4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
8 #include <asm/param.h> /* for HZ */
10 #include <linux/capability.h>
11 #include <linux/threads.h>
12 #include <linux/kernel.h>
13 #include <linux/types.h>
14 #include <linux/timex.h>
15 #include <linux/jiffies.h>
16 #include <linux/mutex.h>
17 #include <linux/plist.h>
18 #include <linux/rbtree.h>
19 #include <linux/thread_info.h>
20 #include <linux/cpumask.h>
21 #include <linux/errno.h>
22 #include <linux/nodemask.h>
23 #include <linux/mm_types.h>
24 #include <linux/preempt.h>
27 #include <asm/ptrace.h>
29 #include <linux/smp.h>
30 #include <linux/sem.h>
31 #include <linux/shm.h>
32 #include <linux/signal.h>
33 #include <linux/compiler.h>
34 #include <linux/completion.h>
35 #include <linux/signal_types.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/seccomp.h>
40 #include <linux/rcupdate.h>
41 #include <linux/rculist.h>
42 #include <linux/rtmutex.h>
44 #include <linux/time.h>
45 #include <linux/param.h>
46 #include <linux/resource.h>
47 #include <linux/timer.h>
48 #include <linux/hrtimer.h>
49 #include <linux/kcov.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
56 #include <linux/topology.h>
57 #include <linux/magic.h>
58 #include <linux/cgroup-defs.h>
60 #include <asm/processor.h>
65 struct futex_pi_state
;
66 struct robust_list_head
;
69 struct perf_event_context
;
75 * These are the constant used to fake the fixed-point load-average
76 * counting. Some notes:
77 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
78 * a load-average precision of 10 bits integer + 11 bits fractional
79 * - if you want to count load-averages more often, you need more
80 * precision, or rounding will get you. With 2-second counting freq,
81 * the EXP_n values would be 1981, 2034 and 2043 if still using only
84 extern unsigned long avenrun
[]; /* Load averages */
85 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
87 #define FSHIFT 11 /* nr of bits of precision */
88 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
89 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
90 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
91 #define EXP_5 2014 /* 1/exp(5sec/5min) */
92 #define EXP_15 2037 /* 1/exp(5sec/15min) */
94 #define CALC_LOAD(load,exp,n) \
96 load += n*(FIXED_1-exp); \
99 extern unsigned long total_forks
;
100 extern int nr_threads
;
101 DECLARE_PER_CPU(unsigned long, process_counts
);
102 extern int nr_processes(void);
103 extern unsigned long nr_running(void);
104 extern bool single_task_running(void);
105 extern unsigned long nr_iowait(void);
106 extern unsigned long nr_iowait_cpu(int cpu
);
107 extern void get_iowait_load(unsigned long *nr_waiters
, unsigned long *load
);
109 extern void calc_global_load(unsigned long ticks
);
111 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
112 extern void cpu_load_update_nohz_start(void);
113 extern void cpu_load_update_nohz_stop(void);
115 static inline void cpu_load_update_nohz_start(void) { }
116 static inline void cpu_load_update_nohz_stop(void) { }
119 extern void dump_cpu_task(int cpu
);
124 #ifdef CONFIG_SCHED_DEBUG
125 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
126 extern void proc_sched_set_task(struct task_struct
*p
);
130 * Task state bitmask. NOTE! These bits are also
131 * encoded in fs/proc/array.c: get_task_state().
133 * We have two separate sets of flags: task->state
134 * is about runnability, while task->exit_state are
135 * about the task exiting. Confusing, but this way
136 * modifying one set can't modify the other one by
139 #define TASK_RUNNING 0
140 #define TASK_INTERRUPTIBLE 1
141 #define TASK_UNINTERRUPTIBLE 2
142 #define __TASK_STOPPED 4
143 #define __TASK_TRACED 8
144 /* in tsk->exit_state */
146 #define EXIT_ZOMBIE 32
147 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
148 /* in tsk->state again */
150 #define TASK_WAKEKILL 128
151 #define TASK_WAKING 256
152 #define TASK_PARKED 512
153 #define TASK_NOLOAD 1024
154 #define TASK_NEW 2048
155 #define TASK_STATE_MAX 4096
157 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
159 /* Convenience macros for the sake of set_current_state */
160 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
161 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
162 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
164 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
166 /* Convenience macros for the sake of wake_up */
167 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
168 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
170 /* get_task_state() */
171 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
172 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
173 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
175 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
176 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
177 #define task_is_stopped_or_traced(task) \
178 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
179 #define task_contributes_to_load(task) \
180 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
181 (task->flags & PF_FROZEN) == 0 && \
182 (task->state & TASK_NOLOAD) == 0)
184 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
186 #define __set_current_state(state_value) \
188 current->task_state_change = _THIS_IP_; \
189 current->state = (state_value); \
191 #define set_current_state(state_value) \
193 current->task_state_change = _THIS_IP_; \
194 smp_store_mb(current->state, (state_value)); \
199 * set_current_state() includes a barrier so that the write of current->state
200 * is correctly serialised wrt the caller's subsequent test of whether to
204 * set_current_state(TASK_UNINTERRUPTIBLE);
210 * __set_current_state(TASK_RUNNING);
212 * If the caller does not need such serialisation (because, for instance, the
213 * condition test and condition change and wakeup are under the same lock) then
214 * use __set_current_state().
216 * The above is typically ordered against the wakeup, which does:
218 * need_sleep = false;
219 * wake_up_state(p, TASK_UNINTERRUPTIBLE);
221 * Where wake_up_state() (and all other wakeup primitives) imply enough
222 * barriers to order the store of the variable against wakeup.
224 * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is,
225 * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a
226 * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING).
228 * This is obviously fine, since they both store the exact same value.
230 * Also see the comments of try_to_wake_up().
232 #define __set_current_state(state_value) \
233 do { current->state = (state_value); } while (0)
234 #define set_current_state(state_value) \
235 smp_store_mb(current->state, (state_value))
239 /* Task command name length */
240 #define TASK_COMM_LEN 16
242 #include <linux/spinlock.h>
245 * This serializes "schedule()" and also protects
246 * the run-queue from deletions/modifications (but
247 * _adding_ to the beginning of the run-queue has
250 extern rwlock_t tasklist_lock
;
251 extern spinlock_t mmlist_lock
;
255 #ifdef CONFIG_PROVE_RCU
256 extern int lockdep_tasklist_lock_is_held(void);
257 #endif /* #ifdef CONFIG_PROVE_RCU */
259 extern void sched_init(void);
260 extern void sched_init_smp(void);
261 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
262 extern void init_idle(struct task_struct
*idle
, int cpu
);
263 extern void init_idle_bootup_task(struct task_struct
*idle
);
265 extern cpumask_var_t cpu_isolated_map
;
267 extern int runqueue_is_locked(int cpu
);
269 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
270 extern void nohz_balance_enter_idle(int cpu
);
271 extern void set_cpu_sd_state_idle(void);
272 extern int get_nohz_timer_target(void);
274 static inline void nohz_balance_enter_idle(int cpu
) { }
275 static inline void set_cpu_sd_state_idle(void) { }
279 * Only dump TASK_* tasks. (0 for all tasks)
281 extern void show_state_filter(unsigned long state_filter
);
283 static inline void show_state(void)
285 show_state_filter(0);
288 extern void show_regs(struct pt_regs
*);
291 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
292 * task), SP is the stack pointer of the first frame that should be shown in the back
293 * trace (or NULL if the entire call-chain of the task should be shown).
295 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
297 extern void cpu_init (void);
298 extern void trap_init(void);
299 extern void update_process_times(int user
);
300 extern void scheduler_tick(void);
301 extern int sched_cpu_starting(unsigned int cpu
);
302 extern int sched_cpu_activate(unsigned int cpu
);
303 extern int sched_cpu_deactivate(unsigned int cpu
);
305 #ifdef CONFIG_HOTPLUG_CPU
306 extern int sched_cpu_dying(unsigned int cpu
);
308 # define sched_cpu_dying NULL
311 extern void sched_show_task(struct task_struct
*p
);
313 #ifdef CONFIG_LOCKUP_DETECTOR
314 extern void touch_softlockup_watchdog_sched(void);
315 extern void touch_softlockup_watchdog(void);
316 extern void touch_softlockup_watchdog_sync(void);
317 extern void touch_all_softlockup_watchdogs(void);
318 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
320 size_t *lenp
, loff_t
*ppos
);
321 extern unsigned int softlockup_panic
;
322 extern unsigned int hardlockup_panic
;
323 void lockup_detector_init(void);
325 static inline void touch_softlockup_watchdog_sched(void)
328 static inline void touch_softlockup_watchdog(void)
331 static inline void touch_softlockup_watchdog_sync(void)
334 static inline void touch_all_softlockup_watchdogs(void)
337 static inline void lockup_detector_init(void)
342 #ifdef CONFIG_DETECT_HUNG_TASK
343 void reset_hung_task_detector(void);
345 static inline void reset_hung_task_detector(void)
350 /* Attach to any functions which should be ignored in wchan output. */
351 #define __sched __attribute__((__section__(".sched.text")))
353 /* Linker adds these: start and end of __sched functions */
354 extern char __sched_text_start
[], __sched_text_end
[];
356 /* Is this address in the __sched functions? */
357 extern int in_sched_functions(unsigned long addr
);
359 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
360 extern signed long schedule_timeout(signed long timeout
);
361 extern signed long schedule_timeout_interruptible(signed long timeout
);
362 extern signed long schedule_timeout_killable(signed long timeout
);
363 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
364 extern signed long schedule_timeout_idle(signed long timeout
);
365 asmlinkage
void schedule(void);
366 extern void schedule_preempt_disabled(void);
368 extern int __must_check
io_schedule_prepare(void);
369 extern void io_schedule_finish(int token
);
370 extern long io_schedule_timeout(long timeout
);
371 extern void io_schedule(void);
373 void __noreturn
do_task_dead(void);
376 struct user_namespace
;
379 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
381 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
382 unsigned long, unsigned long);
384 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
385 unsigned long len
, unsigned long pgoff
,
386 unsigned long flags
);
388 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
391 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
392 #define SUID_DUMP_USER 1 /* Dump as user of process */
393 #define SUID_DUMP_ROOT 2 /* Dump as root */
397 /* for SUID_DUMP_* above */
398 #define MMF_DUMPABLE_BITS 2
399 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
401 extern void set_dumpable(struct mm_struct
*mm
, int value
);
403 * This returns the actual value of the suid_dumpable flag. For things
404 * that are using this for checking for privilege transitions, it must
405 * test against SUID_DUMP_USER rather than treating it as a boolean
408 static inline int __get_dumpable(unsigned long mm_flags
)
410 return mm_flags
& MMF_DUMPABLE_MASK
;
413 static inline int get_dumpable(struct mm_struct
*mm
)
415 return __get_dumpable(mm
->flags
);
418 /* coredump filter bits */
419 #define MMF_DUMP_ANON_PRIVATE 2
420 #define MMF_DUMP_ANON_SHARED 3
421 #define MMF_DUMP_MAPPED_PRIVATE 4
422 #define MMF_DUMP_MAPPED_SHARED 5
423 #define MMF_DUMP_ELF_HEADERS 6
424 #define MMF_DUMP_HUGETLB_PRIVATE 7
425 #define MMF_DUMP_HUGETLB_SHARED 8
426 #define MMF_DUMP_DAX_PRIVATE 9
427 #define MMF_DUMP_DAX_SHARED 10
429 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
430 #define MMF_DUMP_FILTER_BITS 9
431 #define MMF_DUMP_FILTER_MASK \
432 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
433 #define MMF_DUMP_FILTER_DEFAULT \
434 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
435 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
437 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
438 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
440 # define MMF_DUMP_MASK_DEFAULT_ELF 0
442 /* leave room for more dump flags */
443 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
444 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
446 * This one-shot flag is dropped due to necessity of changing exe once again
449 //#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
451 #define MMF_HAS_UPROBES 19 /* has uprobes */
452 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
453 #define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */
454 #define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */
455 #define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */
457 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
459 struct sighand_struct
{
461 struct k_sigaction action
[_NSIG
];
463 wait_queue_head_t signalfd_wqh
;
466 struct pacct_struct
{
469 unsigned long ac_mem
;
470 u64 ac_utime
, ac_stime
;
471 unsigned long ac_minflt
, ac_majflt
;
480 * struct prev_cputime - snaphsot of system and user cputime
481 * @utime: time spent in user mode
482 * @stime: time spent in system mode
483 * @lock: protects the above two fields
485 * Stores previous user/system time values such that we can guarantee
488 struct prev_cputime
{
489 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
496 static inline void prev_cputime_init(struct prev_cputime
*prev
)
498 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
499 prev
->utime
= prev
->stime
= 0;
500 raw_spin_lock_init(&prev
->lock
);
505 * struct task_cputime - collected CPU time counts
506 * @utime: time spent in user mode, in nanoseconds
507 * @stime: time spent in kernel mode, in nanoseconds
508 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
510 * This structure groups together three kinds of CPU time that are tracked for
511 * threads and thread groups. Most things considering CPU time want to group
512 * these counts together and treat all three of them in parallel.
514 struct task_cputime
{
517 unsigned long long sum_exec_runtime
;
520 /* Alternate field names when used to cache expirations. */
521 #define virt_exp utime
522 #define prof_exp stime
523 #define sched_exp sum_exec_runtime
526 * This is the atomic variant of task_cputime, which can be used for
527 * storing and updating task_cputime statistics without locking.
529 struct task_cputime_atomic
{
532 atomic64_t sum_exec_runtime
;
535 #define INIT_CPUTIME_ATOMIC \
536 (struct task_cputime_atomic) { \
537 .utime = ATOMIC64_INIT(0), \
538 .stime = ATOMIC64_INIT(0), \
539 .sum_exec_runtime = ATOMIC64_INIT(0), \
542 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
545 * Disable preemption until the scheduler is running -- use an unconditional
546 * value so that it also works on !PREEMPT_COUNT kernels.
548 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
550 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
553 * Initial preempt_count value; reflects the preempt_count schedule invariant
554 * which states that during context switches:
556 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
558 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
559 * Note: See finish_task_switch().
561 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
564 * struct thread_group_cputimer - thread group interval timer counts
565 * @cputime_atomic: atomic thread group interval timers.
566 * @running: true when there are timers running and
567 * @cputime_atomic receives updates.
568 * @checking_timer: true when a thread in the group is in the
569 * process of checking for thread group timers.
571 * This structure contains the version of task_cputime, above, that is
572 * used for thread group CPU timer calculations.
574 struct thread_group_cputimer
{
575 struct task_cputime_atomic cputime_atomic
;
580 #include <linux/rwsem.h>
584 * NOTE! "signal_struct" does not have its own
585 * locking, because a shared signal_struct always
586 * implies a shared sighand_struct, so locking
587 * sighand_struct is always a proper superset of
588 * the locking of signal_struct.
590 struct signal_struct
{
594 struct list_head thread_head
;
596 wait_queue_head_t wait_chldexit
; /* for wait4() */
598 /* current thread group signal load-balancing target: */
599 struct task_struct
*curr_target
;
601 /* shared signal handling: */
602 struct sigpending shared_pending
;
604 /* thread group exit support */
607 * - notify group_exit_task when ->count is equal to notify_count
608 * - everyone except group_exit_task is stopped during signal delivery
609 * of fatal signals, group_exit_task processes the signal.
612 struct task_struct
*group_exit_task
;
614 /* thread group stop support, overloads group_exit_code too */
615 int group_stop_count
;
616 unsigned int flags
; /* see SIGNAL_* flags below */
619 * PR_SET_CHILD_SUBREAPER marks a process, like a service
620 * manager, to re-parent orphan (double-forking) child processes
621 * to this process instead of 'init'. The service manager is
622 * able to receive SIGCHLD signals and is able to investigate
623 * the process until it calls wait(). All children of this
624 * process will inherit a flag if they should look for a
625 * child_subreaper process at exit.
627 unsigned int is_child_subreaper
:1;
628 unsigned int has_child_subreaper
:1;
630 #ifdef CONFIG_POSIX_TIMERS
632 /* POSIX.1b Interval Timers */
634 struct list_head posix_timers
;
636 /* ITIMER_REAL timer for the process */
637 struct hrtimer real_timer
;
638 ktime_t it_real_incr
;
641 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
642 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
643 * values are defined to 0 and 1 respectively
645 struct cpu_itimer it
[2];
648 * Thread group totals for process CPU timers.
649 * See thread_group_cputimer(), et al, for details.
651 struct thread_group_cputimer cputimer
;
653 /* Earliest-expiration cache. */
654 struct task_cputime cputime_expires
;
656 struct list_head cpu_timers
[3];
660 struct pid
*leader_pid
;
662 #ifdef CONFIG_NO_HZ_FULL
663 atomic_t tick_dep_mask
;
666 struct pid
*tty_old_pgrp
;
668 /* boolean value for session group leader */
671 struct tty_struct
*tty
; /* NULL if no tty */
673 #ifdef CONFIG_SCHED_AUTOGROUP
674 struct autogroup
*autogroup
;
677 * Cumulative resource counters for dead threads in the group,
678 * and for reaped dead child processes forked by this group.
679 * Live threads maintain their own counters and add to these
680 * in __exit_signal, except for the group leader.
682 seqlock_t stats_lock
;
683 u64 utime
, stime
, cutime
, cstime
;
686 struct prev_cputime prev_cputime
;
687 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
688 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
689 unsigned long inblock
, oublock
, cinblock
, coublock
;
690 unsigned long maxrss
, cmaxrss
;
691 struct task_io_accounting ioac
;
694 * Cumulative ns of schedule CPU time fo dead threads in the
695 * group, not including a zombie group leader, (This only differs
696 * from jiffies_to_ns(utime + stime) if sched_clock uses something
697 * other than jiffies.)
699 unsigned long long sum_sched_runtime
;
702 * We don't bother to synchronize most readers of this at all,
703 * because there is no reader checking a limit that actually needs
704 * to get both rlim_cur and rlim_max atomically, and either one
705 * alone is a single word that can safely be read normally.
706 * getrlimit/setrlimit use task_lock(current->group_leader) to
707 * protect this instead of the siglock, because they really
708 * have no need to disable irqs.
710 struct rlimit rlim
[RLIM_NLIMITS
];
712 #ifdef CONFIG_BSD_PROCESS_ACCT
713 struct pacct_struct pacct
; /* per-process accounting information */
715 #ifdef CONFIG_TASKSTATS
716 struct taskstats
*stats
;
720 struct tty_audit_buf
*tty_audit_buf
;
724 * Thread is the potential origin of an oom condition; kill first on
727 bool oom_flag_origin
;
728 short oom_score_adj
; /* OOM kill score adjustment */
729 short oom_score_adj_min
; /* OOM kill score adjustment min value.
730 * Only settable by CAP_SYS_RESOURCE. */
731 struct mm_struct
*oom_mm
; /* recorded mm when the thread group got
732 * killed by the oom killer */
734 struct mutex cred_guard_mutex
; /* guard against foreign influences on
735 * credential calculations
736 * (notably. ptrace) */
740 * Bits in flags field of signal_struct.
742 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
743 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
744 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
745 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
747 * Pending notifications to parent.
749 #define SIGNAL_CLD_STOPPED 0x00000010
750 #define SIGNAL_CLD_CONTINUED 0x00000020
751 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
753 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
755 #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \
756 SIGNAL_STOP_CONTINUED)
758 static inline void signal_set_stop_flags(struct signal_struct
*sig
,
761 WARN_ON(sig
->flags
& (SIGNAL_GROUP_EXIT
|SIGNAL_GROUP_COREDUMP
));
762 sig
->flags
= (sig
->flags
& ~SIGNAL_STOP_MASK
) | flags
;
765 /* If true, all threads except ->group_exit_task have pending SIGKILL */
766 static inline int signal_group_exit(const struct signal_struct
*sig
)
768 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
769 (sig
->group_exit_task
!= NULL
);
773 * Some day this will be a full-fledged user tracking system..
776 atomic_t __count
; /* reference count */
777 atomic_t processes
; /* How many processes does this user have? */
778 atomic_t sigpending
; /* How many pending signals does this user have? */
779 #ifdef CONFIG_FANOTIFY
780 atomic_t fanotify_listeners
;
783 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
785 #ifdef CONFIG_POSIX_MQUEUE
786 /* protected by mq_lock */
787 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
789 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
790 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
791 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
794 struct key
*uid_keyring
; /* UID specific keyring */
795 struct key
*session_keyring
; /* UID's default session keyring */
798 /* Hash table maintenance information */
799 struct hlist_node uidhash_node
;
802 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
803 atomic_long_t locked_vm
;
807 extern int uids_sysfs_init(void);
809 extern struct user_struct
*find_user(kuid_t
);
811 extern struct user_struct root_user
;
812 #define INIT_USER (&root_user)
815 struct backing_dev_info
;
816 struct reclaim_state
;
818 #ifdef CONFIG_SCHED_INFO
820 /* cumulative counters */
821 unsigned long pcount
; /* # of times run on this cpu */
822 unsigned long long run_delay
; /* time spent waiting on a runqueue */
825 unsigned long long last_arrival
,/* when we last ran on a cpu */
826 last_queued
; /* when we were last queued to run */
828 #endif /* CONFIG_SCHED_INFO */
830 struct task_delay_info
;
832 static inline int sched_info_on(void)
834 #ifdef CONFIG_SCHEDSTATS
836 #elif defined(CONFIG_TASK_DELAY_ACCT)
837 extern int delayacct_on
;
844 #ifdef CONFIG_SCHEDSTATS
845 void force_schedstat_enabled(void);
849 * Integer metrics need fixed point arithmetic, e.g., sched/fair
850 * has a few: load, load_avg, util_avg, freq, and capacity.
852 * We define a basic fixed point arithmetic range, and then formalize
853 * all these metrics based on that basic range.
855 # define SCHED_FIXEDPOINT_SHIFT 10
856 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
858 struct io_context
; /* See blkdev.h */
861 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
862 extern void prefetch_stack(struct task_struct
*t
);
864 static inline void prefetch_stack(struct task_struct
*t
) { }
867 struct audit_context
; /* See audit.c */
869 struct pipe_inode_info
;
870 struct uts_namespace
;
873 unsigned long weight
;
878 * The load_avg/util_avg accumulates an infinite geometric series
879 * (see __update_load_avg() in kernel/sched/fair.c).
881 * [load_avg definition]
883 * load_avg = runnable% * scale_load_down(load)
885 * where runnable% is the time ratio that a sched_entity is runnable.
886 * For cfs_rq, it is the aggregated load_avg of all runnable and
887 * blocked sched_entities.
889 * load_avg may also take frequency scaling into account:
891 * load_avg = runnable% * scale_load_down(load) * freq%
893 * where freq% is the CPU frequency normalized to the highest frequency.
895 * [util_avg definition]
897 * util_avg = running% * SCHED_CAPACITY_SCALE
899 * where running% is the time ratio that a sched_entity is running on
900 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
901 * and blocked sched_entities.
903 * util_avg may also factor frequency scaling and CPU capacity scaling:
905 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
907 * where freq% is the same as above, and capacity% is the CPU capacity
908 * normalized to the greatest capacity (due to uarch differences, etc).
910 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
911 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
912 * we therefore scale them to as large a range as necessary. This is for
913 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
917 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
918 * with the highest load (=88761), always runnable on a single cfs_rq,
919 * and should not overflow as the number already hits PID_MAX_LIMIT.
921 * For all other cases (including 32-bit kernels), struct load_weight's
922 * weight will overflow first before we do, because:
924 * Max(load_avg) <= Max(load.weight)
926 * Then it is the load_weight's responsibility to consider overflow
930 u64 last_update_time
, load_sum
;
931 u32 util_sum
, period_contrib
;
932 unsigned long load_avg
, util_avg
;
935 #ifdef CONFIG_SCHEDSTATS
936 struct sched_statistics
{
946 s64 sum_sleep_runtime
;
953 u64 nr_migrations_cold
;
954 u64 nr_failed_migrations_affine
;
955 u64 nr_failed_migrations_running
;
956 u64 nr_failed_migrations_hot
;
957 u64 nr_forced_migrations
;
961 u64 nr_wakeups_migrate
;
962 u64 nr_wakeups_local
;
963 u64 nr_wakeups_remote
;
964 u64 nr_wakeups_affine
;
965 u64 nr_wakeups_affine_attempts
;
966 u64 nr_wakeups_passive
;
971 struct sched_entity
{
972 struct load_weight load
; /* for load-balancing */
973 struct rb_node run_node
;
974 struct list_head group_node
;
978 u64 sum_exec_runtime
;
980 u64 prev_sum_exec_runtime
;
984 #ifdef CONFIG_SCHEDSTATS
985 struct sched_statistics statistics
;
988 #ifdef CONFIG_FAIR_GROUP_SCHED
990 struct sched_entity
*parent
;
991 /* rq on which this entity is (to be) queued: */
992 struct cfs_rq
*cfs_rq
;
993 /* rq "owned" by this entity/group: */
999 * Per entity load average tracking.
1001 * Put into separate cache line so it does not
1002 * collide with read-mostly values above.
1004 struct sched_avg avg ____cacheline_aligned_in_smp
;
1008 struct sched_rt_entity
{
1009 struct list_head run_list
;
1010 unsigned long timeout
;
1011 unsigned long watchdog_stamp
;
1012 unsigned int time_slice
;
1013 unsigned short on_rq
;
1014 unsigned short on_list
;
1016 struct sched_rt_entity
*back
;
1017 #ifdef CONFIG_RT_GROUP_SCHED
1018 struct sched_rt_entity
*parent
;
1019 /* rq on which this entity is (to be) queued: */
1020 struct rt_rq
*rt_rq
;
1021 /* rq "owned" by this entity/group: */
1026 struct sched_dl_entity
{
1027 struct rb_node rb_node
;
1030 * Original scheduling parameters. Copied here from sched_attr
1031 * during sched_setattr(), they will remain the same until
1032 * the next sched_setattr().
1034 u64 dl_runtime
; /* maximum runtime for each instance */
1035 u64 dl_deadline
; /* relative deadline of each instance */
1036 u64 dl_period
; /* separation of two instances (period) */
1037 u64 dl_bw
; /* dl_runtime / dl_deadline */
1040 * Actual scheduling parameters. Initialized with the values above,
1041 * they are continously updated during task execution. Note that
1042 * the remaining runtime could be < 0 in case we are in overrun.
1044 s64 runtime
; /* remaining runtime for this instance */
1045 u64 deadline
; /* absolute deadline for this instance */
1046 unsigned int flags
; /* specifying the scheduler behaviour */
1051 * @dl_throttled tells if we exhausted the runtime. If so, the
1052 * task has to wait for a replenishment to be performed at the
1053 * next firing of dl_timer.
1055 * @dl_boosted tells if we are boosted due to DI. If so we are
1056 * outside bandwidth enforcement mechanism (but only until we
1057 * exit the critical section);
1059 * @dl_yielded tells if task gave up the cpu before consuming
1060 * all its available runtime during the last job.
1062 int dl_throttled
, dl_boosted
, dl_yielded
;
1065 * Bandwidth enforcement timer. Each -deadline task has its
1066 * own bandwidth to be enforced, thus we need one timer per task.
1068 struct hrtimer dl_timer
;
1076 u8 pad
; /* Otherwise the compiler can store garbage here. */
1078 u32 s
; /* Set of bits. */
1082 enum perf_event_task_context
{
1083 perf_invalid_context
= -1,
1084 perf_hw_context
= 0,
1086 perf_nr_task_contexts
,
1089 struct wake_q_node
{
1090 struct wake_q_node
*next
;
1093 /* Track pages that require TLB flushes */
1094 struct tlbflush_unmap_batch
{
1096 * Each bit set is a CPU that potentially has a TLB entry for one of
1097 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1099 struct cpumask cpumask
;
1101 /* True if any bit in cpumask is set */
1102 bool flush_required
;
1105 * If true then the PTE was dirty when unmapped. The entry must be
1106 * flushed before IO is initiated or a stale TLB entry potentially
1107 * allows an update without redirtying the page.
1112 struct task_struct
{
1113 #ifdef CONFIG_THREAD_INFO_IN_TASK
1115 * For reasons of header soup (see current_thread_info()), this
1116 * must be the first element of task_struct.
1118 struct thread_info thread_info
;
1120 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1123 unsigned int flags
; /* per process flags, defined below */
1124 unsigned int ptrace
;
1127 struct llist_node wake_entry
;
1129 #ifdef CONFIG_THREAD_INFO_IN_TASK
1130 unsigned int cpu
; /* current CPU */
1132 unsigned int wakee_flips
;
1133 unsigned long wakee_flip_decay_ts
;
1134 struct task_struct
*last_wakee
;
1140 int prio
, static_prio
, normal_prio
;
1141 unsigned int rt_priority
;
1142 const struct sched_class
*sched_class
;
1143 struct sched_entity se
;
1144 struct sched_rt_entity rt
;
1145 #ifdef CONFIG_CGROUP_SCHED
1146 struct task_group
*sched_task_group
;
1148 struct sched_dl_entity dl
;
1150 #ifdef CONFIG_PREEMPT_NOTIFIERS
1151 /* list of struct preempt_notifier: */
1152 struct hlist_head preempt_notifiers
;
1155 #ifdef CONFIG_BLK_DEV_IO_TRACE
1156 unsigned int btrace_seq
;
1159 unsigned int policy
;
1160 int nr_cpus_allowed
;
1161 cpumask_t cpus_allowed
;
1163 #ifdef CONFIG_PREEMPT_RCU
1164 int rcu_read_lock_nesting
;
1165 union rcu_special rcu_read_unlock_special
;
1166 struct list_head rcu_node_entry
;
1167 struct rcu_node
*rcu_blocked_node
;
1168 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1169 #ifdef CONFIG_TASKS_RCU
1170 unsigned long rcu_tasks_nvcsw
;
1171 bool rcu_tasks_holdout
;
1172 struct list_head rcu_tasks_holdout_list
;
1173 int rcu_tasks_idle_cpu
;
1174 #endif /* #ifdef CONFIG_TASKS_RCU */
1176 #ifdef CONFIG_SCHED_INFO
1177 struct sched_info sched_info
;
1180 struct list_head tasks
;
1182 struct plist_node pushable_tasks
;
1183 struct rb_node pushable_dl_tasks
;
1186 struct mm_struct
*mm
, *active_mm
;
1188 /* Per-thread vma caching: */
1189 struct vmacache vmacache
;
1191 #if defined(SPLIT_RSS_COUNTING)
1192 struct task_rss_stat rss_stat
;
1196 int exit_code
, exit_signal
;
1197 int pdeath_signal
; /* The signal sent when the parent dies */
1198 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1200 /* Used for emulating ABI behavior of previous Linux versions */
1201 unsigned int personality
;
1203 /* scheduler bits, serialized by scheduler locks */
1204 unsigned sched_reset_on_fork
:1;
1205 unsigned sched_contributes_to_load
:1;
1206 unsigned sched_migrated
:1;
1207 unsigned sched_remote_wakeup
:1;
1208 unsigned :0; /* force alignment to the next boundary */
1210 /* unserialized, strictly 'current' */
1211 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1212 unsigned in_iowait
:1;
1213 #if !defined(TIF_RESTORE_SIGMASK)
1214 unsigned restore_sigmask
:1;
1217 unsigned memcg_may_oom
:1;
1219 unsigned memcg_kmem_skip_account
:1;
1222 #ifdef CONFIG_COMPAT_BRK
1223 unsigned brk_randomized
:1;
1226 unsigned long atomic_flags
; /* Flags needing atomic access. */
1228 struct restart_block restart_block
;
1233 #ifdef CONFIG_CC_STACKPROTECTOR
1234 /* Canary value for the -fstack-protector gcc feature */
1235 unsigned long stack_canary
;
1238 * pointers to (original) parent process, youngest child, younger sibling,
1239 * older sibling, respectively. (p->father can be replaced with
1240 * p->real_parent->pid)
1242 struct task_struct __rcu
*real_parent
; /* real parent process */
1243 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1245 * children/sibling forms the list of my natural children
1247 struct list_head children
; /* list of my children */
1248 struct list_head sibling
; /* linkage in my parent's children list */
1249 struct task_struct
*group_leader
; /* threadgroup leader */
1252 * ptraced is the list of tasks this task is using ptrace on.
1253 * This includes both natural children and PTRACE_ATTACH targets.
1254 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1256 struct list_head ptraced
;
1257 struct list_head ptrace_entry
;
1259 /* PID/PID hash table linkage. */
1260 struct pid_link pids
[PIDTYPE_MAX
];
1261 struct list_head thread_group
;
1262 struct list_head thread_node
;
1264 struct completion
*vfork_done
; /* for vfork() */
1265 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1266 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1269 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
1270 u64 utimescaled
, stimescaled
;
1273 struct prev_cputime prev_cputime
;
1274 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1275 seqcount_t vtime_seqcount
;
1276 unsigned long long vtime_snap
;
1278 /* Task is sleeping or running in a CPU with VTIME inactive */
1280 /* Task runs in userspace in a CPU with VTIME active */
1282 /* Task runs in kernelspace in a CPU with VTIME active */
1284 } vtime_snap_whence
;
1287 #ifdef CONFIG_NO_HZ_FULL
1288 atomic_t tick_dep_mask
;
1290 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1291 u64 start_time
; /* monotonic time in nsec */
1292 u64 real_start_time
; /* boot based time in nsec */
1293 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1294 unsigned long min_flt
, maj_flt
;
1296 #ifdef CONFIG_POSIX_TIMERS
1297 struct task_cputime cputime_expires
;
1298 struct list_head cpu_timers
[3];
1301 /* process credentials */
1302 const struct cred __rcu
*ptracer_cred
; /* Tracer's credentials at attach */
1303 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1304 * credentials (COW) */
1305 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1306 * credentials (COW) */
1307 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1308 - access with [gs]et_task_comm (which lock
1309 it with task_lock())
1310 - initialized normally by setup_new_exec */
1311 /* file system info */
1312 struct nameidata
*nameidata
;
1313 #ifdef CONFIG_SYSVIPC
1315 struct sysv_sem sysvsem
;
1316 struct sysv_shm sysvshm
;
1318 #ifdef CONFIG_DETECT_HUNG_TASK
1319 /* hung task detection */
1320 unsigned long last_switch_count
;
1322 /* filesystem information */
1323 struct fs_struct
*fs
;
1324 /* open file information */
1325 struct files_struct
*files
;
1327 struct nsproxy
*nsproxy
;
1328 /* signal handlers */
1329 struct signal_struct
*signal
;
1330 struct sighand_struct
*sighand
;
1332 sigset_t blocked
, real_blocked
;
1333 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1334 struct sigpending pending
;
1336 unsigned long sas_ss_sp
;
1338 unsigned sas_ss_flags
;
1340 struct callback_head
*task_works
;
1342 struct audit_context
*audit_context
;
1343 #ifdef CONFIG_AUDITSYSCALL
1345 unsigned int sessionid
;
1347 struct seccomp seccomp
;
1349 /* Thread group tracking */
1352 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1354 spinlock_t alloc_lock
;
1356 /* Protection of the PI data structures: */
1357 raw_spinlock_t pi_lock
;
1359 struct wake_q_node wake_q
;
1361 #ifdef CONFIG_RT_MUTEXES
1362 /* PI waiters blocked on a rt_mutex held by this task */
1363 struct rb_root pi_waiters
;
1364 struct rb_node
*pi_waiters_leftmost
;
1365 /* Deadlock detection and priority inheritance handling */
1366 struct rt_mutex_waiter
*pi_blocked_on
;
1369 #ifdef CONFIG_DEBUG_MUTEXES
1370 /* mutex deadlock detection */
1371 struct mutex_waiter
*blocked_on
;
1373 #ifdef CONFIG_TRACE_IRQFLAGS
1374 unsigned int irq_events
;
1375 unsigned long hardirq_enable_ip
;
1376 unsigned long hardirq_disable_ip
;
1377 unsigned int hardirq_enable_event
;
1378 unsigned int hardirq_disable_event
;
1379 int hardirqs_enabled
;
1380 int hardirq_context
;
1381 unsigned long softirq_disable_ip
;
1382 unsigned long softirq_enable_ip
;
1383 unsigned int softirq_disable_event
;
1384 unsigned int softirq_enable_event
;
1385 int softirqs_enabled
;
1386 int softirq_context
;
1388 #ifdef CONFIG_LOCKDEP
1389 # define MAX_LOCK_DEPTH 48UL
1392 unsigned int lockdep_recursion
;
1393 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1394 gfp_t lockdep_reclaim_gfp
;
1397 unsigned int in_ubsan
;
1400 /* journalling filesystem info */
1403 /* stacked block device info */
1404 struct bio_list
*bio_list
;
1407 /* stack plugging */
1408 struct blk_plug
*plug
;
1412 struct reclaim_state
*reclaim_state
;
1414 struct backing_dev_info
*backing_dev_info
;
1416 struct io_context
*io_context
;
1418 unsigned long ptrace_message
;
1419 siginfo_t
*last_siginfo
; /* For ptrace use. */
1420 struct task_io_accounting ioac
;
1421 #if defined(CONFIG_TASK_XACCT)
1422 u64 acct_rss_mem1
; /* accumulated rss usage */
1423 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1424 u64 acct_timexpd
; /* stime + utime since last update */
1426 #ifdef CONFIG_CPUSETS
1427 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1428 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1429 int cpuset_mem_spread_rotor
;
1430 int cpuset_slab_spread_rotor
;
1432 #ifdef CONFIG_CGROUPS
1433 /* Control Group info protected by css_set_lock */
1434 struct css_set __rcu
*cgroups
;
1435 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1436 struct list_head cg_list
;
1438 #ifdef CONFIG_INTEL_RDT_A
1442 struct robust_list_head __user
*robust_list
;
1443 #ifdef CONFIG_COMPAT
1444 struct compat_robust_list_head __user
*compat_robust_list
;
1446 struct list_head pi_state_list
;
1447 struct futex_pi_state
*pi_state_cache
;
1449 #ifdef CONFIG_PERF_EVENTS
1450 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1451 struct mutex perf_event_mutex
;
1452 struct list_head perf_event_list
;
1454 #ifdef CONFIG_DEBUG_PREEMPT
1455 unsigned long preempt_disable_ip
;
1458 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1460 short pref_node_fork
;
1462 #ifdef CONFIG_NUMA_BALANCING
1464 unsigned int numa_scan_period
;
1465 unsigned int numa_scan_period_max
;
1466 int numa_preferred_nid
;
1467 unsigned long numa_migrate_retry
;
1468 u64 node_stamp
; /* migration stamp */
1469 u64 last_task_numa_placement
;
1470 u64 last_sum_exec_runtime
;
1471 struct callback_head numa_work
;
1473 struct list_head numa_entry
;
1474 struct numa_group
*numa_group
;
1477 * numa_faults is an array split into four regions:
1478 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1479 * in this precise order.
1481 * faults_memory: Exponential decaying average of faults on a per-node
1482 * basis. Scheduling placement decisions are made based on these
1483 * counts. The values remain static for the duration of a PTE scan.
1484 * faults_cpu: Track the nodes the process was running on when a NUMA
1485 * hinting fault was incurred.
1486 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1487 * during the current scan window. When the scan completes, the counts
1488 * in faults_memory and faults_cpu decay and these values are copied.
1490 unsigned long *numa_faults
;
1491 unsigned long total_numa_faults
;
1494 * numa_faults_locality tracks if faults recorded during the last
1495 * scan window were remote/local or failed to migrate. The task scan
1496 * period is adapted based on the locality of the faults with different
1497 * weights depending on whether they were shared or private faults
1499 unsigned long numa_faults_locality
[3];
1501 unsigned long numa_pages_migrated
;
1502 #endif /* CONFIG_NUMA_BALANCING */
1504 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1505 struct tlbflush_unmap_batch tlb_ubc
;
1508 struct rcu_head rcu
;
1511 * cache last used pipe for splice
1513 struct pipe_inode_info
*splice_pipe
;
1515 struct page_frag task_frag
;
1517 #ifdef CONFIG_TASK_DELAY_ACCT
1518 struct task_delay_info
*delays
;
1521 #ifdef CONFIG_FAULT_INJECTION
1525 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1526 * balance_dirty_pages() for some dirty throttling pause
1529 int nr_dirtied_pause
;
1530 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1532 #ifdef CONFIG_LATENCYTOP
1533 int latency_record_count
;
1534 struct latency_record latency_record
[LT_SAVECOUNT
];
1537 * time slack values; these are used to round up poll() and
1538 * select() etc timeout values. These are in nanoseconds.
1541 u64 default_timer_slack_ns
;
1544 unsigned int kasan_depth
;
1546 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1547 /* Index of current stored address in ret_stack */
1549 /* Stack of return addresses for return function tracing */
1550 struct ftrace_ret_stack
*ret_stack
;
1551 /* time stamp for last schedule */
1552 unsigned long long ftrace_timestamp
;
1554 * Number of functions that haven't been traced
1555 * because of depth overrun.
1557 atomic_t trace_overrun
;
1558 /* Pause for the tracing */
1559 atomic_t tracing_graph_pause
;
1561 #ifdef CONFIG_TRACING
1562 /* state flags for use by tracers */
1563 unsigned long trace
;
1564 /* bitmask and counter of trace recursion */
1565 unsigned long trace_recursion
;
1566 #endif /* CONFIG_TRACING */
1568 /* Coverage collection mode enabled for this task (0 if disabled). */
1569 enum kcov_mode kcov_mode
;
1570 /* Size of the kcov_area. */
1572 /* Buffer for coverage collection. */
1574 /* kcov desciptor wired with this task or NULL. */
1578 struct mem_cgroup
*memcg_in_oom
;
1579 gfp_t memcg_oom_gfp_mask
;
1580 int memcg_oom_order
;
1582 /* number of pages to reclaim on returning to userland */
1583 unsigned int memcg_nr_pages_over_high
;
1585 #ifdef CONFIG_UPROBES
1586 struct uprobe_task
*utask
;
1588 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1589 unsigned int sequential_io
;
1590 unsigned int sequential_io_avg
;
1592 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1593 unsigned long task_state_change
;
1595 int pagefault_disabled
;
1597 struct task_struct
*oom_reaper_list
;
1599 #ifdef CONFIG_VMAP_STACK
1600 struct vm_struct
*stack_vm_area
;
1602 #ifdef CONFIG_THREAD_INFO_IN_TASK
1603 /* A live task holds one reference. */
1604 atomic_t stack_refcount
;
1606 /* CPU-specific state of this task */
1607 struct thread_struct thread
;
1609 * WARNING: on x86, 'thread_struct' contains a variable-sized
1610 * structure. It *MUST* be at the end of 'task_struct'.
1612 * Do not put anything below here!
1616 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1617 extern int arch_task_struct_size __read_mostly
;
1619 # define arch_task_struct_size (sizeof(struct task_struct))
1622 #ifdef CONFIG_VMAP_STACK
1623 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1625 return t
->stack_vm_area
;
1628 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1634 #define TNF_MIGRATED 0x01
1635 #define TNF_NO_GROUP 0x02
1636 #define TNF_SHARED 0x04
1637 #define TNF_FAULT_LOCAL 0x08
1638 #define TNF_MIGRATE_FAIL 0x10
1640 static inline bool in_vfork(struct task_struct
*tsk
)
1645 * need RCU to access ->real_parent if CLONE_VM was used along with
1648 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
1651 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
1652 * ->real_parent is not necessarily the task doing vfork(), so in
1653 * theory we can't rely on task_lock() if we want to dereference it.
1655 * And in this case we can't trust the real_parent->mm == tsk->mm
1656 * check, it can be false negative. But we do not care, if init or
1657 * another oom-unkillable task does this it should blame itself.
1660 ret
= tsk
->vfork_done
&& tsk
->real_parent
->mm
== tsk
->mm
;
1666 #ifdef CONFIG_NUMA_BALANCING
1667 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
1668 extern pid_t
task_numa_group_id(struct task_struct
*p
);
1669 extern void set_numabalancing_state(bool enabled
);
1670 extern void task_numa_free(struct task_struct
*p
);
1671 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
1672 int src_nid
, int dst_cpu
);
1674 static inline void task_numa_fault(int last_node
, int node
, int pages
,
1678 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
1682 static inline void set_numabalancing_state(bool enabled
)
1685 static inline void task_numa_free(struct task_struct
*p
)
1688 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
1689 struct page
*page
, int src_nid
, int dst_cpu
)
1695 static inline struct pid
*task_pid(struct task_struct
*task
)
1697 return task
->pids
[PIDTYPE_PID
].pid
;
1700 static inline struct pid
*task_tgid(struct task_struct
*task
)
1702 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1706 * Without tasklist or rcu lock it is not safe to dereference
1707 * the result of task_pgrp/task_session even if task == current,
1708 * we can race with another thread doing sys_setsid/sys_setpgid.
1710 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1712 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1715 static inline struct pid
*task_session(struct task_struct
*task
)
1717 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1720 struct pid_namespace
;
1723 * the helpers to get the task's different pids as they are seen
1724 * from various namespaces
1726 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1727 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1729 * task_xid_nr_ns() : id seen from the ns specified;
1731 * set_task_vxid() : assigns a virtual id to a task;
1733 * see also pid_nr() etc in include/linux/pid.h
1735 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1736 struct pid_namespace
*ns
);
1738 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1743 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1744 struct pid_namespace
*ns
)
1746 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1749 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1751 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1755 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1760 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1762 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1764 return pid_vnr(task_tgid(tsk
));
1768 static inline int pid_alive(const struct task_struct
*p
);
1769 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1775 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1781 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1783 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
1786 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1787 struct pid_namespace
*ns
)
1789 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1792 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1794 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1798 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1799 struct pid_namespace
*ns
)
1801 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1804 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1806 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1809 /* obsolete, do not use */
1810 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1812 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1816 * pid_alive - check that a task structure is not stale
1817 * @p: Task structure to be checked.
1819 * Test if a process is not yet dead (at most zombie state)
1820 * If pid_alive fails, then pointers within the task structure
1821 * can be stale and must not be dereferenced.
1823 * Return: 1 if the process is alive. 0 otherwise.
1825 static inline int pid_alive(const struct task_struct
*p
)
1827 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1831 * is_global_init - check if a task structure is init. Since init
1832 * is free to have sub-threads we need to check tgid.
1833 * @tsk: Task structure to be checked.
1835 * Check if a task structure is the first user space task the kernel created.
1837 * Return: 1 if the task structure is init. 0 otherwise.
1839 static inline int is_global_init(struct task_struct
*tsk
)
1841 return task_tgid_nr(tsk
) == 1;
1844 extern struct pid
*cad_pid
;
1846 extern void free_task(struct task_struct
*tsk
);
1847 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1849 extern void __put_task_struct(struct task_struct
*t
);
1851 static inline void put_task_struct(struct task_struct
*t
)
1853 if (atomic_dec_and_test(&t
->usage
))
1854 __put_task_struct(t
);
1857 struct task_struct
*task_rcu_dereference(struct task_struct
**ptask
);
1858 struct task_struct
*try_get_task_struct(struct task_struct
**ptask
);
1860 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1861 extern void task_cputime(struct task_struct
*t
,
1862 u64
*utime
, u64
*stime
);
1863 extern u64
task_gtime(struct task_struct
*t
);
1865 static inline void task_cputime(struct task_struct
*t
,
1866 u64
*utime
, u64
*stime
)
1872 static inline u64
task_gtime(struct task_struct
*t
)
1878 #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME
1879 static inline void task_cputime_scaled(struct task_struct
*t
,
1883 *utimescaled
= t
->utimescaled
;
1884 *stimescaled
= t
->stimescaled
;
1887 static inline void task_cputime_scaled(struct task_struct
*t
,
1891 task_cputime(t
, utimescaled
, stimescaled
);
1895 extern void task_cputime_adjusted(struct task_struct
*p
, u64
*ut
, u64
*st
);
1896 extern void thread_group_cputime_adjusted(struct task_struct
*p
, u64
*ut
, u64
*st
);
1901 #define PF_IDLE 0x00000002 /* I am an IDLE thread */
1902 #define PF_EXITING 0x00000004 /* getting shut down */
1903 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1904 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1905 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1906 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1907 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1908 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1909 #define PF_DUMPCORE 0x00000200 /* dumped core */
1910 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1911 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1912 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1913 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1914 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1915 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1916 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1917 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1918 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1919 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1920 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1921 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1922 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1923 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1924 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1925 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1926 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1927 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1928 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1931 * Only the _current_ task can read/write to tsk->flags, but other
1932 * tasks can access tsk->flags in readonly mode for example
1933 * with tsk_used_math (like during threaded core dumping).
1934 * There is however an exception to this rule during ptrace
1935 * or during fork: the ptracer task is allowed to write to the
1936 * child->flags of its traced child (same goes for fork, the parent
1937 * can write to the child->flags), because we're guaranteed the
1938 * child is not running and in turn not changing child->flags
1939 * at the same time the parent does it.
1941 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1942 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1943 #define clear_used_math() clear_stopped_child_used_math(current)
1944 #define set_used_math() set_stopped_child_used_math(current)
1945 #define conditional_stopped_child_used_math(condition, child) \
1946 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1947 #define conditional_used_math(condition) \
1948 conditional_stopped_child_used_math(condition, current)
1949 #define copy_to_stopped_child_used_math(child) \
1950 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1951 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1952 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1953 #define used_math() tsk_used_math(current)
1955 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
1956 * __GFP_FS is also cleared as it implies __GFP_IO.
1958 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
1960 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
1961 flags
&= ~(__GFP_IO
| __GFP_FS
);
1965 static inline unsigned int memalloc_noio_save(void)
1967 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
1968 current
->flags
|= PF_MEMALLOC_NOIO
;
1972 static inline void memalloc_noio_restore(unsigned int flags
)
1974 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
1977 /* Per-process atomic flags. */
1978 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
1979 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
1980 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
1981 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
1984 #define TASK_PFA_TEST(name, func) \
1985 static inline bool task_##func(struct task_struct *p) \
1986 { return test_bit(PFA_##name, &p->atomic_flags); }
1987 #define TASK_PFA_SET(name, func) \
1988 static inline void task_set_##func(struct task_struct *p) \
1989 { set_bit(PFA_##name, &p->atomic_flags); }
1990 #define TASK_PFA_CLEAR(name, func) \
1991 static inline void task_clear_##func(struct task_struct *p) \
1992 { clear_bit(PFA_##name, &p->atomic_flags); }
1994 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
1995 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
1997 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
1998 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
1999 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2001 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2002 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2003 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2005 TASK_PFA_TEST(LMK_WAITING
, lmk_waiting
)
2006 TASK_PFA_SET(LMK_WAITING
, lmk_waiting
)
2009 * task->jobctl flags
2011 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2013 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2014 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2015 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2016 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2017 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2018 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2019 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2021 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2022 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2023 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2024 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2025 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2026 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2027 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2029 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2030 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2032 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2033 unsigned long mask
);
2034 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2035 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2036 unsigned long mask
);
2038 static inline void rcu_copy_process(struct task_struct
*p
)
2040 #ifdef CONFIG_PREEMPT_RCU
2041 p
->rcu_read_lock_nesting
= 0;
2042 p
->rcu_read_unlock_special
.s
= 0;
2043 p
->rcu_blocked_node
= NULL
;
2044 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2045 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2046 #ifdef CONFIG_TASKS_RCU
2047 p
->rcu_tasks_holdout
= false;
2048 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2049 p
->rcu_tasks_idle_cpu
= -1;
2050 #endif /* #ifdef CONFIG_TASKS_RCU */
2053 static inline void tsk_restore_flags(struct task_struct
*task
,
2054 unsigned long orig_flags
, unsigned long flags
)
2056 task
->flags
&= ~flags
;
2057 task
->flags
|= orig_flags
& flags
;
2060 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2061 const struct cpumask
*trial
);
2062 extern int task_can_attach(struct task_struct
*p
,
2063 const struct cpumask
*cs_cpus_allowed
);
2065 extern void do_set_cpus_allowed(struct task_struct
*p
,
2066 const struct cpumask
*new_mask
);
2068 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2069 const struct cpumask
*new_mask
);
2071 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2072 const struct cpumask
*new_mask
)
2075 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2076 const struct cpumask
*new_mask
)
2078 if (!cpumask_test_cpu(0, new_mask
))
2084 #ifdef CONFIG_NO_HZ_COMMON
2085 void calc_load_enter_idle(void);
2086 void calc_load_exit_idle(void);
2088 static inline void calc_load_enter_idle(void) { }
2089 static inline void calc_load_exit_idle(void) { }
2090 #endif /* CONFIG_NO_HZ_COMMON */
2092 #ifndef cpu_relax_yield
2093 #define cpu_relax_yield() cpu_relax()
2096 extern unsigned long long
2097 task_sched_runtime(struct task_struct
*task
);
2099 /* sched_exec is called by processes performing an exec */
2101 extern void sched_exec(void);
2103 #define sched_exec() {}
2106 #ifdef CONFIG_HOTPLUG_CPU
2107 extern void idle_task_exit(void);
2109 static inline void idle_task_exit(void) {}
2112 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2113 extern void wake_up_nohz_cpu(int cpu
);
2115 static inline void wake_up_nohz_cpu(int cpu
) { }
2118 #ifdef CONFIG_NO_HZ_FULL
2119 extern u64
scheduler_tick_max_deferment(void);
2122 #ifdef CONFIG_SCHED_AUTOGROUP
2123 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2124 extern void sched_autogroup_detach(struct task_struct
*p
);
2125 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2126 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2127 extern void sched_autogroup_exit_task(struct task_struct
*p
);
2128 #ifdef CONFIG_PROC_FS
2129 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2130 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2133 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2134 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2135 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2136 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2137 static inline void sched_autogroup_exit_task(struct task_struct
*p
) { }
2140 extern int yield_to(struct task_struct
*p
, bool preempt
);
2141 extern void set_user_nice(struct task_struct
*p
, long nice
);
2142 extern int task_prio(const struct task_struct
*p
);
2144 * task_nice - return the nice value of a given task.
2145 * @p: the task in question.
2147 * Return: The nice value [ -20 ... 0 ... 19 ].
2149 static inline int task_nice(const struct task_struct
*p
)
2151 return PRIO_TO_NICE((p
)->static_prio
);
2153 extern int can_nice(const struct task_struct
*p
, const int nice
);
2154 extern int task_curr(const struct task_struct
*p
);
2155 extern int idle_cpu(int cpu
);
2156 extern int sched_setscheduler(struct task_struct
*, int,
2157 const struct sched_param
*);
2158 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2159 const struct sched_param
*);
2160 extern int sched_setattr(struct task_struct
*,
2161 const struct sched_attr
*);
2162 extern struct task_struct
*idle_task(int cpu
);
2164 * is_idle_task - is the specified task an idle task?
2165 * @p: the task in question.
2167 * Return: 1 if @p is an idle task. 0 otherwise.
2169 static inline bool is_idle_task(const struct task_struct
*p
)
2171 return !!(p
->flags
& PF_IDLE
);
2173 extern struct task_struct
*curr_task(int cpu
);
2174 extern void ia64_set_curr_task(int cpu
, struct task_struct
*p
);
2178 union thread_union
{
2179 #ifndef CONFIG_THREAD_INFO_IN_TASK
2180 struct thread_info thread_info
;
2182 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2185 #ifndef __HAVE_ARCH_KSTACK_END
2186 static inline int kstack_end(void *addr
)
2188 /* Reliable end of stack detection:
2189 * Some APM bios versions misalign the stack
2191 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2195 extern union thread_union init_thread_union
;
2196 extern struct task_struct init_task
;
2198 extern struct mm_struct init_mm
;
2200 extern struct pid_namespace init_pid_ns
;
2203 * find a task by one of its numerical ids
2205 * find_task_by_pid_ns():
2206 * finds a task by its pid in the specified namespace
2207 * find_task_by_vpid():
2208 * finds a task by its virtual pid
2210 * see also find_vpid() etc in include/linux/pid.h
2213 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2214 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2215 struct pid_namespace
*ns
);
2217 /* per-UID process charging. */
2218 extern struct user_struct
* alloc_uid(kuid_t
);
2219 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2221 atomic_inc(&u
->__count
);
2224 extern void free_uid(struct user_struct
*);
2226 #include <asm/current.h>
2228 extern void xtime_update(unsigned long ticks
);
2230 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2231 extern int wake_up_process(struct task_struct
*tsk
);
2232 extern void wake_up_new_task(struct task_struct
*tsk
);
2234 extern void kick_process(struct task_struct
*tsk
);
2236 static inline void kick_process(struct task_struct
*tsk
) { }
2238 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2239 extern void sched_dead(struct task_struct
*p
);
2241 extern void proc_caches_init(void);
2242 extern void flush_signals(struct task_struct
*);
2243 extern void ignore_signals(struct task_struct
*);
2244 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2245 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2247 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2249 struct task_struct
*tsk
= current
;
2253 spin_lock_irq(&tsk
->sighand
->siglock
);
2254 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2255 spin_unlock_irq(&tsk
->sighand
->siglock
);
2260 static inline void kernel_signal_stop(void)
2262 spin_lock_irq(¤t
->sighand
->siglock
);
2263 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2264 __set_current_state(TASK_STOPPED
);
2265 spin_unlock_irq(¤t
->sighand
->siglock
);
2270 extern void release_task(struct task_struct
* p
);
2271 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2272 extern int force_sigsegv(int, struct task_struct
*);
2273 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2274 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2275 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2276 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2277 const struct cred
*, u32
);
2278 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2279 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2280 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2281 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2282 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2283 extern void force_sig(int, struct task_struct
*);
2284 extern int send_sig(int, struct task_struct
*, int);
2285 extern int zap_other_threads(struct task_struct
*p
);
2286 extern struct sigqueue
*sigqueue_alloc(void);
2287 extern void sigqueue_free(struct sigqueue
*);
2288 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2289 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2291 #ifdef TIF_RESTORE_SIGMASK
2293 * Legacy restore_sigmask accessors. These are inefficient on
2294 * SMP architectures because they require atomic operations.
2298 * set_restore_sigmask() - make sure saved_sigmask processing gets done
2300 * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2301 * will run before returning to user mode, to process the flag. For
2302 * all callers, TIF_SIGPENDING is already set or it's no harm to set
2303 * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
2304 * arch code will notice on return to user mode, in case those bits
2305 * are scarce. We set TIF_SIGPENDING here to ensure that the arch
2306 * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2308 static inline void set_restore_sigmask(void)
2310 set_thread_flag(TIF_RESTORE_SIGMASK
);
2311 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2313 static inline void clear_restore_sigmask(void)
2315 clear_thread_flag(TIF_RESTORE_SIGMASK
);
2317 static inline bool test_restore_sigmask(void)
2319 return test_thread_flag(TIF_RESTORE_SIGMASK
);
2321 static inline bool test_and_clear_restore_sigmask(void)
2323 return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK
);
2326 #else /* TIF_RESTORE_SIGMASK */
2328 /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2329 static inline void set_restore_sigmask(void)
2331 current
->restore_sigmask
= true;
2332 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2334 static inline void clear_restore_sigmask(void)
2336 current
->restore_sigmask
= false;
2338 static inline bool test_restore_sigmask(void)
2340 return current
->restore_sigmask
;
2342 static inline bool test_and_clear_restore_sigmask(void)
2344 if (!current
->restore_sigmask
)
2346 current
->restore_sigmask
= false;
2351 static inline void restore_saved_sigmask(void)
2353 if (test_and_clear_restore_sigmask())
2354 __set_current_blocked(¤t
->saved_sigmask
);
2357 static inline sigset_t
*sigmask_to_save(void)
2359 sigset_t
*res
= ¤t
->blocked
;
2360 if (unlikely(test_restore_sigmask()))
2361 res
= ¤t
->saved_sigmask
;
2365 static inline int kill_cad_pid(int sig
, int priv
)
2367 return kill_pid(cad_pid
, sig
, priv
);
2370 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2371 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2372 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2373 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2376 * True if we are on the alternate signal stack.
2378 static inline int on_sig_stack(unsigned long sp
)
2381 * If the signal stack is SS_AUTODISARM then, by construction, we
2382 * can't be on the signal stack unless user code deliberately set
2383 * SS_AUTODISARM when we were already on it.
2385 * This improves reliability: if user state gets corrupted such that
2386 * the stack pointer points very close to the end of the signal stack,
2387 * then this check will enable the signal to be handled anyway.
2389 if (current
->sas_ss_flags
& SS_AUTODISARM
)
2392 #ifdef CONFIG_STACK_GROWSUP
2393 return sp
>= current
->sas_ss_sp
&&
2394 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2396 return sp
> current
->sas_ss_sp
&&
2397 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2401 static inline int sas_ss_flags(unsigned long sp
)
2403 if (!current
->sas_ss_size
)
2406 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2409 static inline void sas_ss_reset(struct task_struct
*p
)
2413 p
->sas_ss_flags
= SS_DISABLE
;
2416 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2418 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2419 #ifdef CONFIG_STACK_GROWSUP
2420 return current
->sas_ss_sp
;
2422 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2428 * Routines for handling mm_structs
2430 extern struct mm_struct
* mm_alloc(void);
2433 * mmgrab() - Pin a &struct mm_struct.
2434 * @mm: The &struct mm_struct to pin.
2436 * Make sure that @mm will not get freed even after the owning task
2437 * exits. This doesn't guarantee that the associated address space
2438 * will still exist later on and mmget_not_zero() has to be used before
2441 * This is a preferred way to to pin @mm for a longer/unbounded amount
2444 * Use mmdrop() to release the reference acquired by mmgrab().
2446 * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
2447 * of &mm_struct.mm_count vs &mm_struct.mm_users.
2449 static inline void mmgrab(struct mm_struct
*mm
)
2451 atomic_inc(&mm
->mm_count
);
2454 /* mmdrop drops the mm and the page tables */
2455 extern void __mmdrop(struct mm_struct
*);
2456 static inline void mmdrop(struct mm_struct
*mm
)
2458 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2462 static inline void mmdrop_async_fn(struct work_struct
*work
)
2464 struct mm_struct
*mm
= container_of(work
, struct mm_struct
, async_put_work
);
2468 static inline void mmdrop_async(struct mm_struct
*mm
)
2470 if (unlikely(atomic_dec_and_test(&mm
->mm_count
))) {
2471 INIT_WORK(&mm
->async_put_work
, mmdrop_async_fn
);
2472 schedule_work(&mm
->async_put_work
);
2477 * mmget() - Pin the address space associated with a &struct mm_struct.
2478 * @mm: The address space to pin.
2480 * Make sure that the address space of the given &struct mm_struct doesn't
2481 * go away. This does not protect against parts of the address space being
2482 * modified or freed, however.
2484 * Never use this function to pin this address space for an
2485 * unbounded/indefinite amount of time.
2487 * Use mmput() to release the reference acquired by mmget().
2489 * See also <Documentation/vm/active_mm.txt> for an in-depth explanation
2490 * of &mm_struct.mm_count vs &mm_struct.mm_users.
2492 static inline void mmget(struct mm_struct
*mm
)
2494 atomic_inc(&mm
->mm_users
);
2497 static inline bool mmget_not_zero(struct mm_struct
*mm
)
2499 return atomic_inc_not_zero(&mm
->mm_users
);
2502 /* mmput gets rid of the mappings and all user-space */
2503 extern void mmput(struct mm_struct
*);
2505 /* same as above but performs the slow path from the async context. Can
2506 * be called from the atomic context as well
2508 extern void mmput_async(struct mm_struct
*);
2511 /* Grab a reference to a task's mm, if it is not already going away */
2512 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2514 * Grab a reference to a task's mm, if it is not already going away
2515 * and ptrace_may_access with the mode parameter passed to it
2518 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2519 /* Remove the current tasks stale references to the old mm_struct */
2520 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2522 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2523 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2524 struct task_struct
*, unsigned long);
2526 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2527 struct task_struct
*);
2529 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2530 * via pt_regs, so ignore the tls argument passed via C. */
2531 static inline int copy_thread_tls(
2532 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2533 struct task_struct
*p
, unsigned long tls
)
2535 return copy_thread(clone_flags
, sp
, arg
, p
);
2538 extern void flush_thread(void);
2540 #ifdef CONFIG_HAVE_EXIT_THREAD
2541 extern void exit_thread(struct task_struct
*tsk
);
2543 static inline void exit_thread(struct task_struct
*tsk
)
2548 extern void exit_files(struct task_struct
*);
2549 extern void __cleanup_sighand(struct sighand_struct
*);
2551 extern void exit_itimers(struct signal_struct
*);
2552 extern void flush_itimer_signals(void);
2554 extern void do_group_exit(int);
2556 extern int do_execve(struct filename
*,
2557 const char __user
* const __user
*,
2558 const char __user
* const __user
*);
2559 extern int do_execveat(int, struct filename
*,
2560 const char __user
* const __user
*,
2561 const char __user
* const __user
*,
2563 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
2564 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2565 struct task_struct
*fork_idle(int);
2566 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2568 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
2569 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
2571 __set_task_comm(tsk
, from
, false);
2573 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2576 void scheduler_ipi(void);
2577 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2579 static inline void scheduler_ipi(void) { }
2580 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2587 #define tasklist_empty() \
2588 list_empty(&init_task.tasks)
2590 #define next_task(p) \
2591 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2593 #define for_each_process(p) \
2594 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2596 extern bool current_is_single_threaded(void);
2599 * Careful: do_each_thread/while_each_thread is a double loop so
2600 * 'break' will not work as expected - use goto instead.
2602 #define do_each_thread(g, t) \
2603 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2605 #define while_each_thread(g, t) \
2606 while ((t = next_thread(t)) != g)
2608 #define __for_each_thread(signal, t) \
2609 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2611 #define for_each_thread(p, t) \
2612 __for_each_thread((p)->signal, t)
2614 /* Careful: this is a double loop, 'break' won't work as expected. */
2615 #define for_each_process_thread(p, t) \
2616 for_each_process(p) for_each_thread(p, t)
2618 typedef int (*proc_visitor
)(struct task_struct
*p
, void *data
);
2619 void walk_process_tree(struct task_struct
*top
, proc_visitor
, void *);
2621 static inline int get_nr_threads(struct task_struct
*tsk
)
2623 return tsk
->signal
->nr_threads
;
2626 static inline bool thread_group_leader(struct task_struct
*p
)
2628 return p
->exit_signal
>= 0;
2631 /* Do to the insanities of de_thread it is possible for a process
2632 * to have the pid of the thread group leader without actually being
2633 * the thread group leader. For iteration through the pids in proc
2634 * all we care about is that we have a task with the appropriate
2635 * pid, we don't actually care if we have the right task.
2637 static inline bool has_group_leader_pid(struct task_struct
*p
)
2639 return task_pid(p
) == p
->signal
->leader_pid
;
2643 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2645 return p1
->signal
== p2
->signal
;
2648 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2650 return list_entry_rcu(p
->thread_group
.next
,
2651 struct task_struct
, thread_group
);
2654 static inline int thread_group_empty(struct task_struct
*p
)
2656 return list_empty(&p
->thread_group
);
2659 #define delay_group_leader(p) \
2660 (thread_group_leader(p) && !thread_group_empty(p))
2663 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2664 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2665 * pins the final release of task.io_context. Also protects ->cpuset and
2666 * ->cgroup.subsys[]. And ->vfork_done.
2668 * Nests both inside and outside of read_lock(&tasklist_lock).
2669 * It must not be nested with write_lock_irq(&tasklist_lock),
2670 * neither inside nor outside.
2672 static inline void task_lock(struct task_struct
*p
)
2674 spin_lock(&p
->alloc_lock
);
2677 static inline void task_unlock(struct task_struct
*p
)
2679 spin_unlock(&p
->alloc_lock
);
2682 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2683 unsigned long *flags
);
2685 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2686 unsigned long *flags
)
2688 struct sighand_struct
*ret
;
2690 ret
= __lock_task_sighand(tsk
, flags
);
2691 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2695 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2696 unsigned long *flags
)
2698 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2701 #ifdef CONFIG_THREAD_INFO_IN_TASK
2703 static inline struct thread_info
*task_thread_info(struct task_struct
*task
)
2705 return &task
->thread_info
;
2709 * When accessing the stack of a non-current task that might exit, use
2710 * try_get_task_stack() instead. task_stack_page will return a pointer
2711 * that could get freed out from under you.
2713 static inline void *task_stack_page(const struct task_struct
*task
)
2718 #define setup_thread_stack(new,old) do { } while(0)
2720 static inline unsigned long *end_of_stack(const struct task_struct
*task
)
2725 #elif !defined(__HAVE_THREAD_FUNCTIONS)
2727 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2728 #define task_stack_page(task) ((void *)(task)->stack)
2730 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2732 *task_thread_info(p
) = *task_thread_info(org
);
2733 task_thread_info(p
)->task
= p
;
2737 * Return the address of the last usable long on the stack.
2739 * When the stack grows down, this is just above the thread
2740 * info struct. Going any lower will corrupt the threadinfo.
2742 * When the stack grows up, this is the highest address.
2743 * Beyond that position, we corrupt data on the next page.
2745 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2747 #ifdef CONFIG_STACK_GROWSUP
2748 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
2750 return (unsigned long *)(task_thread_info(p
) + 1);
2756 #ifdef CONFIG_THREAD_INFO_IN_TASK
2757 static inline void *try_get_task_stack(struct task_struct
*tsk
)
2759 return atomic_inc_not_zero(&tsk
->stack_refcount
) ?
2760 task_stack_page(tsk
) : NULL
;
2763 extern void put_task_stack(struct task_struct
*tsk
);
2765 static inline void *try_get_task_stack(struct task_struct
*tsk
)
2767 return task_stack_page(tsk
);
2770 static inline void put_task_stack(struct task_struct
*tsk
) {}
2773 #define task_stack_end_corrupted(task) \
2774 (*(end_of_stack(task)) != STACK_END_MAGIC)
2776 static inline int object_is_on_stack(void *obj
)
2778 void *stack
= task_stack_page(current
);
2780 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2783 extern void thread_stack_cache_init(void);
2785 #ifdef CONFIG_DEBUG_STACK_USAGE
2786 static inline unsigned long stack_not_used(struct task_struct
*p
)
2788 unsigned long *n
= end_of_stack(p
);
2790 do { /* Skip over canary */
2791 # ifdef CONFIG_STACK_GROWSUP
2798 # ifdef CONFIG_STACK_GROWSUP
2799 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
2801 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2805 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
2807 /* set thread flags in other task's structures
2808 * - see asm/thread_info.h for TIF_xxxx flags available
2810 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2812 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2815 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2817 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2820 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2822 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2825 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2827 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2830 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2832 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2835 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2837 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2840 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2842 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2845 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2847 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2850 static inline int restart_syscall(void)
2852 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2853 return -ERESTARTNOINTR
;
2856 static inline int signal_pending(struct task_struct
*p
)
2858 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2861 static inline int __fatal_signal_pending(struct task_struct
*p
)
2863 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2866 static inline int fatal_signal_pending(struct task_struct
*p
)
2868 return signal_pending(p
) && __fatal_signal_pending(p
);
2871 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2873 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2875 if (!signal_pending(p
))
2878 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2882 * cond_resched() and cond_resched_lock(): latency reduction via
2883 * explicit rescheduling in places that are safe. The return
2884 * value indicates whether a reschedule was done in fact.
2885 * cond_resched_lock() will drop the spinlock before scheduling,
2886 * cond_resched_softirq() will enable bhs before scheduling.
2888 #ifndef CONFIG_PREEMPT
2889 extern int _cond_resched(void);
2891 static inline int _cond_resched(void) { return 0; }
2894 #define cond_resched() ({ \
2895 ___might_sleep(__FILE__, __LINE__, 0); \
2899 extern int __cond_resched_lock(spinlock_t
*lock
);
2901 #define cond_resched_lock(lock) ({ \
2902 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2903 __cond_resched_lock(lock); \
2906 extern int __cond_resched_softirq(void);
2908 #define cond_resched_softirq() ({ \
2909 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2910 __cond_resched_softirq(); \
2913 static inline void cond_resched_rcu(void)
2915 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2923 * Does a critical section need to be broken due to another
2924 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2925 * but a general need for low latency)
2927 static inline int spin_needbreak(spinlock_t
*lock
)
2929 #ifdef CONFIG_PREEMPT
2930 return spin_is_contended(lock
);
2936 static __always_inline
bool need_resched(void)
2938 return unlikely(tif_need_resched());
2942 * Thread group CPU time accounting.
2944 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2945 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2948 * Reevaluate whether the task has signals pending delivery.
2949 * Wake the task if so.
2950 * This is required every time the blocked sigset_t changes.
2951 * callers must hold sighand->siglock.
2953 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2954 extern void recalc_sigpending(void);
2956 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
2958 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
2960 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
2962 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
2964 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
2968 * Wrappers for p->thread_info->cpu access. No-op on UP.
2972 static inline unsigned int task_cpu(const struct task_struct
*p
)
2974 #ifdef CONFIG_THREAD_INFO_IN_TASK
2977 return task_thread_info(p
)->cpu
;
2981 static inline int task_node(const struct task_struct
*p
)
2983 return cpu_to_node(task_cpu(p
));
2986 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2990 static inline unsigned int task_cpu(const struct task_struct
*p
)
2995 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2999 #endif /* CONFIG_SMP */
3002 * In order to reduce various lock holder preemption latencies provide an
3003 * interface to see if a vCPU is currently running or not.
3005 * This allows us to terminate optimistic spin loops and block, analogous to
3006 * the native optimistic spin heuristic of testing if the lock owner task is
3009 #ifndef vcpu_is_preempted
3010 # define vcpu_is_preempted(cpu) false
3013 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3014 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3016 #ifdef CONFIG_CGROUP_SCHED
3017 extern struct task_group root_task_group
;
3018 #endif /* CONFIG_CGROUP_SCHED */
3020 extern int task_can_switch_user(struct user_struct
*up
,
3021 struct task_struct
*tsk
);
3023 #ifdef CONFIG_TASK_XACCT
3024 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3026 tsk
->ioac
.rchar
+= amt
;
3029 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3031 tsk
->ioac
.wchar
+= amt
;
3034 static inline void inc_syscr(struct task_struct
*tsk
)
3039 static inline void inc_syscw(struct task_struct
*tsk
)
3044 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3048 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3052 static inline void inc_syscr(struct task_struct
*tsk
)
3056 static inline void inc_syscw(struct task_struct
*tsk
)
3061 #ifndef TASK_SIZE_OF
3062 #define TASK_SIZE_OF(tsk) TASK_SIZE
3066 extern void mm_update_next_owner(struct mm_struct
*mm
);
3068 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3071 #endif /* CONFIG_MEMCG */
3073 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3076 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3079 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3082 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3085 static inline unsigned long rlimit(unsigned int limit
)
3087 return task_rlimit(current
, limit
);
3090 static inline unsigned long rlimit_max(unsigned int limit
)
3092 return task_rlimit_max(current
, limit
);
3095 #define SCHED_CPUFREQ_RT (1U << 0)
3096 #define SCHED_CPUFREQ_DL (1U << 1)
3097 #define SCHED_CPUFREQ_IOWAIT (1U << 2)
3099 #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
3101 #ifdef CONFIG_CPU_FREQ
3102 struct update_util_data
{
3103 void (*func
)(struct update_util_data
*data
, u64 time
, unsigned int flags
);
3106 void cpufreq_add_update_util_hook(int cpu
, struct update_util_data
*data
,
3107 void (*func
)(struct update_util_data
*data
, u64 time
,
3108 unsigned int flags
));
3109 void cpufreq_remove_update_util_hook(int cpu
);
3110 #endif /* CONFIG_CPU_FREQ */