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>
33 #include <linux/smp.h>
34 #include <linux/sem.h>
35 #include <linux/shm.h>
36 #include <linux/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/signal_types.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/kcov.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/topology.h>
61 #include <linux/magic.h>
62 #include <linux/cgroup-defs.h>
64 #include <asm/processor.h>
66 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
69 * Extended scheduling parameters data structure.
71 * This is needed because the original struct sched_param can not be
72 * altered without introducing ABI issues with legacy applications
73 * (e.g., in sched_getparam()).
75 * However, the possibility of specifying more than just a priority for
76 * the tasks may be useful for a wide variety of application fields, e.g.,
77 * multimedia, streaming, automation and control, and many others.
79 * This variant (sched_attr) is meant at describing a so-called
80 * sporadic time-constrained task. In such model a task is specified by:
81 * - the activation period or minimum instance inter-arrival time;
82 * - the maximum (or average, depending on the actual scheduling
83 * discipline) computation time of all instances, a.k.a. runtime;
84 * - the deadline (relative to the actual activation time) of each
86 * Very briefly, a periodic (sporadic) task asks for the execution of
87 * some specific computation --which is typically called an instance--
88 * (at most) every period. Moreover, each instance typically lasts no more
89 * than the runtime and must be completed by time instant t equal to
90 * the instance activation time + the deadline.
92 * This is reflected by the actual fields of the sched_attr structure:
94 * @size size of the structure, for fwd/bwd compat.
96 * @sched_policy task's scheduling policy
97 * @sched_flags for customizing the scheduler behaviour
98 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
99 * @sched_priority task's static priority (SCHED_FIFO/RR)
100 * @sched_deadline representative of the task's deadline
101 * @sched_runtime representative of the task's runtime
102 * @sched_period representative of the task's period
104 * Given this task model, there are a multiplicity of scheduling algorithms
105 * and policies, that can be used to ensure all the tasks will make their
106 * timing constraints.
108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
109 * only user of this new interface. More information about the algorithm
110 * available in the scheduling class file or in Documentation/.
118 /* SCHED_NORMAL, SCHED_BATCH */
121 /* SCHED_FIFO, SCHED_RR */
130 struct futex_pi_state
;
131 struct robust_list_head
;
134 struct perf_event_context
;
140 * These are the constant used to fake the fixed-point load-average
141 * counting. Some notes:
142 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
143 * a load-average precision of 10 bits integer + 11 bits fractional
144 * - if you want to count load-averages more often, you need more
145 * precision, or rounding will get you. With 2-second counting freq,
146 * the EXP_n values would be 1981, 2034 and 2043 if still using only
149 extern unsigned long avenrun
[]; /* Load averages */
150 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
152 #define FSHIFT 11 /* nr of bits of precision */
153 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
154 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
155 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
156 #define EXP_5 2014 /* 1/exp(5sec/5min) */
157 #define EXP_15 2037 /* 1/exp(5sec/15min) */
159 #define CALC_LOAD(load,exp,n) \
161 load += n*(FIXED_1-exp); \
164 extern unsigned long total_forks
;
165 extern int nr_threads
;
166 DECLARE_PER_CPU(unsigned long, process_counts
);
167 extern int nr_processes(void);
168 extern unsigned long nr_running(void);
169 extern bool single_task_running(void);
170 extern unsigned long nr_iowait(void);
171 extern unsigned long nr_iowait_cpu(int cpu
);
172 extern void get_iowait_load(unsigned long *nr_waiters
, unsigned long *load
);
174 extern void calc_global_load(unsigned long ticks
);
176 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
177 extern void cpu_load_update_nohz_start(void);
178 extern void cpu_load_update_nohz_stop(void);
180 static inline void cpu_load_update_nohz_start(void) { }
181 static inline void cpu_load_update_nohz_stop(void) { }
184 extern void dump_cpu_task(int cpu
);
189 #ifdef CONFIG_SCHED_DEBUG
190 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
191 extern void proc_sched_set_task(struct task_struct
*p
);
195 * Task state bitmask. NOTE! These bits are also
196 * encoded in fs/proc/array.c: get_task_state().
198 * We have two separate sets of flags: task->state
199 * is about runnability, while task->exit_state are
200 * about the task exiting. Confusing, but this way
201 * modifying one set can't modify the other one by
204 #define TASK_RUNNING 0
205 #define TASK_INTERRUPTIBLE 1
206 #define TASK_UNINTERRUPTIBLE 2
207 #define __TASK_STOPPED 4
208 #define __TASK_TRACED 8
209 /* in tsk->exit_state */
211 #define EXIT_ZOMBIE 32
212 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
213 /* in tsk->state again */
215 #define TASK_WAKEKILL 128
216 #define TASK_WAKING 256
217 #define TASK_PARKED 512
218 #define TASK_NOLOAD 1024
219 #define TASK_NEW 2048
220 #define TASK_STATE_MAX 4096
222 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
224 /* Convenience macros for the sake of set_current_state */
225 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
226 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
227 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
229 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
231 /* Convenience macros for the sake of wake_up */
232 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
233 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
235 /* get_task_state() */
236 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
237 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
238 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
240 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
241 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
242 #define task_is_stopped_or_traced(task) \
243 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
244 #define task_contributes_to_load(task) \
245 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
246 (task->flags & PF_FROZEN) == 0 && \
247 (task->state & TASK_NOLOAD) == 0)
249 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
251 #define __set_current_state(state_value) \
253 current->task_state_change = _THIS_IP_; \
254 current->state = (state_value); \
256 #define set_current_state(state_value) \
258 current->task_state_change = _THIS_IP_; \
259 smp_store_mb(current->state, (state_value)); \
264 * set_current_state() includes a barrier so that the write of current->state
265 * is correctly serialised wrt the caller's subsequent test of whether to
269 * set_current_state(TASK_UNINTERRUPTIBLE);
275 * __set_current_state(TASK_RUNNING);
277 * If the caller does not need such serialisation (because, for instance, the
278 * condition test and condition change and wakeup are under the same lock) then
279 * use __set_current_state().
281 * The above is typically ordered against the wakeup, which does:
283 * need_sleep = false;
284 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
286 * Where wake_up_state() (and all other wakeup primitives) imply enough
287 * barriers to order the store of the variable against wakeup.
289 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
290 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
291 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
293 * This is obviously fine, since they both store the exact same value.
295 * Also see the comments of try_to_wake_up().
297 #define __set_current_state(state_value) \
298 do { current->state = (state_value); } while (0)
299 #define set_current_state(state_value) \
300 smp_store_mb(current->state, (state_value))
304 /* Task command name length */
305 #define TASK_COMM_LEN 16
307 #include <linux/spinlock.h>
310 * This serializes "schedule()" and also protects
311 * the run-queue from deletions/modifications (but
312 * _adding_ to the beginning of the run-queue has
315 extern rwlock_t tasklist_lock
;
316 extern spinlock_t mmlist_lock
;
320 #ifdef CONFIG_PROVE_RCU
321 extern int lockdep_tasklist_lock_is_held(void);
322 #endif /* #ifdef CONFIG_PROVE_RCU */
324 extern void sched_init(void);
325 extern void sched_init_smp(void);
326 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
327 extern void init_idle(struct task_struct
*idle
, int cpu
);
328 extern void init_idle_bootup_task(struct task_struct
*idle
);
330 extern cpumask_var_t cpu_isolated_map
;
332 extern int runqueue_is_locked(int cpu
);
334 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
335 extern void nohz_balance_enter_idle(int cpu
);
336 extern void set_cpu_sd_state_idle(void);
337 extern int get_nohz_timer_target(void);
339 static inline void nohz_balance_enter_idle(int cpu
) { }
340 static inline void set_cpu_sd_state_idle(void) { }
344 * Only dump TASK_* tasks. (0 for all tasks)
346 extern void show_state_filter(unsigned long state_filter
);
348 static inline void show_state(void)
350 show_state_filter(0);
353 extern void show_regs(struct pt_regs
*);
356 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
357 * task), SP is the stack pointer of the first frame that should be shown in the back
358 * trace (or NULL if the entire call-chain of the task should be shown).
360 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
362 extern void cpu_init (void);
363 extern void trap_init(void);
364 extern void update_process_times(int user
);
365 extern void scheduler_tick(void);
366 extern int sched_cpu_starting(unsigned int cpu
);
367 extern int sched_cpu_activate(unsigned int cpu
);
368 extern int sched_cpu_deactivate(unsigned int cpu
);
370 #ifdef CONFIG_HOTPLUG_CPU
371 extern int sched_cpu_dying(unsigned int cpu
);
373 # define sched_cpu_dying NULL
376 extern void sched_show_task(struct task_struct
*p
);
378 #ifdef CONFIG_LOCKUP_DETECTOR
379 extern void touch_softlockup_watchdog_sched(void);
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_sched(void)
393 static inline void touch_softlockup_watchdog(void)
396 static inline void touch_softlockup_watchdog_sync(void)
399 static inline void touch_all_softlockup_watchdogs(void)
402 static inline void lockup_detector_init(void)
407 #ifdef CONFIG_DETECT_HUNG_TASK
408 void reset_hung_task_detector(void);
410 static inline void reset_hung_task_detector(void)
415 /* Attach to any functions which should be ignored in wchan output. */
416 #define __sched __attribute__((__section__(".sched.text")))
418 /* Linker adds these: start and end of __sched functions */
419 extern char __sched_text_start
[], __sched_text_end
[];
421 /* Is this address in the __sched functions? */
422 extern int in_sched_functions(unsigned long addr
);
424 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
425 extern signed long schedule_timeout(signed long timeout
);
426 extern signed long schedule_timeout_interruptible(signed long timeout
);
427 extern signed long schedule_timeout_killable(signed long timeout
);
428 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
429 extern signed long schedule_timeout_idle(signed long timeout
);
430 asmlinkage
void schedule(void);
431 extern void schedule_preempt_disabled(void);
433 extern int __must_check
io_schedule_prepare(void);
434 extern void io_schedule_finish(int token
);
435 extern long io_schedule_timeout(long timeout
);
436 extern void io_schedule(void);
438 void __noreturn
do_task_dead(void);
441 struct user_namespace
;
444 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
446 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
447 unsigned long, unsigned long);
449 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
450 unsigned long len
, unsigned long pgoff
,
451 unsigned long flags
);
453 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
456 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
457 #define SUID_DUMP_USER 1 /* Dump as user of process */
458 #define SUID_DUMP_ROOT 2 /* Dump as root */
462 /* for SUID_DUMP_* above */
463 #define MMF_DUMPABLE_BITS 2
464 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
466 extern void set_dumpable(struct mm_struct
*mm
, int value
);
468 * This returns the actual value of the suid_dumpable flag. For things
469 * that are using this for checking for privilege transitions, it must
470 * test against SUID_DUMP_USER rather than treating it as a boolean
473 static inline int __get_dumpable(unsigned long mm_flags
)
475 return mm_flags
& MMF_DUMPABLE_MASK
;
478 static inline int get_dumpable(struct mm_struct
*mm
)
480 return __get_dumpable(mm
->flags
);
483 /* coredump filter bits */
484 #define MMF_DUMP_ANON_PRIVATE 2
485 #define MMF_DUMP_ANON_SHARED 3
486 #define MMF_DUMP_MAPPED_PRIVATE 4
487 #define MMF_DUMP_MAPPED_SHARED 5
488 #define MMF_DUMP_ELF_HEADERS 6
489 #define MMF_DUMP_HUGETLB_PRIVATE 7
490 #define MMF_DUMP_HUGETLB_SHARED 8
491 #define MMF_DUMP_DAX_PRIVATE 9
492 #define MMF_DUMP_DAX_SHARED 10
494 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
495 #define MMF_DUMP_FILTER_BITS 9
496 #define MMF_DUMP_FILTER_MASK \
497 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
498 #define MMF_DUMP_FILTER_DEFAULT \
499 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
500 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
502 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
503 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
505 # define MMF_DUMP_MASK_DEFAULT_ELF 0
507 /* leave room for more dump flags */
508 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
509 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
511 * This one-shot flag is dropped due to necessity of changing exe once again
514 //#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
516 #define MMF_HAS_UPROBES 19 /* has uprobes */
517 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
518 #define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */
519 #define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */
520 #define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */
522 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
524 struct sighand_struct
{
526 struct k_sigaction action
[_NSIG
];
528 wait_queue_head_t signalfd_wqh
;
531 struct pacct_struct
{
534 unsigned long ac_mem
;
535 u64 ac_utime
, ac_stime
;
536 unsigned long ac_minflt
, ac_majflt
;
545 * struct prev_cputime - snaphsot of system and user cputime
546 * @utime: time spent in user mode
547 * @stime: time spent in system mode
548 * @lock: protects the above two fields
550 * Stores previous user/system time values such that we can guarantee
553 struct prev_cputime
{
554 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
561 static inline void prev_cputime_init(struct prev_cputime
*prev
)
563 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
564 prev
->utime
= prev
->stime
= 0;
565 raw_spin_lock_init(&prev
->lock
);
570 * struct task_cputime - collected CPU time counts
571 * @utime: time spent in user mode, in nanoseconds
572 * @stime: time spent in kernel mode, in nanoseconds
573 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
575 * This structure groups together three kinds of CPU time that are tracked for
576 * threads and thread groups. Most things considering CPU time want to group
577 * these counts together and treat all three of them in parallel.
579 struct task_cputime
{
582 unsigned long long sum_exec_runtime
;
585 /* Alternate field names when used to cache expirations. */
586 #define virt_exp utime
587 #define prof_exp stime
588 #define sched_exp sum_exec_runtime
591 * This is the atomic variant of task_cputime, which can be used for
592 * storing and updating task_cputime statistics without locking.
594 struct task_cputime_atomic
{
597 atomic64_t sum_exec_runtime
;
600 #define INIT_CPUTIME_ATOMIC \
601 (struct task_cputime_atomic) { \
602 .utime = ATOMIC64_INIT(0), \
603 .stime = ATOMIC64_INIT(0), \
604 .sum_exec_runtime = ATOMIC64_INIT(0), \
607 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
610 * Disable preemption until the scheduler is running -- use an unconditional
611 * value so that it also works on !PREEMPT_COUNT kernels.
613 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
615 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
618 * Initial preempt_count value; reflects the preempt_count schedule invariant
619 * which states that during context switches:
621 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
623 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
624 * Note: See finish_task_switch().
626 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
629 * struct thread_group_cputimer - thread group interval timer counts
630 * @cputime_atomic: atomic thread group interval timers.
631 * @running: true when there are timers running and
632 * @cputime_atomic receives updates.
633 * @checking_timer: true when a thread in the group is in the
634 * process of checking for thread group timers.
636 * This structure contains the version of task_cputime, above, that is
637 * used for thread group CPU timer calculations.
639 struct thread_group_cputimer
{
640 struct task_cputime_atomic cputime_atomic
;
645 #include <linux/rwsem.h>
649 * NOTE! "signal_struct" does not have its own
650 * locking, because a shared signal_struct always
651 * implies a shared sighand_struct, so locking
652 * sighand_struct is always a proper superset of
653 * the locking of signal_struct.
655 struct signal_struct
{
659 struct list_head thread_head
;
661 wait_queue_head_t wait_chldexit
; /* for wait4() */
663 /* current thread group signal load-balancing target: */
664 struct task_struct
*curr_target
;
666 /* shared signal handling: */
667 struct sigpending shared_pending
;
669 /* thread group exit support */
672 * - notify group_exit_task when ->count is equal to notify_count
673 * - everyone except group_exit_task is stopped during signal delivery
674 * of fatal signals, group_exit_task processes the signal.
677 struct task_struct
*group_exit_task
;
679 /* thread group stop support, overloads group_exit_code too */
680 int group_stop_count
;
681 unsigned int flags
; /* see SIGNAL_* flags below */
684 * PR_SET_CHILD_SUBREAPER marks a process, like a service
685 * manager, to re-parent orphan (double-forking) child processes
686 * to this process instead of 'init'. The service manager is
687 * able to receive SIGCHLD signals and is able to investigate
688 * the process until it calls wait(). All children of this
689 * process will inherit a flag if they should look for a
690 * child_subreaper process at exit.
692 unsigned int is_child_subreaper
:1;
693 unsigned int has_child_subreaper
:1;
695 #ifdef CONFIG_POSIX_TIMERS
697 /* POSIX.1b Interval Timers */
699 struct list_head posix_timers
;
701 /* ITIMER_REAL timer for the process */
702 struct hrtimer real_timer
;
703 ktime_t it_real_incr
;
706 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
707 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
708 * values are defined to 0 and 1 respectively
710 struct cpu_itimer it
[2];
713 * Thread group totals for process CPU timers.
714 * See thread_group_cputimer(), et al, for details.
716 struct thread_group_cputimer cputimer
;
718 /* Earliest-expiration cache. */
719 struct task_cputime cputime_expires
;
721 struct list_head cpu_timers
[3];
725 struct pid
*leader_pid
;
727 #ifdef CONFIG_NO_HZ_FULL
728 atomic_t tick_dep_mask
;
731 struct pid
*tty_old_pgrp
;
733 /* boolean value for session group leader */
736 struct tty_struct
*tty
; /* NULL if no tty */
738 #ifdef CONFIG_SCHED_AUTOGROUP
739 struct autogroup
*autogroup
;
742 * Cumulative resource counters for dead threads in the group,
743 * and for reaped dead child processes forked by this group.
744 * Live threads maintain their own counters and add to these
745 * in __exit_signal, except for the group leader.
747 seqlock_t stats_lock
;
748 u64 utime
, stime
, cutime
, cstime
;
751 struct prev_cputime prev_cputime
;
752 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
753 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
754 unsigned long inblock
, oublock
, cinblock
, coublock
;
755 unsigned long maxrss
, cmaxrss
;
756 struct task_io_accounting ioac
;
759 * Cumulative ns of schedule CPU time fo dead threads in the
760 * group, not including a zombie group leader, (This only differs
761 * from jiffies_to_ns(utime + stime) if sched_clock uses something
762 * other than jiffies.)
764 unsigned long long sum_sched_runtime
;
767 * We don't bother to synchronize most readers of this at all,
768 * because there is no reader checking a limit that actually needs
769 * to get both rlim_cur and rlim_max atomically, and either one
770 * alone is a single word that can safely be read normally.
771 * getrlimit/setrlimit use task_lock(current->group_leader) to
772 * protect this instead of the siglock, because they really
773 * have no need to disable irqs.
775 struct rlimit rlim
[RLIM_NLIMITS
];
777 #ifdef CONFIG_BSD_PROCESS_ACCT
778 struct pacct_struct pacct
; /* per-process accounting information */
780 #ifdef CONFIG_TASKSTATS
781 struct taskstats
*stats
;
785 struct tty_audit_buf
*tty_audit_buf
;
789 * Thread is the potential origin of an oom condition; kill first on
792 bool oom_flag_origin
;
793 short oom_score_adj
; /* OOM kill score adjustment */
794 short oom_score_adj_min
; /* OOM kill score adjustment min value.
795 * Only settable by CAP_SYS_RESOURCE. */
796 struct mm_struct
*oom_mm
; /* recorded mm when the thread group got
797 * killed by the oom killer */
799 struct mutex cred_guard_mutex
; /* guard against foreign influences on
800 * credential calculations
801 * (notably. ptrace) */
805 * Bits in flags field of signal_struct.
807 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
808 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
809 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
810 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
812 * Pending notifications to parent.
814 #define SIGNAL_CLD_STOPPED 0x00000010
815 #define SIGNAL_CLD_CONTINUED 0x00000020
816 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
818 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
820 #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
821 SIGNAL_STOP_CONTINUED)
823 static inline void signal_set_stop_flags(struct signal_struct
*sig
,
826 WARN_ON(sig
->flags
& (SIGNAL_GROUP_EXIT
|SIGNAL_GROUP_COREDUMP
));
827 sig
->flags
= (sig
->flags
& ~SIGNAL_STOP_MASK
) | flags
;
830 /* If true, all threads except ->group_exit_task have pending SIGKILL */
831 static inline int signal_group_exit(const struct signal_struct
*sig
)
833 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
834 (sig
->group_exit_task
!= NULL
);
838 * Some day this will be a full-fledged user tracking system..
841 atomic_t __count
; /* reference count */
842 atomic_t processes
; /* How many processes does this user have? */
843 atomic_t sigpending
; /* How many pending signals does this user have? */
844 #ifdef CONFIG_FANOTIFY
845 atomic_t fanotify_listeners
;
848 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
850 #ifdef CONFIG_POSIX_MQUEUE
851 /* protected by mq_lock */
852 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
854 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
855 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
856 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
859 struct key
*uid_keyring
; /* UID specific keyring */
860 struct key
*session_keyring
; /* UID's default session keyring */
863 /* Hash table maintenance information */
864 struct hlist_node uidhash_node
;
867 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
868 atomic_long_t locked_vm
;
872 extern int uids_sysfs_init(void);
874 extern struct user_struct
*find_user(kuid_t
);
876 extern struct user_struct root_user
;
877 #define INIT_USER (&root_user)
880 struct backing_dev_info
;
881 struct reclaim_state
;
883 #ifdef CONFIG_SCHED_INFO
885 /* cumulative counters */
886 unsigned long pcount
; /* # of times run on this cpu */
887 unsigned long long run_delay
; /* time spent waiting on a runqueue */
890 unsigned long long last_arrival
,/* when we last ran on a cpu */
891 last_queued
; /* when we were last queued to run */
893 #endif /* CONFIG_SCHED_INFO */
895 #ifdef CONFIG_TASK_DELAY_ACCT
896 struct task_delay_info
{
898 unsigned int flags
; /* Private per-task flags */
900 /* For each stat XXX, add following, aligned appropriately
902 * struct timespec XXX_start, XXX_end;
906 * Atomicity of updates to XXX_delay, XXX_count protected by
907 * single lock above (split into XXX_lock if contention is an issue).
911 * XXX_count is incremented on every XXX operation, the delay
912 * associated with the operation is added to XXX_delay.
913 * XXX_delay contains the accumulated delay time in nanoseconds.
915 u64 blkio_start
; /* Shared by blkio, swapin */
916 u64 blkio_delay
; /* wait for sync block io completion */
917 u64 swapin_delay
; /* wait for swapin block io completion */
918 u32 blkio_count
; /* total count of the number of sync block */
919 /* io operations performed */
920 u32 swapin_count
; /* total count of the number of swapin block */
921 /* io operations performed */
924 u64 freepages_delay
; /* wait for memory reclaim */
925 u32 freepages_count
; /* total count of memory reclaim */
927 #endif /* CONFIG_TASK_DELAY_ACCT */
929 static inline int sched_info_on(void)
931 #ifdef CONFIG_SCHEDSTATS
933 #elif defined(CONFIG_TASK_DELAY_ACCT)
934 extern int delayacct_on
;
941 #ifdef CONFIG_SCHEDSTATS
942 void force_schedstat_enabled(void);
953 * Integer metrics need fixed point arithmetic, e.g., sched/fair
954 * has a few: load, load_avg, util_avg, freq, and capacity.
956 * We define a basic fixed point arithmetic range, and then formalize
957 * all these metrics based on that basic range.
959 # define SCHED_FIXEDPOINT_SHIFT 10
960 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
963 * Increase resolution of cpu_capacity calculations
965 #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
966 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
969 * Wake-queues are lists of tasks with a pending wakeup, whose
970 * callers have already marked the task as woken internally,
971 * and can thus carry on. A common use case is being able to
972 * do the wakeups once the corresponding user lock as been
975 * We hold reference to each task in the list across the wakeup,
976 * thus guaranteeing that the memory is still valid by the time
977 * the actual wakeups are performed in wake_up_q().
979 * One per task suffices, because there's never a need for a task to be
980 * in two wake queues simultaneously; it is forbidden to abandon a task
981 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
982 * already in a wake queue, the wakeup will happen soon and the second
983 * waker can just skip it.
985 * The DEFINE_WAKE_Q macro declares and initializes the list head.
986 * wake_up_q() does NOT reinitialize the list; it's expected to be
987 * called near the end of a function. Otherwise, the list can be
988 * re-initialized for later re-use by wake_q_init().
990 * Note that this can cause spurious wakeups. schedule() callers
991 * must ensure the call is done inside a loop, confirming that the
992 * wakeup condition has in fact occurred.
995 struct wake_q_node
*next
;
999 struct wake_q_node
*first
;
1000 struct wake_q_node
**lastp
;
1003 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1005 #define DEFINE_WAKE_Q(name) \
1006 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1008 static inline void wake_q_init(struct wake_q_head
*head
)
1010 head
->first
= WAKE_Q_TAIL
;
1011 head
->lastp
= &head
->first
;
1014 extern void wake_q_add(struct wake_q_head
*head
,
1015 struct task_struct
*task
);
1016 extern void wake_up_q(struct wake_q_head
*head
);
1019 * sched-domains (multiprocessor balancing) declarations:
1022 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1023 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1024 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1025 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
1026 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
1027 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
1028 #define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
1029 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
1030 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1031 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1032 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1033 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1034 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1035 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1036 #define SD_NUMA 0x4000 /* cross-node balancing */
1038 #ifdef CONFIG_SCHED_SMT
1039 static inline int cpu_smt_flags(void)
1041 return SD_SHARE_CPUCAPACITY
| SD_SHARE_PKG_RESOURCES
;
1045 #ifdef CONFIG_SCHED_MC
1046 static inline int cpu_core_flags(void)
1048 return SD_SHARE_PKG_RESOURCES
;
1053 static inline int cpu_numa_flags(void)
1059 extern int arch_asym_cpu_priority(int cpu
);
1061 struct sched_domain_attr
{
1062 int relax_domain_level
;
1065 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1066 .relax_domain_level = -1, \
1069 extern int sched_domain_level_max
;
1073 struct sched_domain_shared
{
1075 atomic_t nr_busy_cpus
;
1079 struct sched_domain
{
1080 /* These fields must be setup */
1081 struct sched_domain
*parent
; /* top domain must be null terminated */
1082 struct sched_domain
*child
; /* bottom domain must be null terminated */
1083 struct sched_group
*groups
; /* the balancing groups of the domain */
1084 unsigned long min_interval
; /* Minimum balance interval ms */
1085 unsigned long max_interval
; /* Maximum balance interval ms */
1086 unsigned int busy_factor
; /* less balancing by factor if busy */
1087 unsigned int imbalance_pct
; /* No balance until over watermark */
1088 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
1089 unsigned int busy_idx
;
1090 unsigned int idle_idx
;
1091 unsigned int newidle_idx
;
1092 unsigned int wake_idx
;
1093 unsigned int forkexec_idx
;
1094 unsigned int smt_gain
;
1096 int nohz_idle
; /* NOHZ IDLE status */
1097 int flags
; /* See SD_* */
1100 /* Runtime fields. */
1101 unsigned long last_balance
; /* init to jiffies. units in jiffies */
1102 unsigned int balance_interval
; /* initialise to 1. units in ms. */
1103 unsigned int nr_balance_failed
; /* initialise to 0 */
1105 /* idle_balance() stats */
1106 u64 max_newidle_lb_cost
;
1107 unsigned long next_decay_max_lb_cost
;
1109 u64 avg_scan_cost
; /* select_idle_sibling */
1111 #ifdef CONFIG_SCHEDSTATS
1112 /* load_balance() stats */
1113 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
1114 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
1115 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
1116 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
1117 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
1118 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
1119 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
1120 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
1122 /* Active load balancing */
1123 unsigned int alb_count
;
1124 unsigned int alb_failed
;
1125 unsigned int alb_pushed
;
1127 /* SD_BALANCE_EXEC stats */
1128 unsigned int sbe_count
;
1129 unsigned int sbe_balanced
;
1130 unsigned int sbe_pushed
;
1132 /* SD_BALANCE_FORK stats */
1133 unsigned int sbf_count
;
1134 unsigned int sbf_balanced
;
1135 unsigned int sbf_pushed
;
1137 /* try_to_wake_up() stats */
1138 unsigned int ttwu_wake_remote
;
1139 unsigned int ttwu_move_affine
;
1140 unsigned int ttwu_move_balance
;
1142 #ifdef CONFIG_SCHED_DEBUG
1146 void *private; /* used during construction */
1147 struct rcu_head rcu
; /* used during destruction */
1149 struct sched_domain_shared
*shared
;
1151 unsigned int span_weight
;
1153 * Span of all CPUs in this domain.
1155 * NOTE: this field is variable length. (Allocated dynamically
1156 * by attaching extra space to the end of the structure,
1157 * depending on how many CPUs the kernel has booted up with)
1159 unsigned long span
[0];
1162 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1164 return to_cpumask(sd
->span
);
1167 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1168 struct sched_domain_attr
*dattr_new
);
1170 /* Allocate an array of sched domains, for partition_sched_domains(). */
1171 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1172 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1174 bool cpus_share_cache(int this_cpu
, int that_cpu
);
1176 typedef const struct cpumask
*(*sched_domain_mask_f
)(int cpu
);
1177 typedef int (*sched_domain_flags_f
)(void);
1179 #define SDTL_OVERLAP 0x01
1182 struct sched_domain
**__percpu sd
;
1183 struct sched_domain_shared
**__percpu sds
;
1184 struct sched_group
**__percpu sg
;
1185 struct sched_group_capacity
**__percpu sgc
;
1188 struct sched_domain_topology_level
{
1189 sched_domain_mask_f mask
;
1190 sched_domain_flags_f sd_flags
;
1193 struct sd_data data
;
1194 #ifdef CONFIG_SCHED_DEBUG
1199 extern void set_sched_topology(struct sched_domain_topology_level
*tl
);
1200 extern void wake_up_if_idle(int cpu
);
1202 #ifdef CONFIG_SCHED_DEBUG
1203 # define SD_INIT_NAME(type) .name = #type
1205 # define SD_INIT_NAME(type)
1208 #else /* CONFIG_SMP */
1210 struct sched_domain_attr
;
1213 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1214 struct sched_domain_attr
*dattr_new
)
1218 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1223 #endif /* !CONFIG_SMP */
1226 struct io_context
; /* See blkdev.h */
1229 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1230 extern void prefetch_stack(struct task_struct
*t
);
1232 static inline void prefetch_stack(struct task_struct
*t
) { }
1235 struct audit_context
; /* See audit.c */
1237 struct pipe_inode_info
;
1238 struct uts_namespace
;
1240 struct load_weight
{
1241 unsigned long weight
;
1246 * The load_avg/util_avg accumulates an infinite geometric series
1247 * (see __update_load_avg() in kernel/sched/fair.c).
1249 * [load_avg definition]
1251 * load_avg = runnable% * scale_load_down(load)
1253 * where runnable% is the time ratio that a sched_entity is runnable.
1254 * For cfs_rq, it is the aggregated load_avg of all runnable and
1255 * blocked sched_entities.
1257 * load_avg may also take frequency scaling into account:
1259 * load_avg = runnable% * scale_load_down(load) * freq%
1261 * where freq% is the CPU frequency normalized to the highest frequency.
1263 * [util_avg definition]
1265 * util_avg = running% * SCHED_CAPACITY_SCALE
1267 * where running% is the time ratio that a sched_entity is running on
1268 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1269 * and blocked sched_entities.
1271 * util_avg may also factor frequency scaling and CPU capacity scaling:
1273 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1275 * where freq% is the same as above, and capacity% is the CPU capacity
1276 * normalized to the greatest capacity (due to uarch differences, etc).
1278 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1279 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1280 * we therefore scale them to as large a range as necessary. This is for
1281 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1285 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1286 * with the highest load (=88761), always runnable on a single cfs_rq,
1287 * and should not overflow as the number already hits PID_MAX_LIMIT.
1289 * For all other cases (including 32-bit kernels), struct load_weight's
1290 * weight will overflow first before we do, because:
1292 * Max(load_avg) <= Max(load.weight)
1294 * Then it is the load_weight's responsibility to consider overflow
1298 u64 last_update_time
, load_sum
;
1299 u32 util_sum
, period_contrib
;
1300 unsigned long load_avg
, util_avg
;
1303 #ifdef CONFIG_SCHEDSTATS
1304 struct sched_statistics
{
1314 s64 sum_sleep_runtime
;
1321 u64 nr_migrations_cold
;
1322 u64 nr_failed_migrations_affine
;
1323 u64 nr_failed_migrations_running
;
1324 u64 nr_failed_migrations_hot
;
1325 u64 nr_forced_migrations
;
1328 u64 nr_wakeups_sync
;
1329 u64 nr_wakeups_migrate
;
1330 u64 nr_wakeups_local
;
1331 u64 nr_wakeups_remote
;
1332 u64 nr_wakeups_affine
;
1333 u64 nr_wakeups_affine_attempts
;
1334 u64 nr_wakeups_passive
;
1335 u64 nr_wakeups_idle
;
1339 struct sched_entity
{
1340 struct load_weight load
; /* for load-balancing */
1341 struct rb_node run_node
;
1342 struct list_head group_node
;
1346 u64 sum_exec_runtime
;
1348 u64 prev_sum_exec_runtime
;
1352 #ifdef CONFIG_SCHEDSTATS
1353 struct sched_statistics statistics
;
1356 #ifdef CONFIG_FAIR_GROUP_SCHED
1358 struct sched_entity
*parent
;
1359 /* rq on which this entity is (to be) queued: */
1360 struct cfs_rq
*cfs_rq
;
1361 /* rq "owned" by this entity/group: */
1362 struct cfs_rq
*my_q
;
1367 * Per entity load average tracking.
1369 * Put into separate cache line so it does not
1370 * collide with read-mostly values above.
1372 struct sched_avg avg ____cacheline_aligned_in_smp
;
1376 struct sched_rt_entity
{
1377 struct list_head run_list
;
1378 unsigned long timeout
;
1379 unsigned long watchdog_stamp
;
1380 unsigned int time_slice
;
1381 unsigned short on_rq
;
1382 unsigned short on_list
;
1384 struct sched_rt_entity
*back
;
1385 #ifdef CONFIG_RT_GROUP_SCHED
1386 struct sched_rt_entity
*parent
;
1387 /* rq on which this entity is (to be) queued: */
1388 struct rt_rq
*rt_rq
;
1389 /* rq "owned" by this entity/group: */
1394 struct sched_dl_entity
{
1395 struct rb_node rb_node
;
1398 * Original scheduling parameters. Copied here from sched_attr
1399 * during sched_setattr(), they will remain the same until
1400 * the next sched_setattr().
1402 u64 dl_runtime
; /* maximum runtime for each instance */
1403 u64 dl_deadline
; /* relative deadline of each instance */
1404 u64 dl_period
; /* separation of two instances (period) */
1405 u64 dl_bw
; /* dl_runtime / dl_deadline */
1408 * Actual scheduling parameters. Initialized with the values above,
1409 * they are continously updated during task execution. Note that
1410 * the remaining runtime could be < 0 in case we are in overrun.
1412 s64 runtime
; /* remaining runtime for this instance */
1413 u64 deadline
; /* absolute deadline for this instance */
1414 unsigned int flags
; /* specifying the scheduler behaviour */
1419 * @dl_throttled tells if we exhausted the runtime. If so, the
1420 * task has to wait for a replenishment to be performed at the
1421 * next firing of dl_timer.
1423 * @dl_boosted tells if we are boosted due to DI. If so we are
1424 * outside bandwidth enforcement mechanism (but only until we
1425 * exit the critical section);
1427 * @dl_yielded tells if task gave up the cpu before consuming
1428 * all its available runtime during the last job.
1430 int dl_throttled
, dl_boosted
, dl_yielded
;
1433 * Bandwidth enforcement timer. Each -deadline task has its
1434 * own bandwidth to be enforced, thus we need one timer per task.
1436 struct hrtimer dl_timer
;
1444 u8 pad
; /* Otherwise the compiler can store garbage here. */
1446 u32 s
; /* Set of bits. */
1450 enum perf_event_task_context
{
1451 perf_invalid_context
= -1,
1452 perf_hw_context
= 0,
1454 perf_nr_task_contexts
,
1457 /* Track pages that require TLB flushes */
1458 struct tlbflush_unmap_batch
{
1460 * Each bit set is a CPU that potentially has a TLB entry for one of
1461 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1463 struct cpumask cpumask
;
1465 /* True if any bit in cpumask is set */
1466 bool flush_required
;
1469 * If true then the PTE was dirty when unmapped. The entry must be
1470 * flushed before IO is initiated or a stale TLB entry potentially
1471 * allows an update without redirtying the page.
1476 struct task_struct
{
1477 #ifdef CONFIG_THREAD_INFO_IN_TASK
1479 * For reasons of header soup (see current_thread_info()), this
1480 * must be the first element of task_struct.
1482 struct thread_info thread_info
;
1484 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1487 unsigned int flags
; /* per process flags, defined below */
1488 unsigned int ptrace
;
1491 struct llist_node wake_entry
;
1493 #ifdef CONFIG_THREAD_INFO_IN_TASK
1494 unsigned int cpu
; /* current CPU */
1496 unsigned int wakee_flips
;
1497 unsigned long wakee_flip_decay_ts
;
1498 struct task_struct
*last_wakee
;
1504 int prio
, static_prio
, normal_prio
;
1505 unsigned int rt_priority
;
1506 const struct sched_class
*sched_class
;
1507 struct sched_entity se
;
1508 struct sched_rt_entity rt
;
1509 #ifdef CONFIG_CGROUP_SCHED
1510 struct task_group
*sched_task_group
;
1512 struct sched_dl_entity dl
;
1514 #ifdef CONFIG_PREEMPT_NOTIFIERS
1515 /* list of struct preempt_notifier: */
1516 struct hlist_head preempt_notifiers
;
1519 #ifdef CONFIG_BLK_DEV_IO_TRACE
1520 unsigned int btrace_seq
;
1523 unsigned int policy
;
1524 int nr_cpus_allowed
;
1525 cpumask_t cpus_allowed
;
1527 #ifdef CONFIG_PREEMPT_RCU
1528 int rcu_read_lock_nesting
;
1529 union rcu_special rcu_read_unlock_special
;
1530 struct list_head rcu_node_entry
;
1531 struct rcu_node
*rcu_blocked_node
;
1532 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1533 #ifdef CONFIG_TASKS_RCU
1534 unsigned long rcu_tasks_nvcsw
;
1535 bool rcu_tasks_holdout
;
1536 struct list_head rcu_tasks_holdout_list
;
1537 int rcu_tasks_idle_cpu
;
1538 #endif /* #ifdef CONFIG_TASKS_RCU */
1540 #ifdef CONFIG_SCHED_INFO
1541 struct sched_info sched_info
;
1544 struct list_head tasks
;
1546 struct plist_node pushable_tasks
;
1547 struct rb_node pushable_dl_tasks
;
1550 struct mm_struct
*mm
, *active_mm
;
1552 /* Per-thread vma caching: */
1553 struct vmacache vmacache
;
1555 #if defined(SPLIT_RSS_COUNTING)
1556 struct task_rss_stat rss_stat
;
1560 int exit_code
, exit_signal
;
1561 int pdeath_signal
; /* The signal sent when the parent dies */
1562 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1564 /* Used for emulating ABI behavior of previous Linux versions */
1565 unsigned int personality
;
1567 /* scheduler bits, serialized by scheduler locks */
1568 unsigned sched_reset_on_fork
:1;
1569 unsigned sched_contributes_to_load
:1;
1570 unsigned sched_migrated
:1;
1571 unsigned sched_remote_wakeup
:1;
1572 unsigned :0; /* force alignment to the next boundary */
1574 /* unserialized, strictly 'current' */
1575 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1576 unsigned in_iowait
:1;
1577 #if !defined(TIF_RESTORE_SIGMASK)
1578 unsigned restore_sigmask
:1;
1581 unsigned memcg_may_oom
:1;
1583 unsigned memcg_kmem_skip_account
:1;
1586 #ifdef CONFIG_COMPAT_BRK
1587 unsigned brk_randomized
:1;
1590 unsigned long atomic_flags
; /* Flags needing atomic access. */
1592 struct restart_block restart_block
;
1597 #ifdef CONFIG_CC_STACKPROTECTOR
1598 /* Canary value for the -fstack-protector gcc feature */
1599 unsigned long stack_canary
;
1602 * pointers to (original) parent process, youngest child, younger sibling,
1603 * older sibling, respectively. (p->father can be replaced with
1604 * p->real_parent->pid)
1606 struct task_struct __rcu
*real_parent
; /* real parent process */
1607 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1609 * children/sibling forms the list of my natural children
1611 struct list_head children
; /* list of my children */
1612 struct list_head sibling
; /* linkage in my parent's children list */
1613 struct task_struct
*group_leader
; /* threadgroup leader */
1616 * ptraced is the list of tasks this task is using ptrace on.
1617 * This includes both natural children and PTRACE_ATTACH targets.
1618 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1620 struct list_head ptraced
;
1621 struct list_head ptrace_entry
;
1623 /* PID/PID hash table linkage. */
1624 struct pid_link pids
[PIDTYPE_MAX
];
1625 struct list_head thread_group
;
1626 struct list_head thread_node
;
1628 struct completion
*vfork_done
; /* for vfork() */
1629 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1630 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1633 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
1634 u64 utimescaled
, stimescaled
;
1637 struct prev_cputime prev_cputime
;
1638 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1639 seqcount_t vtime_seqcount
;
1640 unsigned long long vtime_snap
;
1642 /* Task is sleeping or running in a CPU with VTIME inactive */
1644 /* Task runs in userspace in a CPU with VTIME active */
1646 /* Task runs in kernelspace in a CPU with VTIME active */
1648 } vtime_snap_whence
;
1651 #ifdef CONFIG_NO_HZ_FULL
1652 atomic_t tick_dep_mask
;
1654 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1655 u64 start_time
; /* monotonic time in nsec */
1656 u64 real_start_time
; /* boot based time in nsec */
1657 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1658 unsigned long min_flt
, maj_flt
;
1660 #ifdef CONFIG_POSIX_TIMERS
1661 struct task_cputime cputime_expires
;
1662 struct list_head cpu_timers
[3];
1665 /* process credentials */
1666 const struct cred __rcu
*ptracer_cred
; /* Tracer's credentials at attach */
1667 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1668 * credentials (COW) */
1669 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1670 * credentials (COW) */
1671 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1672 - access with [gs]et_task_comm (which lock
1673 it with task_lock())
1674 - initialized normally by setup_new_exec */
1675 /* file system info */
1676 struct nameidata
*nameidata
;
1677 #ifdef CONFIG_SYSVIPC
1679 struct sysv_sem sysvsem
;
1680 struct sysv_shm sysvshm
;
1682 #ifdef CONFIG_DETECT_HUNG_TASK
1683 /* hung task detection */
1684 unsigned long last_switch_count
;
1686 /* filesystem information */
1687 struct fs_struct
*fs
;
1688 /* open file information */
1689 struct files_struct
*files
;
1691 struct nsproxy
*nsproxy
;
1692 /* signal handlers */
1693 struct signal_struct
*signal
;
1694 struct sighand_struct
*sighand
;
1696 sigset_t blocked
, real_blocked
;
1697 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1698 struct sigpending pending
;
1700 unsigned long sas_ss_sp
;
1702 unsigned sas_ss_flags
;
1704 struct callback_head
*task_works
;
1706 struct audit_context
*audit_context
;
1707 #ifdef CONFIG_AUDITSYSCALL
1709 unsigned int sessionid
;
1711 struct seccomp seccomp
;
1713 /* Thread group tracking */
1716 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1718 spinlock_t alloc_lock
;
1720 /* Protection of the PI data structures: */
1721 raw_spinlock_t pi_lock
;
1723 struct wake_q_node wake_q
;
1725 #ifdef CONFIG_RT_MUTEXES
1726 /* PI waiters blocked on a rt_mutex held by this task */
1727 struct rb_root pi_waiters
;
1728 struct rb_node
*pi_waiters_leftmost
;
1729 /* Deadlock detection and priority inheritance handling */
1730 struct rt_mutex_waiter
*pi_blocked_on
;
1733 #ifdef CONFIG_DEBUG_MUTEXES
1734 /* mutex deadlock detection */
1735 struct mutex_waiter
*blocked_on
;
1737 #ifdef CONFIG_TRACE_IRQFLAGS
1738 unsigned int irq_events
;
1739 unsigned long hardirq_enable_ip
;
1740 unsigned long hardirq_disable_ip
;
1741 unsigned int hardirq_enable_event
;
1742 unsigned int hardirq_disable_event
;
1743 int hardirqs_enabled
;
1744 int hardirq_context
;
1745 unsigned long softirq_disable_ip
;
1746 unsigned long softirq_enable_ip
;
1747 unsigned int softirq_disable_event
;
1748 unsigned int softirq_enable_event
;
1749 int softirqs_enabled
;
1750 int softirq_context
;
1752 #ifdef CONFIG_LOCKDEP
1753 # define MAX_LOCK_DEPTH 48UL
1756 unsigned int lockdep_recursion
;
1757 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1758 gfp_t lockdep_reclaim_gfp
;
1761 unsigned int in_ubsan
;
1764 /* journalling filesystem info */
1767 /* stacked block device info */
1768 struct bio_list
*bio_list
;
1771 /* stack plugging */
1772 struct blk_plug
*plug
;
1776 struct reclaim_state
*reclaim_state
;
1778 struct backing_dev_info
*backing_dev_info
;
1780 struct io_context
*io_context
;
1782 unsigned long ptrace_message
;
1783 siginfo_t
*last_siginfo
; /* For ptrace use. */
1784 struct task_io_accounting ioac
;
1785 #if defined(CONFIG_TASK_XACCT)
1786 u64 acct_rss_mem1
; /* accumulated rss usage */
1787 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1788 u64 acct_timexpd
; /* stime + utime since last update */
1790 #ifdef CONFIG_CPUSETS
1791 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1792 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1793 int cpuset_mem_spread_rotor
;
1794 int cpuset_slab_spread_rotor
;
1796 #ifdef CONFIG_CGROUPS
1797 /* Control Group info protected by css_set_lock */
1798 struct css_set __rcu
*cgroups
;
1799 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1800 struct list_head cg_list
;
1802 #ifdef CONFIG_INTEL_RDT_A
1806 struct robust_list_head __user
*robust_list
;
1807 #ifdef CONFIG_COMPAT
1808 struct compat_robust_list_head __user
*compat_robust_list
;
1810 struct list_head pi_state_list
;
1811 struct futex_pi_state
*pi_state_cache
;
1813 #ifdef CONFIG_PERF_EVENTS
1814 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1815 struct mutex perf_event_mutex
;
1816 struct list_head perf_event_list
;
1818 #ifdef CONFIG_DEBUG_PREEMPT
1819 unsigned long preempt_disable_ip
;
1822 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1824 short pref_node_fork
;
1826 #ifdef CONFIG_NUMA_BALANCING
1828 unsigned int numa_scan_period
;
1829 unsigned int numa_scan_period_max
;
1830 int numa_preferred_nid
;
1831 unsigned long numa_migrate_retry
;
1832 u64 node_stamp
; /* migration stamp */
1833 u64 last_task_numa_placement
;
1834 u64 last_sum_exec_runtime
;
1835 struct callback_head numa_work
;
1837 struct list_head numa_entry
;
1838 struct numa_group
*numa_group
;
1841 * numa_faults is an array split into four regions:
1842 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1843 * in this precise order.
1845 * faults_memory: Exponential decaying average of faults on a per-node
1846 * basis. Scheduling placement decisions are made based on these
1847 * counts. The values remain static for the duration of a PTE scan.
1848 * faults_cpu: Track the nodes the process was running on when a NUMA
1849 * hinting fault was incurred.
1850 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1851 * during the current scan window. When the scan completes, the counts
1852 * in faults_memory and faults_cpu decay and these values are copied.
1854 unsigned long *numa_faults
;
1855 unsigned long total_numa_faults
;
1858 * numa_faults_locality tracks if faults recorded during the last
1859 * scan window were remote/local or failed to migrate. The task scan
1860 * period is adapted based on the locality of the faults with different
1861 * weights depending on whether they were shared or private faults
1863 unsigned long numa_faults_locality
[3];
1865 unsigned long numa_pages_migrated
;
1866 #endif /* CONFIG_NUMA_BALANCING */
1868 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1869 struct tlbflush_unmap_batch tlb_ubc
;
1872 struct rcu_head rcu
;
1875 * cache last used pipe for splice
1877 struct pipe_inode_info
*splice_pipe
;
1879 struct page_frag task_frag
;
1881 #ifdef CONFIG_TASK_DELAY_ACCT
1882 struct task_delay_info
*delays
;
1884 #ifdef CONFIG_FAULT_INJECTION
1888 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1889 * balance_dirty_pages() for some dirty throttling pause
1892 int nr_dirtied_pause
;
1893 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1895 #ifdef CONFIG_LATENCYTOP
1896 int latency_record_count
;
1897 struct latency_record latency_record
[LT_SAVECOUNT
];
1900 * time slack values; these are used to round up poll() and
1901 * select() etc timeout values. These are in nanoseconds.
1904 u64 default_timer_slack_ns
;
1907 unsigned int kasan_depth
;
1909 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1910 /* Index of current stored address in ret_stack */
1912 /* Stack of return addresses for return function tracing */
1913 struct ftrace_ret_stack
*ret_stack
;
1914 /* time stamp for last schedule */
1915 unsigned long long ftrace_timestamp
;
1917 * Number of functions that haven't been traced
1918 * because of depth overrun.
1920 atomic_t trace_overrun
;
1921 /* Pause for the tracing */
1922 atomic_t tracing_graph_pause
;
1924 #ifdef CONFIG_TRACING
1925 /* state flags for use by tracers */
1926 unsigned long trace
;
1927 /* bitmask and counter of trace recursion */
1928 unsigned long trace_recursion
;
1929 #endif /* CONFIG_TRACING */
1931 /* Coverage collection mode enabled for this task (0 if disabled). */
1932 enum kcov_mode kcov_mode
;
1933 /* Size of the kcov_area. */
1935 /* Buffer for coverage collection. */
1937 /* kcov desciptor wired with this task or NULL. */
1941 struct mem_cgroup
*memcg_in_oom
;
1942 gfp_t memcg_oom_gfp_mask
;
1943 int memcg_oom_order
;
1945 /* number of pages to reclaim on returning to userland */
1946 unsigned int memcg_nr_pages_over_high
;
1948 #ifdef CONFIG_UPROBES
1949 struct uprobe_task
*utask
;
1951 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1952 unsigned int sequential_io
;
1953 unsigned int sequential_io_avg
;
1955 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1956 unsigned long task_state_change
;
1958 int pagefault_disabled
;
1960 struct task_struct
*oom_reaper_list
;
1962 #ifdef CONFIG_VMAP_STACK
1963 struct vm_struct
*stack_vm_area
;
1965 #ifdef CONFIG_THREAD_INFO_IN_TASK
1966 /* A live task holds one reference. */
1967 atomic_t stack_refcount
;
1969 /* CPU-specific state of this task */
1970 struct thread_struct thread
;
1972 * WARNING: on x86, 'thread_struct' contains a variable-sized
1973 * structure. It *MUST* be at the end of 'task_struct'.
1975 * Do not put anything below here!
1979 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1980 extern int arch_task_struct_size __read_mostly
;
1982 # define arch_task_struct_size (sizeof(struct task_struct))
1985 #ifdef CONFIG_VMAP_STACK
1986 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1988 return t
->stack_vm_area
;
1991 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1997 #define TNF_MIGRATED 0x01
1998 #define TNF_NO_GROUP 0x02
1999 #define TNF_SHARED 0x04
2000 #define TNF_FAULT_LOCAL 0x08
2001 #define TNF_MIGRATE_FAIL 0x10
2003 static inline bool in_vfork(struct task_struct
*tsk
)
2008 * need RCU to access ->real_parent if CLONE_VM was used along with
2011 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
2014 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
2015 * ->real_parent is not necessarily the task doing vfork(), so in
2016 * theory we can't rely on task_lock() if we want to dereference it.
2018 * And in this case we can't trust the real_parent->mm == tsk->mm
2019 * check, it can be false negative. But we do not care, if init or
2020 * another oom-unkillable task does this it should blame itself.
2023 ret
= tsk
->vfork_done
&& tsk
->real_parent
->mm
== tsk
->mm
;
2029 #ifdef CONFIG_NUMA_BALANCING
2030 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
2031 extern pid_t
task_numa_group_id(struct task_struct
*p
);
2032 extern void set_numabalancing_state(bool enabled
);
2033 extern void task_numa_free(struct task_struct
*p
);
2034 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
2035 int src_nid
, int dst_cpu
);
2037 static inline void task_numa_fault(int last_node
, int node
, int pages
,
2041 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
2045 static inline void set_numabalancing_state(bool enabled
)
2048 static inline void task_numa_free(struct task_struct
*p
)
2051 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
2052 struct page
*page
, int src_nid
, int dst_cpu
)
2058 static inline struct pid
*task_pid(struct task_struct
*task
)
2060 return task
->pids
[PIDTYPE_PID
].pid
;
2063 static inline struct pid
*task_tgid(struct task_struct
*task
)
2065 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
2069 * Without tasklist or rcu lock it is not safe to dereference
2070 * the result of task_pgrp/task_session even if task == current,
2071 * we can race with another thread doing sys_setsid/sys_setpgid.
2073 static inline struct pid
*task_pgrp(struct task_struct
*task
)
2075 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
2078 static inline struct pid
*task_session(struct task_struct
*task
)
2080 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
2083 struct pid_namespace
;
2086 * the helpers to get the task's different pids as they are seen
2087 * from various namespaces
2089 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
2090 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2092 * task_xid_nr_ns() : id seen from the ns specified;
2094 * set_task_vxid() : assigns a virtual id to a task;
2096 * see also pid_nr() etc in include/linux/pid.h
2098 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
2099 struct pid_namespace
*ns
);
2101 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
2106 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
2107 struct pid_namespace
*ns
)
2109 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
2112 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
2114 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
2118 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
2123 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
2125 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
2127 return pid_vnr(task_tgid(tsk
));
2131 static inline int pid_alive(const struct task_struct
*p
);
2132 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
2138 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
2144 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
2146 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
2149 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
2150 struct pid_namespace
*ns
)
2152 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
2155 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
2157 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
2161 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
2162 struct pid_namespace
*ns
)
2164 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
2167 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
2169 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
2172 /* obsolete, do not use */
2173 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
2175 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
2179 * pid_alive - check that a task structure is not stale
2180 * @p: Task structure to be checked.
2182 * Test if a process is not yet dead (at most zombie state)
2183 * If pid_alive fails, then pointers within the task structure
2184 * can be stale and must not be dereferenced.
2186 * Return: 1 if the process is alive. 0 otherwise.
2188 static inline int pid_alive(const struct task_struct
*p
)
2190 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
2194 * is_global_init - check if a task structure is init. Since init
2195 * is free to have sub-threads we need to check tgid.
2196 * @tsk: Task structure to be checked.
2198 * Check if a task structure is the first user space task the kernel created.
2200 * Return: 1 if the task structure is init. 0 otherwise.
2202 static inline int is_global_init(struct task_struct
*tsk
)
2204 return task_tgid_nr(tsk
) == 1;
2207 extern struct pid
*cad_pid
;
2209 extern void free_task(struct task_struct
*tsk
);
2210 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2212 extern void __put_task_struct(struct task_struct
*t
);
2214 static inline void put_task_struct(struct task_struct
*t
)
2216 if (atomic_dec_and_test(&t
->usage
))
2217 __put_task_struct(t
);
2220 struct task_struct
*task_rcu_dereference(struct task_struct
**ptask
);
2221 struct task_struct
*try_get_task_struct(struct task_struct
**ptask
);
2223 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2224 extern void task_cputime(struct task_struct
*t
,
2225 u64
*utime
, u64
*stime
);
2226 extern u64
task_gtime(struct task_struct
*t
);
2228 static inline void task_cputime(struct task_struct
*t
,
2229 u64
*utime
, u64
*stime
)
2235 static inline u64
task_gtime(struct task_struct
*t
)
2241 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
2242 static inline void task_cputime_scaled(struct task_struct
*t
,
2246 *utimescaled
= t
->utimescaled
;
2247 *stimescaled
= t
->stimescaled
;
2250 static inline void task_cputime_scaled(struct task_struct
*t
,
2254 task_cputime(t
, utimescaled
, stimescaled
);
2258 extern void task_cputime_adjusted(struct task_struct
*p
, u64
*ut
, u64
*st
);
2259 extern void thread_group_cputime_adjusted(struct task_struct
*p
, u64
*ut
, u64
*st
);
2264 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
2265 #define PF_EXITING 0x00000004 /* getting shut down */
2266 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2267 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2268 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2269 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2270 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2271 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2272 #define PF_DUMPCORE 0x00000200 /* dumped core */
2273 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2274 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2275 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2276 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2277 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2278 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2279 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2280 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2281 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2282 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2283 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2284 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2285 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2286 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2287 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2288 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2289 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2290 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2291 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2294 * Only the _current_ task can read/write to tsk->flags, but other
2295 * tasks can access tsk->flags in readonly mode for example
2296 * with tsk_used_math (like during threaded core dumping).
2297 * There is however an exception to this rule during ptrace
2298 * or during fork: the ptracer task is allowed to write to the
2299 * child->flags of its traced child (same goes for fork, the parent
2300 * can write to the child->flags), because we're guaranteed the
2301 * child is not running and in turn not changing child->flags
2302 * at the same time the parent does it.
2304 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2305 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2306 #define clear_used_math() clear_stopped_child_used_math(current)
2307 #define set_used_math() set_stopped_child_used_math(current)
2308 #define conditional_stopped_child_used_math(condition, child) \
2309 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2310 #define conditional_used_math(condition) \
2311 conditional_stopped_child_used_math(condition, current)
2312 #define copy_to_stopped_child_used_math(child) \
2313 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2314 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2315 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2316 #define used_math() tsk_used_math(current)
2318 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2319 * __GFP_FS is also cleared as it implies __GFP_IO.
2321 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
2323 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
2324 flags
&= ~(__GFP_IO
| __GFP_FS
);
2328 static inline unsigned int memalloc_noio_save(void)
2330 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
2331 current
->flags
|= PF_MEMALLOC_NOIO
;
2335 static inline void memalloc_noio_restore(unsigned int flags
)
2337 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
2340 /* Per-process atomic flags. */
2341 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2342 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2343 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2344 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
2347 #define TASK_PFA_TEST(name, func) \
2348 static inline bool task_##func(struct task_struct *p) \
2349 { return test_bit(PFA_##name, &p->atomic_flags); }
2350 #define TASK_PFA_SET(name, func) \
2351 static inline void task_set_##func(struct task_struct *p) \
2352 { set_bit(PFA_##name, &p->atomic_flags); }
2353 #define TASK_PFA_CLEAR(name, func) \
2354 static inline void task_clear_##func(struct task_struct *p) \
2355 { clear_bit(PFA_##name, &p->atomic_flags); }
2357 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
2358 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
2360 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
2361 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
2362 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2364 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2365 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2366 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2368 TASK_PFA_TEST(LMK_WAITING
, lmk_waiting
)
2369 TASK_PFA_SET(LMK_WAITING
, lmk_waiting
)
2372 * task->jobctl flags
2374 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2376 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2377 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2378 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2379 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2380 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2381 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2382 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2384 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2385 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2386 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2387 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2388 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2389 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2390 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2392 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2393 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2395 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2396 unsigned long mask
);
2397 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2398 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2399 unsigned long mask
);
2401 static inline void rcu_copy_process(struct task_struct
*p
)
2403 #ifdef CONFIG_PREEMPT_RCU
2404 p
->rcu_read_lock_nesting
= 0;
2405 p
->rcu_read_unlock_special
.s
= 0;
2406 p
->rcu_blocked_node
= NULL
;
2407 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2408 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2409 #ifdef CONFIG_TASKS_RCU
2410 p
->rcu_tasks_holdout
= false;
2411 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2412 p
->rcu_tasks_idle_cpu
= -1;
2413 #endif /* #ifdef CONFIG_TASKS_RCU */
2416 static inline void tsk_restore_flags(struct task_struct
*task
,
2417 unsigned long orig_flags
, unsigned long flags
)
2419 task
->flags
&= ~flags
;
2420 task
->flags
|= orig_flags
& flags
;
2423 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2424 const struct cpumask
*trial
);
2425 extern int task_can_attach(struct task_struct
*p
,
2426 const struct cpumask
*cs_cpus_allowed
);
2428 extern void do_set_cpus_allowed(struct task_struct
*p
,
2429 const struct cpumask
*new_mask
);
2431 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2432 const struct cpumask
*new_mask
);
2434 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2435 const struct cpumask
*new_mask
)
2438 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2439 const struct cpumask
*new_mask
)
2441 if (!cpumask_test_cpu(0, new_mask
))
2447 #ifdef CONFIG_NO_HZ_COMMON
2448 void calc_load_enter_idle(void);
2449 void calc_load_exit_idle(void);
2451 static inline void calc_load_enter_idle(void) { }
2452 static inline void calc_load_exit_idle(void) { }
2453 #endif /* CONFIG_NO_HZ_COMMON */
2455 #ifndef cpu_relax_yield
2456 #define cpu_relax_yield() cpu_relax()
2460 * Do not use outside of architecture code which knows its limitations.
2462 * sched_clock() has no promise of monotonicity or bounded drift between
2463 * CPUs, use (which you should not) requires disabling IRQs.
2465 * Please use one of the three interfaces below.
2467 extern unsigned long long notrace
sched_clock(void);
2469 * See the comment in kernel/sched/clock.c
2471 extern u64
running_clock(void);
2472 extern u64
sched_clock_cpu(int cpu
);
2475 extern void sched_clock_init(void);
2477 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2478 static inline void sched_clock_init_late(void)
2482 static inline void sched_clock_tick(void)
2486 static inline void clear_sched_clock_stable(void)
2490 static inline void sched_clock_idle_sleep_event(void)
2494 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
2498 static inline u64
cpu_clock(int cpu
)
2500 return sched_clock();
2503 static inline u64
local_clock(void)
2505 return sched_clock();
2508 extern void sched_clock_init_late(void);
2510 * Architectures can set this to 1 if they have specified
2511 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2512 * but then during bootup it turns out that sched_clock()
2513 * is reliable after all:
2515 extern int sched_clock_stable(void);
2516 extern void clear_sched_clock_stable(void);
2518 extern void sched_clock_tick(void);
2519 extern void sched_clock_idle_sleep_event(void);
2520 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2523 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2524 * time source that is monotonic per cpu argument and has bounded drift
2527 * ######################### BIG FAT WARNING ##########################
2528 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2529 * # go backwards !! #
2530 * ####################################################################
2532 static inline u64
cpu_clock(int cpu
)
2534 return sched_clock_cpu(cpu
);
2537 static inline u64
local_clock(void)
2539 return sched_clock_cpu(raw_smp_processor_id());
2543 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2545 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2546 * The reason for this explicit opt-in is not to have perf penalty with
2547 * slow sched_clocks.
2549 extern void enable_sched_clock_irqtime(void);
2550 extern void disable_sched_clock_irqtime(void);
2552 static inline void enable_sched_clock_irqtime(void) {}
2553 static inline void disable_sched_clock_irqtime(void) {}
2556 extern unsigned long long
2557 task_sched_runtime(struct task_struct
*task
);
2559 /* sched_exec is called by processes performing an exec */
2561 extern void sched_exec(void);
2563 #define sched_exec() {}
2566 extern void sched_clock_idle_sleep_event(void);
2567 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2569 #ifdef CONFIG_HOTPLUG_CPU
2570 extern void idle_task_exit(void);
2572 static inline void idle_task_exit(void) {}
2575 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2576 extern void wake_up_nohz_cpu(int cpu
);
2578 static inline void wake_up_nohz_cpu(int cpu
) { }
2581 #ifdef CONFIG_NO_HZ_FULL
2582 extern u64
scheduler_tick_max_deferment(void);
2585 #ifdef CONFIG_SCHED_AUTOGROUP
2586 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2587 extern void sched_autogroup_detach(struct task_struct
*p
);
2588 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2589 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2590 extern void sched_autogroup_exit_task(struct task_struct
*p
);
2591 #ifdef CONFIG_PROC_FS
2592 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2593 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2596 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2597 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2598 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2599 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2600 static inline void sched_autogroup_exit_task(struct task_struct
*p
) { }
2603 extern int yield_to(struct task_struct
*p
, bool preempt
);
2604 extern void set_user_nice(struct task_struct
*p
, long nice
);
2605 extern int task_prio(const struct task_struct
*p
);
2607 * task_nice - return the nice value of a given task.
2608 * @p: the task in question.
2610 * Return: The nice value [ -20 ... 0 ... 19 ].
2612 static inline int task_nice(const struct task_struct
*p
)
2614 return PRIO_TO_NICE((p
)->static_prio
);
2616 extern int can_nice(const struct task_struct
*p
, const int nice
);
2617 extern int task_curr(const struct task_struct
*p
);
2618 extern int idle_cpu(int cpu
);
2619 extern int sched_setscheduler(struct task_struct
*, int,
2620 const struct sched_param
*);
2621 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2622 const struct sched_param
*);
2623 extern int sched_setattr(struct task_struct
*,
2624 const struct sched_attr
*);
2625 extern struct task_struct
*idle_task(int cpu
);
2627 * is_idle_task - is the specified task an idle task?
2628 * @p: the task in question.
2630 * Return: 1 if @p is an idle task. 0 otherwise.
2632 static inline bool is_idle_task(const struct task_struct
*p
)
2634 return !!(p
->flags
& PF_IDLE
);
2636 extern struct task_struct
*curr_task(int cpu
);
2637 extern void ia64_set_curr_task(int cpu
, struct task_struct
*p
);
2641 union thread_union
{
2642 #ifndef CONFIG_THREAD_INFO_IN_TASK
2643 struct thread_info thread_info
;
2645 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2648 #ifndef __HAVE_ARCH_KSTACK_END
2649 static inline int kstack_end(void *addr
)
2651 /* Reliable end of stack detection:
2652 * Some APM bios versions misalign the stack
2654 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2658 extern union thread_union init_thread_union
;
2659 extern struct task_struct init_task
;
2661 extern struct mm_struct init_mm
;
2663 extern struct pid_namespace init_pid_ns
;
2666 * find a task by one of its numerical ids
2668 * find_task_by_pid_ns():
2669 * finds a task by its pid in the specified namespace
2670 * find_task_by_vpid():
2671 * finds a task by its virtual pid
2673 * see also find_vpid() etc in include/linux/pid.h
2676 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2677 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2678 struct pid_namespace
*ns
);
2680 /* per-UID process charging. */
2681 extern struct user_struct
* alloc_uid(kuid_t
);
2682 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2684 atomic_inc(&u
->__count
);
2687 extern void free_uid(struct user_struct
*);
2689 #include <asm/current.h>
2691 extern void xtime_update(unsigned long ticks
);
2693 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2694 extern int wake_up_process(struct task_struct
*tsk
);
2695 extern void wake_up_new_task(struct task_struct
*tsk
);
2697 extern void kick_process(struct task_struct
*tsk
);
2699 static inline void kick_process(struct task_struct
*tsk
) { }
2701 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2702 extern void sched_dead(struct task_struct
*p
);
2704 extern void proc_caches_init(void);
2705 extern void flush_signals(struct task_struct
*);
2706 extern void ignore_signals(struct task_struct
*);
2707 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2708 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2710 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2712 struct task_struct
*tsk
= current
;
2716 spin_lock_irq(&tsk
->sighand
->siglock
);
2717 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2718 spin_unlock_irq(&tsk
->sighand
->siglock
);
2723 static inline void kernel_signal_stop(void)
2725 spin_lock_irq(¤t
->sighand
->siglock
);
2726 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2727 __set_current_state(TASK_STOPPED
);
2728 spin_unlock_irq(¤t
->sighand
->siglock
);
2733 extern void release_task(struct task_struct
* p
);
2734 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2735 extern int force_sigsegv(int, struct task_struct
*);
2736 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2737 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2738 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2739 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2740 const struct cred
*, u32
);
2741 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2742 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2743 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2744 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2745 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2746 extern void force_sig(int, struct task_struct
*);
2747 extern int send_sig(int, struct task_struct
*, int);
2748 extern int zap_other_threads(struct task_struct
*p
);
2749 extern struct sigqueue
*sigqueue_alloc(void);
2750 extern void sigqueue_free(struct sigqueue
*);
2751 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2752 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2754 #ifdef TIF_RESTORE_SIGMASK
2756 * Legacy restore_sigmask accessors. These are inefficient on
2757 * SMP architectures because they require atomic operations.
2761 * set_restore_sigmask() - make sure saved_sigmask processing gets done
2763 * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2764 * will run before returning to user mode, to process the flag. For
2765 * all callers, TIF_SIGPENDING is already set or it's no harm to set
2766 * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
2767 * arch code will notice on return to user mode, in case those bits
2768 * are scarce. We set TIF_SIGPENDING here to ensure that the arch
2769 * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2771 static inline void set_restore_sigmask(void)
2773 set_thread_flag(TIF_RESTORE_SIGMASK
);
2774 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2776 static inline void clear_restore_sigmask(void)
2778 clear_thread_flag(TIF_RESTORE_SIGMASK
);
2780 static inline bool test_restore_sigmask(void)
2782 return test_thread_flag(TIF_RESTORE_SIGMASK
);
2784 static inline bool test_and_clear_restore_sigmask(void)
2786 return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK
);
2789 #else /* TIF_RESTORE_SIGMASK */
2791 /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2792 static inline void set_restore_sigmask(void)
2794 current
->restore_sigmask
= true;
2795 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2797 static inline void clear_restore_sigmask(void)
2799 current
->restore_sigmask
= false;
2801 static inline bool test_restore_sigmask(void)
2803 return current
->restore_sigmask
;
2805 static inline bool test_and_clear_restore_sigmask(void)
2807 if (!current
->restore_sigmask
)
2809 current
->restore_sigmask
= false;
2814 static inline void restore_saved_sigmask(void)
2816 if (test_and_clear_restore_sigmask())
2817 __set_current_blocked(¤t
->saved_sigmask
);
2820 static inline sigset_t
*sigmask_to_save(void)
2822 sigset_t
*res
= ¤t
->blocked
;
2823 if (unlikely(test_restore_sigmask()))
2824 res
= ¤t
->saved_sigmask
;
2828 static inline int kill_cad_pid(int sig
, int priv
)
2830 return kill_pid(cad_pid
, sig
, priv
);
2833 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2834 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2835 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2836 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2839 * True if we are on the alternate signal stack.
2841 static inline int on_sig_stack(unsigned long sp
)
2844 * If the signal stack is SS_AUTODISARM then, by construction, we
2845 * can't be on the signal stack unless user code deliberately set
2846 * SS_AUTODISARM when we were already on it.
2848 * This improves reliability: if user state gets corrupted such that
2849 * the stack pointer points very close to the end of the signal stack,
2850 * then this check will enable the signal to be handled anyway.
2852 if (current
->sas_ss_flags
& SS_AUTODISARM
)
2855 #ifdef CONFIG_STACK_GROWSUP
2856 return sp
>= current
->sas_ss_sp
&&
2857 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2859 return sp
> current
->sas_ss_sp
&&
2860 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2864 static inline int sas_ss_flags(unsigned long sp
)
2866 if (!current
->sas_ss_size
)
2869 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2872 static inline void sas_ss_reset(struct task_struct
*p
)
2876 p
->sas_ss_flags
= SS_DISABLE
;
2879 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2881 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2882 #ifdef CONFIG_STACK_GROWSUP
2883 return current
->sas_ss_sp
;
2885 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2891 * Routines for handling mm_structs
2893 extern struct mm_struct
* mm_alloc(void);
2896 * mmgrab() - Pin a &struct mm_struct.
2897 * @mm: The &struct mm_struct to pin.
2899 * Make sure that @mm will not get freed even after the owning task
2900 * exits. This doesn't guarantee that the associated address space
2901 * will still exist later on and mmget_not_zero() has to be used before
2904 * This is a preferred way to to pin @mm for a longer/unbounded amount
2907 * Use mmdrop() to release the reference acquired by mmgrab().
2909 * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
2910 * of &mm_struct.mm_count vs &mm_struct.mm_users.
2912 static inline void mmgrab(struct mm_struct
*mm
)
2914 atomic_inc(&mm
->mm_count
);
2917 /* mmdrop drops the mm and the page tables */
2918 extern void __mmdrop(struct mm_struct
*);
2919 static inline void mmdrop(struct mm_struct
*mm
)
2921 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2925 static inline void mmdrop_async_fn(struct work_struct
*work
)
2927 struct mm_struct
*mm
= container_of(work
, struct mm_struct
, async_put_work
);
2931 static inline void mmdrop_async(struct mm_struct
*mm
)
2933 if (unlikely(atomic_dec_and_test(&mm
->mm_count
))) {
2934 INIT_WORK(&mm
->async_put_work
, mmdrop_async_fn
);
2935 schedule_work(&mm
->async_put_work
);
2940 * mmget() - Pin the address space associated with a &struct mm_struct.
2941 * @mm: The address space to pin.
2943 * Make sure that the address space of the given &struct mm_struct doesn't
2944 * go away. This does not protect against parts of the address space being
2945 * modified or freed, however.
2947 * Never use this function to pin this address space for an
2948 * unbounded/indefinite amount of time.
2950 * Use mmput() to release the reference acquired by mmget().
2952 * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
2953 * of &mm_struct.mm_count vs &mm_struct.mm_users.
2955 static inline void mmget(struct mm_struct
*mm
)
2957 atomic_inc(&mm
->mm_users
);
2960 static inline bool mmget_not_zero(struct mm_struct
*mm
)
2962 return atomic_inc_not_zero(&mm
->mm_users
);
2965 /* mmput gets rid of the mappings and all user-space */
2966 extern void mmput(struct mm_struct
*);
2968 /* same as above but performs the slow path from the async context. Can
2969 * be called from the atomic context as well
2971 extern void mmput_async(struct mm_struct
*);
2974 /* Grab a reference to a task's mm, if it is not already going away */
2975 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2977 * Grab a reference to a task's mm, if it is not already going away
2978 * and ptrace_may_access with the mode parameter passed to it
2981 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2982 /* Remove the current tasks stale references to the old mm_struct */
2983 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2985 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2986 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2987 struct task_struct
*, unsigned long);
2989 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2990 struct task_struct
*);
2992 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2993 * via pt_regs, so ignore the tls argument passed via C. */
2994 static inline int copy_thread_tls(
2995 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2996 struct task_struct
*p
, unsigned long tls
)
2998 return copy_thread(clone_flags
, sp
, arg
, p
);
3001 extern void flush_thread(void);
3003 #ifdef CONFIG_HAVE_EXIT_THREAD
3004 extern void exit_thread(struct task_struct
*tsk
);
3006 static inline void exit_thread(struct task_struct
*tsk
)
3011 extern void exit_files(struct task_struct
*);
3012 extern void __cleanup_sighand(struct sighand_struct
*);
3014 extern void exit_itimers(struct signal_struct
*);
3015 extern void flush_itimer_signals(void);
3017 extern void do_group_exit(int);
3019 extern int do_execve(struct filename
*,
3020 const char __user
* const __user
*,
3021 const char __user
* const __user
*);
3022 extern int do_execveat(int, struct filename
*,
3023 const char __user
* const __user
*,
3024 const char __user
* const __user
*,
3026 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
3027 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
3028 struct task_struct
*fork_idle(int);
3029 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
3031 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
3032 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
3034 __set_task_comm(tsk
, from
, false);
3036 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
3039 void scheduler_ipi(void);
3040 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
3042 static inline void scheduler_ipi(void) { }
3043 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
3050 #define tasklist_empty() \
3051 list_empty(&init_task.tasks)
3053 #define next_task(p) \
3054 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
3056 #define for_each_process(p) \
3057 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
3059 extern bool current_is_single_threaded(void);
3062 * Careful: do_each_thread/while_each_thread is a double loop so
3063 * 'break' will not work as expected - use goto instead.
3065 #define do_each_thread(g, t) \
3066 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
3068 #define while_each_thread(g, t) \
3069 while ((t = next_thread(t)) != g)
3071 #define __for_each_thread(signal, t) \
3072 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
3074 #define for_each_thread(p, t) \
3075 __for_each_thread((p)->signal, t)
3077 /* Careful: this is a double loop, 'break' won't work as expected. */
3078 #define for_each_process_thread(p, t) \
3079 for_each_process(p) for_each_thread(p, t)
3081 typedef int (*proc_visitor
)(struct task_struct
*p
, void *data
);
3082 void walk_process_tree(struct task_struct
*top
, proc_visitor
, void *);
3084 static inline int get_nr_threads(struct task_struct
*tsk
)
3086 return tsk
->signal
->nr_threads
;
3089 static inline bool thread_group_leader(struct task_struct
*p
)
3091 return p
->exit_signal
>= 0;
3094 /* Do to the insanities of de_thread it is possible for a process
3095 * to have the pid of the thread group leader without actually being
3096 * the thread group leader. For iteration through the pids in proc
3097 * all we care about is that we have a task with the appropriate
3098 * pid, we don't actually care if we have the right task.
3100 static inline bool has_group_leader_pid(struct task_struct
*p
)
3102 return task_pid(p
) == p
->signal
->leader_pid
;
3106 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
3108 return p1
->signal
== p2
->signal
;
3111 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
3113 return list_entry_rcu(p
->thread_group
.next
,
3114 struct task_struct
, thread_group
);
3117 static inline int thread_group_empty(struct task_struct
*p
)
3119 return list_empty(&p
->thread_group
);
3122 #define delay_group_leader(p) \
3123 (thread_group_leader(p) && !thread_group_empty(p))
3126 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
3127 * subscriptions and synchronises with wait4(). Also used in procfs. Also
3128 * pins the final release of task.io_context. Also protects ->cpuset and
3129 * ->cgroup.subsys[]. And ->vfork_done.
3131 * Nests both inside and outside of read_lock(&tasklist_lock).
3132 * It must not be nested with write_lock_irq(&tasklist_lock),
3133 * neither inside nor outside.
3135 static inline void task_lock(struct task_struct
*p
)
3137 spin_lock(&p
->alloc_lock
);
3140 static inline void task_unlock(struct task_struct
*p
)
3142 spin_unlock(&p
->alloc_lock
);
3145 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
3146 unsigned long *flags
);
3148 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
3149 unsigned long *flags
)
3151 struct sighand_struct
*ret
;
3153 ret
= __lock_task_sighand(tsk
, flags
);
3154 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
3158 static inline void unlock_task_sighand(struct task_struct
*tsk
,
3159 unsigned long *flags
)
3161 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
3164 #ifdef CONFIG_THREAD_INFO_IN_TASK
3166 static inline struct thread_info
*task_thread_info(struct task_struct
*task
)
3168 return &task
->thread_info
;
3172 * When accessing the stack of a non-current task that might exit, use
3173 * try_get_task_stack() instead. task_stack_page will return a pointer
3174 * that could get freed out from under you.
3176 static inline void *task_stack_page(const struct task_struct
*task
)
3181 #define setup_thread_stack(new,old) do { } while(0)
3183 static inline unsigned long *end_of_stack(const struct task_struct
*task
)
3188 #elif !defined(__HAVE_THREAD_FUNCTIONS)
3190 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
3191 #define task_stack_page(task) ((void *)(task)->stack)
3193 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
3195 *task_thread_info(p
) = *task_thread_info(org
);
3196 task_thread_info(p
)->task
= p
;
3200 * Return the address of the last usable long on the stack.
3202 * When the stack grows down, this is just above the thread
3203 * info struct. Going any lower will corrupt the threadinfo.
3205 * When the stack grows up, this is the highest address.
3206 * Beyond that position, we corrupt data on the next page.
3208 static inline unsigned long *end_of_stack(struct task_struct
*p
)
3210 #ifdef CONFIG_STACK_GROWSUP
3211 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
3213 return (unsigned long *)(task_thread_info(p
) + 1);
3219 #ifdef CONFIG_THREAD_INFO_IN_TASK
3220 static inline void *try_get_task_stack(struct task_struct
*tsk
)
3222 return atomic_inc_not_zero(&tsk
->stack_refcount
) ?
3223 task_stack_page(tsk
) : NULL
;
3226 extern void put_task_stack(struct task_struct
*tsk
);
3228 static inline void *try_get_task_stack(struct task_struct
*tsk
)
3230 return task_stack_page(tsk
);
3233 static inline void put_task_stack(struct task_struct
*tsk
) {}
3236 #define task_stack_end_corrupted(task) \
3237 (*(end_of_stack(task)) != STACK_END_MAGIC)
3239 static inline int object_is_on_stack(void *obj
)
3241 void *stack
= task_stack_page(current
);
3243 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
3246 extern void thread_stack_cache_init(void);
3248 #ifdef CONFIG_DEBUG_STACK_USAGE
3249 static inline unsigned long stack_not_used(struct task_struct
*p
)
3251 unsigned long *n
= end_of_stack(p
);
3253 do { /* Skip over canary */
3254 # ifdef CONFIG_STACK_GROWSUP
3261 # ifdef CONFIG_STACK_GROWSUP
3262 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
3264 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
3268 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
3270 /* set thread flags in other task's structures
3271 * - see asm/thread_info.h for TIF_xxxx flags available
3273 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3275 set_ti_thread_flag(task_thread_info(tsk
), flag
);
3278 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3280 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3283 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3285 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
3288 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3290 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3293 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3295 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
3298 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
3300 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3303 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
3305 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3308 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
3310 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
3313 static inline int restart_syscall(void)
3315 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
3316 return -ERESTARTNOINTR
;
3319 static inline int signal_pending(struct task_struct
*p
)
3321 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
3324 static inline int __fatal_signal_pending(struct task_struct
*p
)
3326 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
3329 static inline int fatal_signal_pending(struct task_struct
*p
)
3331 return signal_pending(p
) && __fatal_signal_pending(p
);
3334 static inline int signal_pending_state(long state
, struct task_struct
*p
)
3336 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
3338 if (!signal_pending(p
))
3341 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
3345 * cond_resched() and cond_resched_lock(): latency reduction via
3346 * explicit rescheduling in places that are safe. The return
3347 * value indicates whether a reschedule was done in fact.
3348 * cond_resched_lock() will drop the spinlock before scheduling,
3349 * cond_resched_softirq() will enable bhs before scheduling.
3351 #ifndef CONFIG_PREEMPT
3352 extern int _cond_resched(void);
3354 static inline int _cond_resched(void) { return 0; }
3357 #define cond_resched() ({ \
3358 ___might_sleep(__FILE__, __LINE__, 0); \
3362 extern int __cond_resched_lock(spinlock_t
*lock
);
3364 #define cond_resched_lock(lock) ({ \
3365 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3366 __cond_resched_lock(lock); \
3369 extern int __cond_resched_softirq(void);
3371 #define cond_resched_softirq() ({ \
3372 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3373 __cond_resched_softirq(); \
3376 static inline void cond_resched_rcu(void)
3378 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3386 * Does a critical section need to be broken due to another
3387 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3388 * but a general need for low latency)
3390 static inline int spin_needbreak(spinlock_t
*lock
)
3392 #ifdef CONFIG_PREEMPT
3393 return spin_is_contended(lock
);
3400 * Idle thread specific functions to determine the need_resched
3403 #ifdef TIF_POLLING_NRFLAG
3404 static inline int tsk_is_polling(struct task_struct
*p
)
3406 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
3409 static inline void __current_set_polling(void)
3411 set_thread_flag(TIF_POLLING_NRFLAG
);
3414 static inline bool __must_check
current_set_polling_and_test(void)
3416 __current_set_polling();
3419 * Polling state must be visible before we test NEED_RESCHED,
3420 * paired by resched_curr()
3422 smp_mb__after_atomic();
3424 return unlikely(tif_need_resched());
3427 static inline void __current_clr_polling(void)
3429 clear_thread_flag(TIF_POLLING_NRFLAG
);
3432 static inline bool __must_check
current_clr_polling_and_test(void)
3434 __current_clr_polling();
3437 * Polling state must be visible before we test NEED_RESCHED,
3438 * paired by resched_curr()
3440 smp_mb__after_atomic();
3442 return unlikely(tif_need_resched());
3446 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
3447 static inline void __current_set_polling(void) { }
3448 static inline void __current_clr_polling(void) { }
3450 static inline bool __must_check
current_set_polling_and_test(void)
3452 return unlikely(tif_need_resched());
3454 static inline bool __must_check
current_clr_polling_and_test(void)
3456 return unlikely(tif_need_resched());
3460 static inline void current_clr_polling(void)
3462 __current_clr_polling();
3465 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3466 * Once the bit is cleared, we'll get IPIs with every new
3467 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3470 smp_mb(); /* paired with resched_curr() */
3472 preempt_fold_need_resched();
3475 static __always_inline
bool need_resched(void)
3477 return unlikely(tif_need_resched());
3481 * Thread group CPU time accounting.
3483 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
3484 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
3487 * Reevaluate whether the task has signals pending delivery.
3488 * Wake the task if so.
3489 * This is required every time the blocked sigset_t changes.
3490 * callers must hold sighand->siglock.
3492 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
3493 extern void recalc_sigpending(void);
3495 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
3497 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
3499 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
3501 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
3503 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
3507 * Wrappers for p->thread_info->cpu access. No-op on UP.
3511 static inline unsigned int task_cpu(const struct task_struct
*p
)
3513 #ifdef CONFIG_THREAD_INFO_IN_TASK
3516 return task_thread_info(p
)->cpu
;
3520 static inline int task_node(const struct task_struct
*p
)
3522 return cpu_to_node(task_cpu(p
));
3525 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
3529 static inline unsigned int task_cpu(const struct task_struct
*p
)
3534 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
3538 #endif /* CONFIG_SMP */
3541 * In order to reduce various lock holder preemption latencies provide an
3542 * interface to see if a vCPU is currently running or not.
3544 * This allows us to terminate optimistic spin loops and block, analogous to
3545 * the native optimistic spin heuristic of testing if the lock owner task is
3548 #ifndef vcpu_is_preempted
3549 # define vcpu_is_preempted(cpu) false
3552 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3553 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3555 #ifdef CONFIG_CGROUP_SCHED
3556 extern struct task_group root_task_group
;
3557 #endif /* CONFIG_CGROUP_SCHED */
3559 extern int task_can_switch_user(struct user_struct
*up
,
3560 struct task_struct
*tsk
);
3562 #ifdef CONFIG_TASK_XACCT
3563 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3565 tsk
->ioac
.rchar
+= amt
;
3568 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3570 tsk
->ioac
.wchar
+= amt
;
3573 static inline void inc_syscr(struct task_struct
*tsk
)
3578 static inline void inc_syscw(struct task_struct
*tsk
)
3583 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3587 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3591 static inline void inc_syscr(struct task_struct
*tsk
)
3595 static inline void inc_syscw(struct task_struct
*tsk
)
3600 #ifndef TASK_SIZE_OF
3601 #define TASK_SIZE_OF(tsk) TASK_SIZE
3605 extern void mm_update_next_owner(struct mm_struct
*mm
);
3607 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3610 #endif /* CONFIG_MEMCG */
3612 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3615 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3618 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3621 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3624 static inline unsigned long rlimit(unsigned int limit
)
3626 return task_rlimit(current
, limit
);
3629 static inline unsigned long rlimit_max(unsigned int limit
)
3631 return task_rlimit_max(current
, limit
);
3634 #define SCHED_CPUFREQ_RT (1U << 0)
3635 #define SCHED_CPUFREQ_DL (1U << 1)
3636 #define SCHED_CPUFREQ_IOWAIT (1U << 2)
3638 #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
3640 #ifdef CONFIG_CPU_FREQ
3641 struct update_util_data
{
3642 void (*func
)(struct update_util_data
*data
, u64 time
, unsigned int flags
);
3645 void cpufreq_add_update_util_hook(int cpu
, struct update_util_data
*data
,
3646 void (*func
)(struct update_util_data
*data
, u64 time
,
3647 unsigned int flags
));
3648 void cpufreq_remove_update_util_hook(int cpu
);
3649 #endif /* CONFIG_CPU_FREQ */