
1288 struct rt_rq *rt_rq;
1289 /* rq "owned" by this entity/group: */
1294 struct sched_dl_entity {
1295 struct rb_node rb_node;
1298 * Original scheduling parameters. Copied here from sched_attr
1299 * during sched_setattr(), they will remain the same until
1300 * the next sched_setattr().
1302 u64 dl_runtime; /* maximum runtime for each instance */
1303 u64 dl_deadline; /* relative deadline of each instance */
1304 u64 dl_period; /* separation of two instances (period) */
1305 u64 dl_bw; /* dl_runtime / dl_deadline */
1308 * Actual scheduling parameters. Initialized with the values above,
1309 * they are continously updated during task execution. Note that
1310 * the remaining runtime could be < 0 in case we are in overrun.
1312 s64 runtime; /* remaining runtime for this instance */
1313 u64 deadline; /* absolute deadline for this instance */
1314 unsigned int flags; /* specifying the scheduler behaviour */
1319 * @dl_throttled tells if we exhausted the runtime. If so, the
1320 * task has to wait for a replenishment to be performed at the
1321 * next firing of dl_timer.
1323 * @dl_new tells if a new instance arrived. If so we must
1324 * start executing it with full runtime and reset its absolute
1327 * @dl_boosted tells if we are boosted due to DI. If so we are
1328 * outside bandwidth enforcement mechanism (but only until we
1329 * exit the critical section);
1331 * @dl_yielded tells if task gave up the cpu before consuming
1332 * all its available runtime during the last job.
1334 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1337 * Bandwidth enforcement timer. Each -deadline task has its
1338 * own bandwidth to be enforced, thus we need one timer per task.
1340 struct hrtimer dl_timer;
1348 u8 pad; /* Otherwise the compiler can store garbage here. */
1350 u32 s; /* Set of bits. */
1354 enum perf_event_task_context {
1355 perf_invalid_context = -1,
1356 perf_hw_context = 0,
1358 perf_nr_task_contexts,
1361 /* Track pages that require TLB flushes */
1362 struct tlbflush_unmap_batch {
1364 * Each bit set is a CPU that potentially has a TLB entry for one of
1365 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1367 struct cpumask cpumask;
1369 /* True if any bit in cpumask is set */
1370 bool flush_required;
1373 * If true then the PTE was dirty when unmapped. The entry must be
1374 * flushed before IO is initiated or a stale TLB entry potentially
1375 * allows an update without redirtying the page.
1380 struct task_struct {
1381 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1384 unsigned int flags; /* per process flags, defined below */
1385 unsigned int ptrace;
1388 struct llist_node wake_entry;
1390 unsigned int wakee_flips;
1391 unsigned long wakee_flip_decay_ts;
1392 struct task_struct *last_wakee;
1398 int prio, static_prio, normal_prio;
1399 unsigned int rt_priority;
1400 const struct sched_class *sched_class;
1401 struct sched_entity se;
1402 struct sched_rt_entity rt;
1403 #ifdef CONFIG_CGROUP_SCHED
1404 struct task_group *sched_task_group;
1406 struct sched_dl_entity dl;
1408 #ifdef CONFIG_PREEMPT_NOTIFIERS
1409 /* list of struct preempt_notifier: */
1410 struct hlist_head preempt_notifiers;
1413 #ifdef CONFIG_BLK_DEV_IO_TRACE
1414 unsigned int btrace_seq;
1417 unsigned int policy;
1418 int nr_cpus_allowed;
1419 cpumask_t cpus_allowed;
1421 #ifdef CONFIG_PREEMPT_RCU
1422 int rcu_read_lock_nesting;
1423 union rcu_special rcu_read_unlock_special;
1424 struct list_head rcu_node_entry;
1425 struct rcu_node *rcu_blocked_node;
1426 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1427 #ifdef CONFIG_TASKS_RCU
1428 unsigned long rcu_tasks_nvcsw;
1429 bool rcu_tasks_holdout;
1430 struct list_head rcu_tasks_holdout_list;
1431 int rcu_tasks_idle_cpu;
1432 #endif /* #ifdef CONFIG_TASKS_RCU */
1434 #ifdef CONFIG_SCHED_INFO
1435 struct sched_info sched_info;
1438 struct list_head tasks;
1440 struct plist_node pushable_tasks;
1441 struct rb_node pushable_dl_tasks;
1444 struct mm_struct *mm, *active_mm;
1445 /* per-thread vma caching */
1446 u32 vmacache_seqnum;
1447 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1448 #if defined(SPLIT_RSS_COUNTING)
1449 struct task_rss_stat rss_stat;
1453 int exit_code, exit_signal;
1454 int pdeath_signal; /* The signal sent when the parent dies */
1455 unsigned long jobctl; /* JOBCTL_*, siglock protected */
1457 /* Used for emulating ABI behavior of previous Linux versions */
1458 unsigned int personality;
1460 /* scheduler bits, serialized by scheduler locks */
1461 unsigned sched_reset_on_fork:1;
1462 unsigned sched_contributes_to_load:1;
1463 unsigned sched_migrated:1;
1464 unsigned :0; /* force alignment to the next boundary */
1466 /* unserialized, strictly 'current' */
1467 unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1468 unsigned in_iowait:1;
1470 unsigned memcg_may_oom:1;
1472 #ifdef CONFIG_MEMCG_KMEM
1473 unsigned memcg_kmem_skip_account:1;
1475 #ifdef CONFIG_COMPAT_BRK
1476 unsigned brk_randomized:1;
1479 unsigned long atomic_flags; /* Flags needing atomic access. */
1481 struct restart_block restart_block;
1486 #ifdef CONFIG_CC_STACKPROTECTOR
1487 /* Canary value for the -fstack-protector gcc feature */
1488 unsigned long stack_canary;
1491 * pointers to (original) parent process, youngest child, younger sibling,
1492 * older sibling, respectively. (p->father can be replaced with
1493 * p->real_parent->pid)
1495 struct task_struct __rcu *real_parent; /* real parent process */
1496 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1498 * children/sibling forms the list of my natural children
1500 struct list_head children; /* list of my children */
1501 struct list_head sibling; /* linkage in my parent's children list */
1502 struct task_struct *group_leader; /* threadgroup leader */
1505 * ptraced is the list of tasks this task is using ptrace on.
1506 * This includes both natural children and PTRACE_ATTACH targets.
1507 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1509 struct list_head ptraced;
1510 struct list_head ptrace_entry;
1512 /* PID/PID hash table linkage. */
1513 struct pid_link pids[PIDTYPE_MAX];
1514 struct list_head thread_group;
1515 struct list_head thread_node;
1517 struct completion *vfork_done; /* for vfork() */
1518 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1519 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1521 cputime_t utime, stime, utimescaled, stimescaled;
1523 struct prev_cputime prev_cputime;
1524 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1525 seqlock_t vtime_seqlock;
1526 unsigned long long vtime_snap;
1531 } vtime_snap_whence;
1533 unsigned long nvcsw, nivcsw; /* context switch counts */
1534 u64 start_time; /* monotonic time in nsec */
1535 u64 real_start_time; /* boot based time in nsec */
1536 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1537 unsigned long min_flt, maj_flt;
1539 struct task_cputime cputime_expires;
1540 struct list_head cpu_timers[3];
1542 /* process credentials */
1543 const struct cred __rcu *real_cred; /* objective and real subjective task
1544 * credentials (COW) */
1545 const struct cred __rcu *cred; /* effective (overridable) subjective task
1546 * credentials (COW) */
1547 char comm[TASK_COMM_LEN]; /* executable name excluding path
1548 - access with [gs]et_task_comm (which lock
1549 it with task_lock())
1550 - initialized normally by setup_new_exec */
1551 /* file system info */
1552 struct nameidata *nameidata;
1553 #ifdef CONFIG_SYSVIPC
1555 struct sysv_sem sysvsem;
1556 struct sysv_shm sysvshm;
1558 #ifdef CONFIG_DETECT_HUNG_TASK
1559 /* hung task detection */
1560 unsigned long last_switch_count;
1562 /* filesystem information */
1563 struct fs_struct *fs;
1564 /* open file information */
1565 struct files_struct *files;
1567 struct nsproxy *nsproxy;
1568 /* signal handlers */
1569 struct signal_struct *signal;
1570 struct sighand_struct *sighand;
1572 sigset_t blocked, real_blocked;
1573 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1574 struct sigpending pending;
1576 unsigned long sas_ss_sp;
1579 struct callback_head *task_works;
1581 struct audit_context *audit_context;
1582 #ifdef CONFIG_AUDITSYSCALL
1584 unsigned int sessionid;
1586 struct seccomp seccomp;
1588 /* Thread group tracking */
1591 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1593 spinlock_t alloc_lock;
1595 /* Protection of the PI data structures: */
1596 raw_spinlock_t pi_lock;
1598 struct wake_q_node wake_q;
1600 #ifdef CONFIG_RT_MUTEXES
1601 /* PI waiters blocked on a rt_mutex held by this task */
1602 struct rb_root pi_waiters;
1603 struct rb_node *pi_waiters_leftmost;
1604 /* Deadlock detection and priority inheritance handling */
1605 struct rt_mutex_waiter *pi_blocked_on;
1608 #ifdef CONFIG_DEBUG_MUTEXES
1609 /* mutex deadlock detection */
1610 struct mutex_waiter *blocked_on;
1612 #ifdef CONFIG_TRACE_IRQFLAGS
1613 unsigned int irq_events;
1614 unsigned long hardirq_enable_ip;
1615 unsigned long hardirq_disable_ip;
1616 unsigned int hardirq_enable_event;
1617 unsigned int hardirq_disable_event;
1618 int hardirqs_enabled;
1619 int hardirq_context;
1620 unsigned long softirq_disable_ip;
1621 unsigned long softirq_enable_ip;
1622 unsigned int softirq_disable_event;
1623 unsigned int softirq_enable_event;
1624 int softirqs_enabled;
1625 int softirq_context;
1627 #ifdef CONFIG_LOCKDEP
1628 # define MAX_LOCK_DEPTH 48UL
1631 unsigned int lockdep_recursion;
1632 struct held_lock held_locks[MAX_LOCK_DEPTH];
1633 gfp_t lockdep_reclaim_gfp;
1636 /* journalling filesystem info */
1639 /* stacked block device info */
1640 struct bio_list *bio_list;
1643 /* stack plugging */
1644 struct blk_plug *plug;
1648 struct reclaim_state *reclaim_state;
1650 struct backing_dev_info *backing_dev_info;
1652 struct io_context *io_context;
1654 unsigned long ptrace_message;
1655 siginfo_t *last_siginfo; /* For ptrace use. */
1656 struct task_io_accounting ioac;
1657 #if defined(CONFIG_TASK_XACCT)
1658 u64 acct_rss_mem1; /* accumulated rss usage */
1659 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1660 cputime_t acct_timexpd; /* stime + utime since last update */
1662 #ifdef CONFIG_CPUSETS
1663 nodemask_t mems_allowed; /* Protected by alloc_lock */
1664 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1665 int cpuset_mem_spread_rotor;
1666 int cpuset_slab_spread_rotor;
1668 #ifdef CONFIG_CGROUPS
1669 /* Control Group info protected by css_set_lock */
1670 struct css_set __rcu *cgroups;
1671 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1672 struct list_head cg_list;
1675 struct robust_list_head __user *robust_list;
1676 #ifdef CONFIG_COMPAT
1677 struct compat_robust_list_head __user *compat_robust_list;
1679 struct list_head pi_state_list;
1680 struct futex_pi_state *pi_state_cache;
1682 #ifdef CONFIG_PERF_EVENTS
1683 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1684 struct mutex perf_event_mutex;
1685 struct list_head perf_event_list;
1687 #ifdef CONFIG_DEBUG_PREEMPT
1688 unsigned long preempt_disable_ip;
1691 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1693 short pref_node_fork;
1695 #ifdef CONFIG_NUMA_BALANCING
1697 unsigned int numa_scan_period;
1698 unsigned int numa_scan_period_max;
1699 int numa_preferred_nid;
1700 unsigned long numa_migrate_retry;
1701 u64 node_stamp; /* migration stamp */
1702 u64 last_task_numa_placement;
1703 u64 last_sum_exec_runtime;
1704 struct callback_head numa_work;
1706 struct list_head numa_entry;
1707 struct numa_group *numa_group;
1710 * numa_faults is an array split into four regions:
1711 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1712 * in this precise order.
1714 * faults_memory: Exponential decaying average of faults on a per-node
1715 * basis. Scheduling placement decisions are made based on these
1716 * counts. The values remain static for the duration of a PTE scan.
1717 * faults_cpu: Track the nodes the process was running on when a NUMA
1718 * hinting fault was incurred.
1719 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1720 * during the current scan window. When the scan completes, the counts
1721 * in faults_memory and faults_cpu decay and these values are copied.
1723 unsigned long *numa_faults;
1724 unsigned long total_numa_faults;
1727 * numa_faults_locality tracks if faults recorded during the last
1728 * scan window were remote/local or failed to migrate. The task scan
1729 * period is adapted based on the locality of the faults with different
1730 * weights depending on whether they were shared or private faults
1732 unsigned long numa_faults_locality[3];
1734 unsigned long numa_pages_migrated;
1735 #endif /* CONFIG_NUMA_BALANCING */
1737 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1738 struct tlbflush_unmap_batch tlb_ubc;
1741 struct rcu_head rcu;
1744 * cache last used pipe for splice
1746 struct pipe_inode_info *splice_pipe;
1748 struct page_frag task_frag;
1750 #ifdef CONFIG_TASK_DELAY_ACCT
1751 struct task_delay_info *delays;
1753 #ifdef CONFIG_FAULT_INJECTION
1757 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1758 * balance_dirty_pages() for some dirty throttling pause
1761 int nr_dirtied_pause;
1762 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1764 #ifdef CONFIG_LATENCYTOP
1765 int latency_record_count;
1766 struct latency_record latency_record[LT_SAVECOUNT];
1769 * time slack values; these are used to round up poll() and
1770 * select() etc timeout values. These are in nanoseconds.
1772 unsigned long timer_slack_ns;
1773 unsigned long default_timer_slack_ns;
1776 unsigned int kasan_depth;
1778 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1779 /* Index of current stored address in ret_stack */
1781 /* Stack of return addresses for return function tracing */
1782 struct ftrace_ret_stack *ret_stack;
1783 /* time stamp for last schedule */
1784 unsigned long long ftrace_timestamp;
1786 * Number of functions that haven't been traced
1787 * because of depth overrun.
1789 atomic_t trace_overrun;
1790 /* Pause for the tracing */
1791 atomic_t tracing_graph_pause;
1793 #ifdef CONFIG_TRACING
1794 /* state flags for use by tracers */
1795 unsigned long trace;
1796 /* bitmask and counter of trace recursion */
1797 unsigned long trace_recursion;
1798 #endif /* CONFIG_TRACING */
1800 struct mem_cgroup *memcg_in_oom;
1801 gfp_t memcg_oom_gfp_mask;
1802 int memcg_oom_order;
1804 /* number of pages to reclaim on returning to userland */
1805 unsigned int memcg_nr_pages_over_high;
1807 #ifdef CONFIG_UPROBES
1808 struct uprobe_task *utask;
1810 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1811 unsigned int sequential_io;
1812 unsigned int sequential_io_avg;
1814 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1815 unsigned long task_state_change;
1817 int pagefault_disabled;
1818 /* CPU-specific state of this task */
1819 struct thread_struct thread;
1821 * WARNING: on x86, 'thread_struct' contains a variable-sized
1822 * structure. It *MUST* be at the end of 'task_struct'.
1824 * Do not put anything below here!
1828 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1829 extern int arch_task_struct_size __read_mostly;
1831 # define arch_task_struct_size (sizeof(struct task_struct))
1834 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1835 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1837 #define TNF_MIGRATED 0x01
1838 #define TNF_NO_GROUP 0x02
1839 #define TNF_SHARED 0x04
1840 #define TNF_FAULT_LOCAL 0x08
1841 #define TNF_MIGRATE_FAIL 0x10
1843 #ifdef CONFIG_NUMA_BALANCING
1844 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1845 extern pid_t task_numa_group_id(struct task_struct *p);
1846 extern void set_numabalancing_state(bool enabled);
1847 extern void task_numa_free(struct task_struct *p);
1848 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1849 int src_nid, int dst_cpu);
1851 static inline void task_numa_fault(int last_node, int node, int pages,
1855 static inline pid_t task_numa_group_id(struct task_struct *p)
1859 static inline void set_numabalancing_state(bool enabled)
1862 static inline void task_numa_free(struct task_struct *p)
1865 static inline bool should_numa_migrate_memory(struct task_struct *p,
1866 struct page *page, int src_nid, int dst_cpu)
1872 static inline struct pid *task_pid(struct task_struct *task)
1874 return task->pids[PIDTYPE_PID].pid;
1877 static inline struct pid *task_tgid(struct task_struct *task)
1879 return task->group_leader->pids[PIDTYPE_PID].pid;
1883 * Without tasklist or rcu lock it is not safe to dereference
1884 * the result of task_pgrp/task_session even if task == current,
1885 * we can race with another thread doing sys_setsid/sys_setpgid.
1887 static inline struct pid *task_pgrp(struct task_struct *task)
1889 return task->group_leader->pids[PIDTYPE_PGID].pid;
1892 static inline struct pid *task_session(struct task_struct *task)
1894 return task->group_leader->pids[PIDTYPE_SID].pid;
1897 struct pid_namespace;
1900 * the helpers to get the task's different pids as they are seen
1901 * from various namespaces
1903 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1904 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1906 * task_xid_nr_ns() : id seen from the ns specified;
1908 * set_task_vxid() : assigns a virtual id to a task;
1910 * see also pid_nr() etc in include/linux/pid.h
1912 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1913 struct pid_namespace *ns);
1915 static inline pid_t task_pid_nr(struct task_struct *tsk)
1920 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1921 struct pid_namespace *ns)
1923 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1926 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1928 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1932 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1937 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1939 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1941 return pid_vnr(task_tgid(tsk));
1945 static inline int pid_alive(const struct task_struct *p);
1946 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1952 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1958 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1960 return task_ppid_nr_ns(tsk, &init_pid_ns);
1963 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1964 struct pid_namespace *ns)
1966 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1969 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1971 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1975 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1976 struct pid_namespace *ns)
1978 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1981 static inline pid_t task_session_vnr(struct task_struct *tsk)
1983 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1986 /* obsolete, do not use */
1987 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1989 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1993 * pid_alive - check that a task structure is not stale
1994 * @p: Task structure to be checked.
1996 * Test if a process is not yet dead (at most zombie state)
1997 * If pid_alive fails, then pointers within the task structure
1998 * can be stale and must not be dereferenced.
2000 * Return: 1 if the process is alive. 0 otherwise.
2002 static inline int pid_alive(const struct task_struct *p)
2004 return p->pids[PIDTYPE_PID].pid != NULL;
2008 * is_global_init - check if a task structure is init. Since init
2009 * is free to have sub-threads we need to check tgid.
2010 * @tsk: Task structure to be checked.
2012 * Check if a task structure is the first user space task the kernel created.
2014 * Return: 1 if the task structure is init. 0 otherwise.
2016 static inline int is_global_init(struct task_struct *tsk)
2018 return task_tgid_nr(tsk) == 1;
2021 extern struct pid *cad_pid;
2023 extern void free_task(struct task_struct *tsk);
2024 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2026 extern void __put_task_struct(struct task_struct *t);
2028 static inline void put_task_struct(struct task_struct *t)
2030 if (atomic_dec_and_test(&t->usage))
2031 __put_task_struct(t);
2034 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2035 extern void task_cputime(struct task_struct *t,
2036 cputime_t *utime, cputime_t *stime);
2037 extern void task_cputime_scaled(struct task_struct *t,
2038 cputime_t *utimescaled, cputime_t *stimescaled);
2039 extern cputime_t task_gtime(struct task_struct *t);
2041 static inline void task_cputime(struct task_struct *t,
2042 cputime_t *utime, cputime_t *stime)
2050 static inline void task_cputime_scaled(struct task_struct *t,
2051 cputime_t *utimescaled,
2052 cputime_t *stimescaled)
2055 *utimescaled = t->utimescaled;
2057 *stimescaled = t->stimescaled;
2060 static inline cputime_t task_gtime(struct task_struct *t)
2065 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2066 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2071 #define PF_EXITING 0x00000004 /* getting shut down */
2072 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2073 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2074 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2075 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2076 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2077 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2078 #define PF_DUMPCORE 0x00000200 /* dumped core */
2079 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2080 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2081 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2082 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2083 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2084 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2085 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2086 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2087 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2088 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2089 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2090 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2091 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2092 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2093 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2094 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2095 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2096 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2097 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2100 * Only the _current_ task can read/write to tsk->flags, but other
2101 * tasks can access tsk->flags in readonly mode for example
2102 * with tsk_used_math (like during threaded core dumping).
2103 * There is however an exception to this rule during ptrace
2104 * or during fork: the ptracer task is allowed to write to the
2105 * child->flags of its traced child (same goes for fork, the parent
2106 * can write to the child->flags), because we're guaranteed the
2107 * child is not running and in turn not changing child->flags
2108 * at the same time the parent does it.
2110 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2111 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2112 #define clear_used_math() clear_stopped_child_used_math(current)
2113 #define set_used_math() set_stopped_child_used_math(current)
2114 #define conditional_stopped_child_used_math(condition, child) \
2115 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2116 #define conditional_used_math(condition) \
2117 conditional_stopped_child_used_math(condition, current)
2118 #define copy_to_stopped_child_used_math(child) \
2119 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2120 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2121 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2122 #define used_math() tsk_used_math(current)
2124 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2125 * __GFP_FS is also cleared as it implies __GFP_IO.
2127 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2129 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2130 flags &= ~(__GFP_IO | __GFP_FS);
2134 static inline unsigned int memalloc_noio_save(void)
2136 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2137 current->flags |= PF_MEMALLOC_NOIO;
2141 static inline void memalloc_noio_restore(unsigned int flags)
2143 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2146 /* Per-process atomic flags. */
2147 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2148 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2149 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2152 #define TASK_PFA_TEST(name, func) \
2153 static inline bool task_##func(struct task_struct *p) \
2154 { return test_bit(PFA_##name, &p->atomic_flags); }
2155 #define TASK_PFA_SET(name, func) \
2156 static inline void task_set_##func(struct task_struct *p) \
2157 { set_bit(PFA_##name, &p->atomic_flags); }
2158 #define TASK_PFA_CLEAR(name, func) \
2159 static inline void task_clear_##func(struct task_struct *p) \
2160 { clear_bit(PFA_##name, &p->atomic_flags); }
2162 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2163 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2165 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2166 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2167 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2169 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2170 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2171 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2174 * task->jobctl flags
2176 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2178 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2179 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2180 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2181 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2182 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2183 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2184 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2186 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2187 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2188 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2189 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2190 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2191 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2192 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2194 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2195 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2197 extern bool task_set_jobctl_pending(struct task_struct *task,
2198 unsigned long mask);
2199 extern void task_clear_jobctl_trapping(struct task_struct *task);
2200 extern void task_clear_jobctl_pending(struct task_struct *task,
2201 unsigned long mask);
2203 static inline void rcu_copy_process(struct task_struct *p)
2205 #ifdef CONFIG_PREEMPT_RCU
2206 p->rcu_read_lock_nesting = 0;
2207 p->rcu_read_unlock_special.s = 0;
2208 p->rcu_blocked_node = NULL;
2209 INIT_LIST_HEAD(&p->rcu_node_entry);
2210 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2211 #ifdef CONFIG_TASKS_RCU
2212 p->rcu_tasks_holdout = false;
2213 INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2214 p->rcu_tasks_idle_cpu = -1;
2215 #endif /* #ifdef CONFIG_TASKS_RCU */
2218 static inline void tsk_restore_flags(struct task_struct *task,
2219 unsigned long orig_flags, unsigned long flags)
2221 task->flags &= ~flags;
2222 task->flags |= orig_flags & flags;
2225 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2226 const struct cpumask *trial);
2227 extern int task_can_attach(struct task_struct *p,
2228 const struct cpumask *cs_cpus_allowed);
2230 extern void do_set_cpus_allowed(struct task_struct *p,
2231 const struct cpumask *new_mask);
2233 extern int set_cpus_allowed_ptr(struct task_struct *p,
2234 const struct cpumask *new_mask);
2236 static inline void do_set_cpus_allowed(struct task_struct *p,
2237 const struct cpumask *new_mask)
2240 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2241 const struct cpumask *new_mask)
2243 if (!cpumask_test_cpu(0, new_mask))
2249 #ifdef CONFIG_NO_HZ_COMMON
2250 void calc_load_enter_idle(void);
2251 void calc_load_exit_idle(void);
2253 static inline void calc_load_enter_idle(void) { }
2254 static inline void calc_load_exit_idle(void) { }
2255 #endif /* CONFIG_NO_HZ_COMMON */
2258 * Do not use outside of architecture code which knows its limitations.
2260 * sched_clock() has no promise of monotonicity or bounded drift between
2261 * CPUs, use (which you should not) requires disabling IRQs.
2263 * Please use one of the three interfaces below.
2265 extern unsigned long long notrace sched_clock(void);
2267 * See the comment in kernel/sched/clock.c
2269 extern u64 cpu_clock(int cpu);
2270 extern u64 local_clock(void);
2271 extern u64 running_clock(void);
2272 extern u64 sched_clock_cpu(int cpu);
2275 extern void sched_clock_init(void);
2277 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2278 static inline void sched_clock_tick(void)
2282 static inline void sched_clock_idle_sleep_event(void)
2286 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2291 * Architectures can set this to 1 if they have specified
2292 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2293 * but then during bootup it turns out that sched_clock()
2294 * is reliable after all:
2296 extern int sched_clock_stable(void);
2297 extern void set_sched_clock_stable(void);
2298 extern void clear_sched_clock_stable(void);
2300 extern void sched_clock_tick(void);
2301 extern void sched_clock_idle_sleep_event(void);
2302 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2305 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2307 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2308 * The reason for this explicit opt-in is not to have perf penalty with
2309 * slow sched_clocks.
2311 extern void enable_sched_clock_irqtime(void);
2312 extern void disable_sched_clock_irqtime(void);
2314 static inline void enable_sched_clock_irqtime(void) {}
2315 static inline void disable_sched_clock_irqtime(void) {}
2318 extern unsigned long long
2319 task_sched_runtime(struct task_struct *task);
2321 /* sched_exec is called by processes performing an exec */
2323 extern void sched_exec(void);
2325 #define sched_exec() {}
2328 extern void sched_clock_idle_sleep_event(void);
2329 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2331 #ifdef CONFIG_HOTPLUG_CPU
2332 extern void idle_task_exit(void);
2334 static inline void idle_task_exit(void) {}
2337 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2338 extern void wake_up_nohz_cpu(int cpu);
2340 static inline void wake_up_nohz_cpu(int cpu) { }
2343 #ifdef CONFIG_NO_HZ_FULL
2344 extern bool sched_can_stop_tick(void);
2345 extern u64 scheduler_tick_max_deferment(void);
2347 static inline bool sched_can_stop_tick(void) { return false; }
2350 #ifdef CONFIG_SCHED_AUTOGROUP
2351 extern void sched_autogroup_create_attach(struct task_struct *p);
2352 extern void sched_autogroup_detach(struct task_struct *p);
2353 extern void sched_autogroup_fork(struct signal_struct *sig);
2354 extern void sched_autogroup_exit(struct signal_struct *sig);
2355 #ifdef CONFIG_PROC_FS
2356 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2357 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2360 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2361 static inline void sched_autogroup_detach(struct task_struct *p) { }
2362 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2363 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2366 extern int yield_to(struct task_struct *p, bool preempt);
2367 extern void set_user_nice(struct task_struct *p, long nice);
2368 extern int task_prio(const struct task_struct *p);
2370 * task_nice - return the nice value of a given task.
2371 * @p: the task in question.
2373 * Return: The nice value [ -20 ... 0 ... 19 ].
2375 static inline int task_nice(const struct task_struct *p)
2377 return PRIO_TO_NICE((p)->static_prio);
2379 extern int can_nice(const struct task_struct *p, const int nice);
2380 extern int task_curr(const struct task_struct *p);
2381 extern int idle_cpu(int cpu);
2382 extern int sched_setscheduler(struct task_struct *, int,
2383 const struct sched_param *);
2384 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2385 const struct sched_param *);
2386 extern int sched_setattr(struct task_struct *,
2387 const struct sched_attr *);
2388 extern struct task_struct *idle_task(int cpu);
2390 * is_idle_task - is the specified task an idle task?
2391 * @p: the task in question.
2393 * Return: 1 if @p is an idle task. 0 otherwise.
2395 static inline bool is_idle_task(const struct task_struct *p)
2399 extern struct task_struct *curr_task(int cpu);
2400 extern void set_curr_task(int cpu, struct task_struct *p);
2404 union thread_union {
2405 struct thread_info thread_info;
2406 unsigned long stack[THREAD_SIZE/sizeof(long)];
2409 #ifndef __HAVE_ARCH_KSTACK_END
2410 static inline int kstack_end(void *addr)
2412 /* Reliable end of stack detection:
2413 * Some APM bios versions misalign the stack
2415 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2419 extern union thread_union init_thread_union;
2420 extern struct task_struct init_task;
2422 extern struct mm_struct init_mm;
2424 extern struct pid_namespace init_pid_ns;
2427 * find a task by one of its numerical ids
2429 * find_task_by_pid_ns():
2430 * finds a task by its pid in the specified namespace
2431 * find_task_by_vpid():
2432 * finds a task by its virtual pid
2434 * see also find_vpid() etc in include/linux/pid.h
2437 extern struct task_struct *find_task_by_vpid(pid_t nr);
2438 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2439 struct pid_namespace *ns);
2441 /* per-UID process charging. */
2442 extern struct user_struct * alloc_uid(kuid_t);
2443 static inline struct user_struct *get_uid(struct user_struct *u)
2445 atomic_inc(&u->__count);
2448 extern void free_uid(struct user_struct *);
2450 #include <asm/current.h>
2452 extern void xtime_update(unsigned long ticks);
2454 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2455 extern int wake_up_process(struct task_struct *tsk);
2456 extern void wake_up_new_task(struct task_struct *tsk);
2458 extern void kick_process(struct task_struct *tsk);
2460 static inline void kick_process(struct task_struct *tsk) { }
2462 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2463 extern void sched_dead(struct task_struct *p);
2465 extern void proc_caches_init(void);
2466 extern void flush_signals(struct task_struct *);
2467 extern void ignore_signals(struct task_struct *);
2468 extern void flush_signal_handlers(struct task_struct *, int force_default);
2469 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2471 static inline int kernel_dequeue_signal(siginfo_t *info)
2473 struct task_struct *tsk = current;
2477 spin_lock_irq(&tsk->sighand->siglock);
2478 ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2479 spin_unlock_irq(&tsk->sighand->siglock);
2484 static inline void kernel_signal_stop(void)
2486 spin_lock_irq(¤t->sighand->siglock);
2487 if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2488 __set_current_state(TASK_STOPPED);
2489 spin_unlock_irq(¤t->sighand->siglock);
2494 extern void release_task(struct task_struct * p);
2495 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2496 extern int force_sigsegv(int, struct task_struct *);
2497 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2498 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2499 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2500 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2501 const struct cred *, u32);
2502 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2503 extern int kill_pid(struct pid *pid, int sig, int priv);
2504 extern int kill_proc_info(int, struct siginfo *, pid_t);
2505 extern __must_check bool do_notify_parent(struct task_struct *, int);
2506 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2507 extern void force_sig(int, struct task_struct *);
2508 extern int send_sig(int, struct task_struct *, int);
2509 extern int zap_other_threads(struct task_struct *p);
2510 extern struct sigqueue *sigqueue_alloc(void);
2511 extern void sigqueue_free(struct sigqueue *);
2512 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2513 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2515 static inline void restore_saved_sigmask(void)
2517 if (test_and_clear_restore_sigmask())
2518 __set_current_blocked(¤t->saved_sigmask);
2521 static inline sigset_t *sigmask_to_save(void)
2523 sigset_t *res = ¤t->blocked;
2524 if (unlikely(test_restore_sigmask()))
2525 res = ¤t->saved_sigmask;
2529 static inline int kill_cad_pid(int sig, int priv)
2531 return kill_pid(cad_pid, sig, priv);
2534 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2535 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2536 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2537 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2540 * True if we are on the alternate signal stack.
2542 static inline int on_sig_stack(unsigned long sp)
2544 #ifdef CONFIG_STACK_GROWSUP
2545 return sp >= current->sas_ss_sp &&
2546 sp - current->sas_ss_sp < current->sas_ss_size;
2548 return sp > current->sas_ss_sp &&
2549 sp - current->sas_ss_sp <= current->sas_ss_size;
2553 static inline int sas_ss_flags(unsigned long sp)
2555 if (!current->sas_ss_size)
2558 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2561 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2563 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2564 #ifdef CONFIG_STACK_GROWSUP
2565 return current->sas_ss_sp;
2567 return current->sas_ss_sp + current->sas_ss_size;
2573 * Routines for handling mm_structs
2575 extern struct mm_struct * mm_alloc(void);
2577 /* mmdrop drops the mm and the page tables */
2578 extern void __mmdrop(struct mm_struct *);
2579 static inline void mmdrop(struct mm_struct * mm)
2581 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2585 /* mmput gets rid of the mappings and all user-space */
2586 extern void mmput(struct mm_struct *);
2587 /* Grab a reference to a task's mm, if it is not already going away */
2588 extern struct mm_struct *get_task_mm(struct task_struct *task);
2590 * Grab a reference to a task's mm, if it is not already going away
2591 * and ptrace_may_access with the mode parameter passed to it
2594 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2595 /* Remove the current tasks stale references to the old mm_struct */
2596 extern void mm_release(struct task_struct *, struct mm_struct *);
2598 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2599 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2600 struct task_struct *, unsigned long);
2602 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2603 struct task_struct *);
2605 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2606 * via pt_regs, so ignore the tls argument passed via C. */
2607 static inline int copy_thread_tls(
2608 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2609 struct task_struct *p, unsigned long tls)
2611 return copy_thread(clone_flags, sp, arg, p);
2614 extern void flush_thread(void);
2615 extern void exit_thread(void);
2617 extern void exit_files(struct task_struct *);
2618 extern void __cleanup_sighand(struct sighand_struct *);
2620 extern void exit_itimers(struct signal_struct *);
2621 extern void flush_itimer_signals(void);
2623 extern void do_group_exit(int);
2625 extern int do_execve(struct filename *,
2626 const char __user * const __user *,
2627 const char __user * const __user *);
2628 extern int do_execveat(int, struct filename *,
2629 const char __user * const __user *,
2630 const char __user * const __user *,
2632 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2633 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2634 struct task_struct *fork_idle(int);
2635 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2637 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2638 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2640 __set_task_comm(tsk, from, false);
2642 extern char *get_task_comm(char *to, struct task_struct *tsk);
2645 void scheduler_ipi(void);
2646 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2648 static inline void scheduler_ipi(void) { }
2649 static inline unsigned long wait_task_inactive(struct task_struct *p,
2656 #define tasklist_empty() \
2657 list_empty(&init_task.tasks)
2659 #define next_task(p) \
2660 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2662 #define for_each_process(p) \
2663 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2665 extern bool current_is_single_threaded(void);
2668 * Careful: do_each_thread/while_each_thread is a double loop so
2669 * 'break' will not work as expected - use goto instead.
2671 #define do_each_thread(g, t) \
2672 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2674 #define while_each_thread(g, t) \
2675 while ((t = next_thread(t)) != g)
2677 #define __for_each_thread(signal, t) \
2678 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2680 #define for_each_thread(p, t) \
2681 __for_each_thread((p)->signal, t)
2683 /* Careful: this is a double loop, 'break' won't work as expected. */
2684 #define for_each_process_thread(p, t) \
2685 for_each_process(p) for_each_thread(p, t)
2687 static inline int get_nr_threads(struct task_struct *tsk)
2689 return tsk->signal->nr_threads;
2692 static inline bool thread_group_leader(struct task_struct *p)
2694 return p->exit_signal >= 0;
2697 /* Do to the insanities of de_thread it is possible for a process
2698 * to have the pid of the thread group leader without actually being
2699 * the thread group leader. For iteration through the pids in proc
2700 * all we care about is that we have a task with the appropriate
2701 * pid, we don't actually care if we have the right task.
2703 static inline bool has_group_leader_pid(struct task_struct *p)
2705 return task_pid(p) == p->signal->leader_pid;
2709 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2711 return p1->signal == p2->signal;
2714 static inline struct task_struct *next_thread(const struct task_struct *p)
2716 return list_entry_rcu(p->thread_group.next,
2717 struct task_struct, thread_group);
2720 static inline int thread_group_empty(struct task_struct *p)
2722 return list_empty(&p->thread_group);
2725 #define delay_group_leader(p) \
2726 (thread_group_leader(p) && !thread_group_empty(p))
2729 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2730 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2731 * pins the final release of task.io_context. Also protects ->cpuset and
2732 * ->cgroup.subsys[]. And ->vfork_done.
2734 * Nests both inside and outside of read_lock(&tasklist_lock).
2735 * It must not be nested with write_lock_irq(&tasklist_lock),
2736 * neither inside nor outside.
2738 static inline void task_lock(struct task_struct *p)
2740 spin_lock(&p->alloc_lock);
2743 static inline void task_unlock(struct task_struct *p)
2745 spin_unlock(&p->alloc_lock);
2748 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2749 unsigned long *flags);
2751 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2752 unsigned long *flags)
2754 struct sighand_struct *ret;
2756 ret = __lock_task_sighand(tsk, flags);
2757 (void)__cond_lock(&tsk->sighand->siglock, ret);
2761 static inline void unlock_task_sighand(struct task_struct *tsk,
2762 unsigned long *flags)
2764 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2768 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2769 * @tsk: task causing the changes
2771 * All operations which modify a threadgroup - a new thread joining the
2772 * group, death of a member thread (the assertion of PF_EXITING) and
2773 * exec(2) dethreading the process and replacing the leader - are wrapped
2774 * by threadgroup_change_{begin|end}(). This is to provide a place which
2775 * subsystems needing threadgroup stability can hook into for
2778 static inline void threadgroup_change_begin(struct task_struct *tsk)
2781 cgroup_threadgroup_change_begin(tsk);
2785 * threadgroup_change_end - mark the end of changes to a threadgroup
2786 * @tsk: task causing the changes
2788 * See threadgroup_change_begin().
2790 static inline void threadgroup_change_end(struct task_struct *tsk)
2792 cgroup_threadgroup_change_end(tsk);
2795 #ifndef __HAVE_THREAD_FUNCTIONS
2797 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2798 #define task_stack_page(task) ((task)->stack)
2800 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2802 *task_thread_info(p) = *task_thread_info(org);
2803 task_thread_info(p)->task = p;
2807 * Return the address of the last usable long on the stack.
2809 * When the stack grows down, this is just above the thread
2810 * info struct. Going any lower will corrupt the threadinfo.
2812 * When the stack grows up, this is the highest address.
2813 * Beyond that position, we corrupt data on the next page.
2815 static inline unsigned long *end_of_stack(struct task_struct *p)
2817 #ifdef CONFIG_STACK_GROWSUP
2818 return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
2820 return (unsigned long *)(task_thread_info(p) + 1);
2825 #define task_stack_end_corrupted(task) \
2826 (*(end_of_stack(task)) != STACK_END_MAGIC)
2828 static inline int object_is_on_stack(void *obj)
2830 void *stack = task_stack_page(current);
2832 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2835 extern void thread_info_cache_init(void);
2837 #ifdef CONFIG_DEBUG_STACK_USAGE
2838 static inline unsigned long stack_not_used(struct task_struct *p)
2840 unsigned long *n = end_of_stack(p);
2842 do { /* Skip over canary */
2846 return (unsigned long)n - (unsigned long)end_of_stack(p);
2849 extern void set_task_stack_end_magic(struct task_struct *tsk);
2851 /* set thread flags in other task's structures
2852 * - see asm/thread_info.h for TIF_xxxx flags available
2854 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2856 set_ti_thread_flag(task_thread_info(tsk), flag);
2859 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2861 clear_ti_thread_flag(task_thread_info(tsk), flag);
2864 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2866 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2869 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2871 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2874 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2876 return test_ti_thread_flag(task_thread_info(tsk), flag);
2879 static inline void set_tsk_need_resched(struct task_struct *tsk)
2881 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2884 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2886 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2889 static inline int test_tsk_need_resched(struct task_struct *tsk)
2891 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2894 static inline int restart_syscall(void)
2896 set_tsk_thread_flag(current, TIF_SIGPENDING);
2897 return -ERESTARTNOINTR;
2900 static inline int signal_pending(struct task_struct *p)
2902 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2905 static inline int __fatal_signal_pending(struct task_struct *p)
2907 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2910 static inline int fatal_signal_pending(struct task_struct *p)
2912 return signal_pending(p) && __fatal_signal_pending(p);
2915 static inline int signal_pending_state(long state, struct task_struct *p)
2917 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2919 if (!signal_pending(p))
2922 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2926 * cond_resched() and cond_resched_lock(): latency reduction via
2927 * explicit rescheduling in places that are safe. The return
2928 * value indicates whether a reschedule was done in fact.
2929 * cond_resched_lock() will drop the spinlock before scheduling,
2930 * cond_resched_softirq() will enable bhs before scheduling.
2932 extern int _cond_resched(void);
2934 #define cond_resched() ({ \
2935 ___might_sleep(__FILE__, __LINE__, 0); \
2939 extern int __cond_resched_lock(spinlock_t *lock);
2941 #define cond_resched_lock(lock) ({ \
2942 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2943 __cond_resched_lock(lock); \
2946 extern int __cond_resched_softirq(void);
2948 #define cond_resched_softirq() ({ \
2949 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2950 __cond_resched_softirq(); \
2953 static inline void cond_resched_rcu(void)
2955 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2963 * Does a critical section need to be broken due to another
2964 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2965 * but a general need for low latency)
2967 static inline int spin_needbreak(spinlock_t *lock)
2969 #ifdef CONFIG_PREEMPT
2970 return spin_is_contended(lock);
2977 * Idle thread specific functions to determine the need_resched
2980 #ifdef TIF_POLLING_NRFLAG
2981 static inline int tsk_is_polling(struct task_struct *p)
2983 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2986 static inline void __current_set_polling(void)
2988 set_thread_flag(TIF_POLLING_NRFLAG);
2991 static inline bool __must_check current_set_polling_and_test(void)
2993 __current_set_polling();
2996 * Polling state must be visible before we test NEED_RESCHED,
2997 * paired by resched_curr()
2999 smp_mb__after_atomic();
3001 return unlikely(tif_need_resched());
3004 static inline void __current_clr_polling(void)
3006 clear_thread_flag(TIF_POLLING_NRFLAG);
3009 static inline bool __must_check current_clr_polling_and_test(void)
3011 __current_clr_polling();
3014 * Polling state must be visible before we test NEED_RESCHED,
3015 * paired by resched_curr()
3017 smp_mb__after_atomic();
3019 return unlikely(tif_need_resched());
3023 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3024 static inline void __current_set_polling(void) { }
3025 static inline void __current_clr_polling(void) { }
3027 static inline bool __must_check current_set_polling_and_test(void)
3029 return unlikely(tif_need_resched());
3031 static inline bool __must_check current_clr_polling_and_test(void)
3033 return unlikely(tif_need_resched());
3037 static inline void current_clr_polling(void)
3039 __current_clr_polling();
3042 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3043 * Once the bit is cleared, we'll get IPIs with every new
3044 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3047 smp_mb(); /* paired with resched_curr() */
3049 preempt_fold_need_resched();
3052 static __always_inline bool need_resched(void)
3054 return unlikely(tif_need_resched());
3058 * Thread group CPU time accounting.
3060 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3061 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3064 * Reevaluate whether the task has signals pending delivery.
3065 * Wake the task if so.
3066 * This is required every time the blocked sigset_t changes.
3067 * callers must hold sighand->siglock.
3069 extern void recalc_sigpending_and_wake(struct task_struct *t);
3070 extern void recalc_sigpending(void);
3072 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3074 static inline void signal_wake_up(struct task_struct *t, bool resume)
3076 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3078 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3080 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3084 * Wrappers for p->thread_info->cpu access. No-op on UP.
3088 static inline unsigned int task_cpu(const struct task_struct *p)
3090 return task_thread_info(p)->cpu;
3093 static inline int task_node(const struct task_struct *p)
3095 return cpu_to_node(task_cpu(p));
3098 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3102 static inline unsigned int task_cpu(const struct task_struct *p)
3107 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3111 #endif /* CONFIG_SMP */
3113 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3114 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3116 #ifdef CONFIG_CGROUP_SCHED
3117 extern struct task_group root_task_group;
3118 #endif /* CONFIG_CGROUP_SCHED */
3120 extern int task_can_switch_user(struct user_struct *up,
3121 struct task_struct *tsk);
3123 #ifdef CONFIG_TASK_XACCT
3124 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3126 tsk->ioac.rchar += amt;
3129 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3131 tsk->ioac.wchar += amt;
3134 static inline void inc_syscr(struct task_struct *tsk)
3139 static inline void inc_syscw(struct task_struct *tsk)
3144 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3148 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3152 static inline void inc_syscr(struct task_struct *tsk)
3156 static inline void inc_syscw(struct task_struct *tsk)
3161 #ifndef TASK_SIZE_OF
3162 #define TASK_SIZE_OF(tsk) TASK_SIZE
3166 extern void mm_update_next_owner(struct mm_struct *mm);
3168 static inline void mm_update_next_owner(struct mm_struct *mm)
3171 #endif /* CONFIG_MEMCG */
3173 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3176 return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3179 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3182 return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3185 static inline unsigned long rlimit(unsigned int limit)
3187 return task_rlimit(current, limit);
3190 static inline unsigned long rlimit_max(unsigned int limit)
3192 return task_rlimit_max(current, limit);