4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/kcov.h>
55 #include <linux/task_io_accounting.h>
56 #include <linux/latencytop.h>
57 #include <linux/cred.h>
58 #include <linux/llist.h>
59 #include <linux/uidgid.h>
60 #include <linux/gfp.h>
61 #include <linux/magic.h>
62 #include <linux/cgroup-defs.h>
64 #include <asm/processor.h>
66 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
69 * Extended scheduling parameters data structure.
71 * This is needed because the original struct sched_param can not be
72 * altered without introducing ABI issues with legacy applications
73 * (e.g., in sched_getparam()).
75 * However, the possibility of specifying more than just a priority for
76 * the tasks may be useful for a wide variety of application fields, e.g.,
77 * multimedia, streaming, automation and control, and many others.
79 * This variant (sched_attr) is meant at describing a so-called
80 * sporadic time-constrained task. In such model a task is specified by:
81 * - the activation period or minimum instance inter-arrival time;
82 * - the maximum (or average, depending on the actual scheduling
83 * discipline) computation time of all instances, a.k.a. runtime;
84 * - the deadline (relative to the actual activation time) of each
86 * Very briefly, a periodic (sporadic) task asks for the execution of
87 * some specific computation --which is typically called an instance--
88 * (at most) every period. Moreover, each instance typically lasts no more
89 * than the runtime and must be completed by time instant t equal to
90 * the instance activation time + the deadline.
92 * This is reflected by the actual fields of the sched_attr structure:
94 * @size size of the structure, for fwd/bwd compat.
96 * @sched_policy task's scheduling policy
97 * @sched_flags for customizing the scheduler behaviour
98 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
99 * @sched_priority task's static priority (SCHED_FIFO/RR)
100 * @sched_deadline representative of the task's deadline
101 * @sched_runtime representative of the task's runtime
102 * @sched_period representative of the task's period
104 * Given this task model, there are a multiplicity of scheduling algorithms
105 * and policies, that can be used to ensure all the tasks will make their
106 * timing constraints.
108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
109 * only user of this new interface. More information about the algorithm
110 * available in the scheduling class file or in Documentation/.
118 /* SCHED_NORMAL, SCHED_BATCH */
121 /* SCHED_FIFO, SCHED_RR */
130 struct futex_pi_state
;
131 struct robust_list_head
;
134 struct perf_event_context
;
139 #define VMACACHE_BITS 2
140 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
141 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
144 * These are the constant used to fake the fixed-point load-average
145 * counting. Some notes:
146 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
147 * a load-average precision of 10 bits integer + 11 bits fractional
148 * - if you want to count load-averages more often, you need more
149 * precision, or rounding will get you. With 2-second counting freq,
150 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 extern unsigned long avenrun
[]; /* Load averages */
154 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
156 #define FSHIFT 11 /* nr of bits of precision */
157 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
158 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
159 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
160 #define EXP_5 2014 /* 1/exp(5sec/5min) */
161 #define EXP_15 2037 /* 1/exp(5sec/15min) */
163 #define CALC_LOAD(load,exp,n) \
165 load += n*(FIXED_1-exp); \
168 extern unsigned long total_forks
;
169 extern int nr_threads
;
170 DECLARE_PER_CPU(unsigned long, process_counts
);
171 extern int nr_processes(void);
172 extern unsigned long nr_running(void);
173 extern bool single_task_running(void);
174 extern unsigned long nr_iowait(void);
175 extern unsigned long nr_iowait_cpu(int cpu
);
176 extern void get_iowait_load(unsigned long *nr_waiters
, unsigned long *load
);
178 extern void calc_global_load(unsigned long ticks
);
180 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
181 extern void update_cpu_load_nohz(int active
);
183 static inline void update_cpu_load_nohz(int active
) { }
186 extern void dump_cpu_task(int cpu
);
191 #ifdef CONFIG_SCHED_DEBUG
192 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
193 extern void proc_sched_set_task(struct task_struct
*p
);
197 * Task state bitmask. NOTE! These bits are also
198 * encoded in fs/proc/array.c: get_task_state().
200 * We have two separate sets of flags: task->state
201 * is about runnability, while task->exit_state are
202 * about the task exiting. Confusing, but this way
203 * modifying one set can't modify the other one by
206 #define TASK_RUNNING 0
207 #define TASK_INTERRUPTIBLE 1
208 #define TASK_UNINTERRUPTIBLE 2
209 #define __TASK_STOPPED 4
210 #define __TASK_TRACED 8
211 /* in tsk->exit_state */
213 #define EXIT_ZOMBIE 32
214 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
215 /* in tsk->state again */
217 #define TASK_WAKEKILL 128
218 #define TASK_WAKING 256
219 #define TASK_PARKED 512
220 #define TASK_NOLOAD 1024
221 #define TASK_STATE_MAX 2048
223 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
225 extern char ___assert_task_state
[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
233 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
235 /* Convenience macros for the sake of wake_up */
236 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
237 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
239 /* get_task_state() */
240 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
241 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
242 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
244 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
245 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
246 #define task_is_stopped_or_traced(task) \
247 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
248 #define task_contributes_to_load(task) \
249 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
250 (task->flags & PF_FROZEN) == 0 && \
251 (task->state & TASK_NOLOAD) == 0)
253 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
255 #define __set_task_state(tsk, state_value) \
257 (tsk)->task_state_change = _THIS_IP_; \
258 (tsk)->state = (state_value); \
260 #define set_task_state(tsk, state_value) \
262 (tsk)->task_state_change = _THIS_IP_; \
263 smp_store_mb((tsk)->state, (state_value)); \
267 * set_current_state() includes a barrier so that the write of current->state
268 * is correctly serialised wrt the caller's subsequent test of whether to
271 * set_current_state(TASK_UNINTERRUPTIBLE);
272 * if (do_i_need_to_sleep())
275 * If the caller does not need such serialisation then use __set_current_state()
277 #define __set_current_state(state_value) \
279 current->task_state_change = _THIS_IP_; \
280 current->state = (state_value); \
282 #define set_current_state(state_value) \
284 current->task_state_change = _THIS_IP_; \
285 smp_store_mb(current->state, (state_value)); \
290 #define __set_task_state(tsk, state_value) \
291 do { (tsk)->state = (state_value); } while (0)
292 #define set_task_state(tsk, state_value) \
293 smp_store_mb((tsk)->state, (state_value))
296 * set_current_state() includes a barrier so that the write of current->state
297 * is correctly serialised wrt the caller's subsequent test of whether to
300 * set_current_state(TASK_UNINTERRUPTIBLE);
301 * if (do_i_need_to_sleep())
304 * If the caller does not need such serialisation then use __set_current_state()
306 #define __set_current_state(state_value) \
307 do { current->state = (state_value); } while (0)
308 #define set_current_state(state_value) \
309 smp_store_mb(current->state, (state_value))
313 /* Task command name length */
314 #define TASK_COMM_LEN 16
316 #include <linux/spinlock.h>
319 * This serializes "schedule()" and also protects
320 * the run-queue from deletions/modifications (but
321 * _adding_ to the beginning of the run-queue has
324 extern rwlock_t tasklist_lock
;
325 extern spinlock_t mmlist_lock
;
329 #ifdef CONFIG_PROVE_RCU
330 extern int lockdep_tasklist_lock_is_held(void);
331 #endif /* #ifdef CONFIG_PROVE_RCU */
333 extern void sched_init(void);
334 extern void sched_init_smp(void);
335 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
336 extern void init_idle(struct task_struct
*idle
, int cpu
);
337 extern void init_idle_bootup_task(struct task_struct
*idle
);
339 extern cpumask_var_t cpu_isolated_map
;
341 extern int runqueue_is_locked(int cpu
);
343 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
344 extern void nohz_balance_enter_idle(int cpu
);
345 extern void set_cpu_sd_state_idle(void);
346 extern int get_nohz_timer_target(void);
348 static inline void nohz_balance_enter_idle(int cpu
) { }
349 static inline void set_cpu_sd_state_idle(void) { }
353 * Only dump TASK_* tasks. (0 for all tasks)
355 extern void show_state_filter(unsigned long state_filter
);
357 static inline void show_state(void)
359 show_state_filter(0);
362 extern void show_regs(struct pt_regs
*);
365 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
366 * task), SP is the stack pointer of the first frame that should be shown in the back
367 * trace (or NULL if the entire call-chain of the task should be shown).
369 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
371 extern void cpu_init (void);
372 extern void trap_init(void);
373 extern void update_process_times(int user
);
374 extern void scheduler_tick(void);
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 long io_schedule_timeout(long timeout
);
435 static inline void io_schedule(void)
437 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT
);
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 */
510 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
512 #define MMF_HAS_UPROBES 19 /* has uprobes */
513 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
515 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
517 struct sighand_struct
{
519 struct k_sigaction action
[_NSIG
];
521 wait_queue_head_t signalfd_wqh
;
524 struct pacct_struct
{
527 unsigned long ac_mem
;
528 cputime_t ac_utime
, ac_stime
;
529 unsigned long ac_minflt
, ac_majflt
;
540 * struct prev_cputime - snaphsot of system and user cputime
541 * @utime: time spent in user mode
542 * @stime: time spent in system mode
543 * @lock: protects the above two fields
545 * Stores previous user/system time values such that we can guarantee
548 struct prev_cputime
{
549 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
556 static inline void prev_cputime_init(struct prev_cputime
*prev
)
558 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
559 prev
->utime
= prev
->stime
= 0;
560 raw_spin_lock_init(&prev
->lock
);
565 * struct task_cputime - collected CPU time counts
566 * @utime: time spent in user mode, in &cputime_t units
567 * @stime: time spent in kernel mode, in &cputime_t units
568 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
570 * This structure groups together three kinds of CPU time that are tracked for
571 * threads and thread groups. Most things considering CPU time want to group
572 * these counts together and treat all three of them in parallel.
574 struct task_cputime
{
577 unsigned long long sum_exec_runtime
;
580 /* Alternate field names when used to cache expirations. */
581 #define virt_exp utime
582 #define prof_exp stime
583 #define sched_exp sum_exec_runtime
585 #define INIT_CPUTIME \
586 (struct task_cputime) { \
589 .sum_exec_runtime = 0, \
593 * This is the atomic variant of task_cputime, which can be used for
594 * storing and updating task_cputime statistics without locking.
596 struct task_cputime_atomic
{
599 atomic64_t sum_exec_runtime
;
602 #define INIT_CPUTIME_ATOMIC \
603 (struct task_cputime_atomic) { \
604 .utime = ATOMIC64_INIT(0), \
605 .stime = ATOMIC64_INIT(0), \
606 .sum_exec_runtime = ATOMIC64_INIT(0), \
609 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
612 * Disable preemption until the scheduler is running -- use an unconditional
613 * value so that it also works on !PREEMPT_COUNT kernels.
615 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
617 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
620 * Initial preempt_count value; reflects the preempt_count schedule invariant
621 * which states that during context switches:
623 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
625 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
626 * Note: See finish_task_switch().
628 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
631 * struct thread_group_cputimer - thread group interval timer counts
632 * @cputime_atomic: atomic thread group interval timers.
633 * @running: true when there are timers running and
634 * @cputime_atomic receives updates.
635 * @checking_timer: true when a thread in the group is in the
636 * process of checking for thread group timers.
638 * This structure contains the version of task_cputime, above, that is
639 * used for thread group CPU timer calculations.
641 struct thread_group_cputimer
{
642 struct task_cputime_atomic cputime_atomic
;
647 #include <linux/rwsem.h>
651 * NOTE! "signal_struct" does not have its own
652 * locking, because a shared signal_struct always
653 * implies a shared sighand_struct, so locking
654 * sighand_struct is always a proper superset of
655 * the locking of signal_struct.
657 struct signal_struct
{
661 struct list_head thread_head
;
663 wait_queue_head_t wait_chldexit
; /* for wait4() */
665 /* current thread group signal load-balancing target: */
666 struct task_struct
*curr_target
;
668 /* shared signal handling: */
669 struct sigpending shared_pending
;
671 /* thread group exit support */
674 * - notify group_exit_task when ->count is equal to notify_count
675 * - everyone except group_exit_task is stopped during signal delivery
676 * of fatal signals, group_exit_task processes the signal.
679 struct task_struct
*group_exit_task
;
681 /* thread group stop support, overloads group_exit_code too */
682 int group_stop_count
;
683 unsigned int flags
; /* see SIGNAL_* flags below */
686 * PR_SET_CHILD_SUBREAPER marks a process, like a service
687 * manager, to re-parent orphan (double-forking) child processes
688 * to this process instead of 'init'. The service manager is
689 * able to receive SIGCHLD signals and is able to investigate
690 * the process until it calls wait(). All children of this
691 * process will inherit a flag if they should look for a
692 * child_subreaper process at exit.
694 unsigned int is_child_subreaper
:1;
695 unsigned int has_child_subreaper
:1;
697 /* POSIX.1b Interval Timers */
699 struct list_head posix_timers
;
701 /* ITIMER_REAL timer for the process */
702 struct hrtimer real_timer
;
703 struct pid
*leader_pid
;
704 ktime_t it_real_incr
;
707 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
708 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
709 * values are defined to 0 and 1 respectively
711 struct cpu_itimer it
[2];
714 * Thread group totals for process CPU timers.
715 * See thread_group_cputimer(), et al, for details.
717 struct thread_group_cputimer cputimer
;
719 /* Earliest-expiration cache. */
720 struct task_cputime cputime_expires
;
722 #ifdef CONFIG_NO_HZ_FULL
723 atomic_t tick_dep_mask
;
726 struct list_head cpu_timers
[3];
728 struct pid
*tty_old_pgrp
;
730 /* boolean value for session group leader */
733 struct tty_struct
*tty
; /* NULL if no tty */
735 #ifdef CONFIG_SCHED_AUTOGROUP
736 struct autogroup
*autogroup
;
739 * Cumulative resource counters for dead threads in the group,
740 * and for reaped dead child processes forked by this group.
741 * Live threads maintain their own counters and add to these
742 * in __exit_signal, except for the group leader.
744 seqlock_t stats_lock
;
745 cputime_t utime
, stime
, cutime
, cstime
;
748 struct prev_cputime prev_cputime
;
749 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
750 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
751 unsigned long inblock
, oublock
, cinblock
, coublock
;
752 unsigned long maxrss
, cmaxrss
;
753 struct task_io_accounting ioac
;
756 * Cumulative ns of schedule CPU time fo dead threads in the
757 * group, not including a zombie group leader, (This only differs
758 * from jiffies_to_ns(utime + stime) if sched_clock uses something
759 * other than jiffies.)
761 unsigned long long sum_sched_runtime
;
764 * We don't bother to synchronize most readers of this at all,
765 * because there is no reader checking a limit that actually needs
766 * to get both rlim_cur and rlim_max atomically, and either one
767 * alone is a single word that can safely be read normally.
768 * getrlimit/setrlimit use task_lock(current->group_leader) to
769 * protect this instead of the siglock, because they really
770 * have no need to disable irqs.
772 struct rlimit rlim
[RLIM_NLIMITS
];
774 #ifdef CONFIG_BSD_PROCESS_ACCT
775 struct pacct_struct pacct
; /* per-process accounting information */
777 #ifdef CONFIG_TASKSTATS
778 struct taskstats
*stats
;
782 struct tty_audit_buf
*tty_audit_buf
;
785 oom_flags_t oom_flags
;
786 short oom_score_adj
; /* OOM kill score adjustment */
787 short oom_score_adj_min
; /* OOM kill score adjustment min value.
788 * Only settable by CAP_SYS_RESOURCE. */
790 struct mutex cred_guard_mutex
; /* guard against foreign influences on
791 * credential calculations
792 * (notably. ptrace) */
796 * Bits in flags field of signal_struct.
798 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
799 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
800 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
801 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
803 * Pending notifications to parent.
805 #define SIGNAL_CLD_STOPPED 0x00000010
806 #define SIGNAL_CLD_CONTINUED 0x00000020
807 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
809 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
811 /* If true, all threads except ->group_exit_task have pending SIGKILL */
812 static inline int signal_group_exit(const struct signal_struct
*sig
)
814 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
815 (sig
->group_exit_task
!= NULL
);
819 * Some day this will be a full-fledged user tracking system..
822 atomic_t __count
; /* reference count */
823 atomic_t processes
; /* How many processes does this user have? */
824 atomic_t sigpending
; /* How many pending signals does this user have? */
825 #ifdef CONFIG_INOTIFY_USER
826 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
827 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
829 #ifdef CONFIG_FANOTIFY
830 atomic_t fanotify_listeners
;
833 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
835 #ifdef CONFIG_POSIX_MQUEUE
836 /* protected by mq_lock */
837 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
839 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
840 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
841 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
844 struct key
*uid_keyring
; /* UID specific keyring */
845 struct key
*session_keyring
; /* UID's default session keyring */
848 /* Hash table maintenance information */
849 struct hlist_node uidhash_node
;
852 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
853 atomic_long_t locked_vm
;
857 extern int uids_sysfs_init(void);
859 extern struct user_struct
*find_user(kuid_t
);
861 extern struct user_struct root_user
;
862 #define INIT_USER (&root_user)
865 struct backing_dev_info
;
866 struct reclaim_state
;
868 #ifdef CONFIG_SCHED_INFO
870 /* cumulative counters */
871 unsigned long pcount
; /* # of times run on this cpu */
872 unsigned long long run_delay
; /* time spent waiting on a runqueue */
875 unsigned long long last_arrival
,/* when we last ran on a cpu */
876 last_queued
; /* when we were last queued to run */
878 #endif /* CONFIG_SCHED_INFO */
880 #ifdef CONFIG_TASK_DELAY_ACCT
881 struct task_delay_info
{
883 unsigned int flags
; /* Private per-task flags */
885 /* For each stat XXX, add following, aligned appropriately
887 * struct timespec XXX_start, XXX_end;
891 * Atomicity of updates to XXX_delay, XXX_count protected by
892 * single lock above (split into XXX_lock if contention is an issue).
896 * XXX_count is incremented on every XXX operation, the delay
897 * associated with the operation is added to XXX_delay.
898 * XXX_delay contains the accumulated delay time in nanoseconds.
900 u64 blkio_start
; /* Shared by blkio, swapin */
901 u64 blkio_delay
; /* wait for sync block io completion */
902 u64 swapin_delay
; /* wait for swapin block io completion */
903 u32 blkio_count
; /* total count of the number of sync block */
904 /* io operations performed */
905 u32 swapin_count
; /* total count of the number of swapin block */
906 /* io operations performed */
909 u64 freepages_delay
; /* wait for memory reclaim */
910 u32 freepages_count
; /* total count of memory reclaim */
912 #endif /* CONFIG_TASK_DELAY_ACCT */
914 static inline int sched_info_on(void)
916 #ifdef CONFIG_SCHEDSTATS
918 #elif defined(CONFIG_TASK_DELAY_ACCT)
919 extern int delayacct_on
;
926 #ifdef CONFIG_SCHEDSTATS
927 void force_schedstat_enabled(void);
938 * Increase resolution of cpu_capacity calculations
940 #define SCHED_CAPACITY_SHIFT 10
941 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
944 * Wake-queues are lists of tasks with a pending wakeup, whose
945 * callers have already marked the task as woken internally,
946 * and can thus carry on. A common use case is being able to
947 * do the wakeups once the corresponding user lock as been
950 * We hold reference to each task in the list across the wakeup,
951 * thus guaranteeing that the memory is still valid by the time
952 * the actual wakeups are performed in wake_up_q().
954 * One per task suffices, because there's never a need for a task to be
955 * in two wake queues simultaneously; it is forbidden to abandon a task
956 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
957 * already in a wake queue, the wakeup will happen soon and the second
958 * waker can just skip it.
960 * The WAKE_Q macro declares and initializes the list head.
961 * wake_up_q() does NOT reinitialize the list; it's expected to be
962 * called near the end of a function, where the fact that the queue is
963 * not used again will be easy to see by inspection.
965 * Note that this can cause spurious wakeups. schedule() callers
966 * must ensure the call is done inside a loop, confirming that the
967 * wakeup condition has in fact occurred.
970 struct wake_q_node
*next
;
974 struct wake_q_node
*first
;
975 struct wake_q_node
**lastp
;
978 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
980 #define WAKE_Q(name) \
981 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
983 extern void wake_q_add(struct wake_q_head
*head
,
984 struct task_struct
*task
);
985 extern void wake_up_q(struct wake_q_head
*head
);
988 * sched-domains (multiprocessor balancing) declarations:
991 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
992 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
993 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
994 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
995 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
996 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
997 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
998 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
999 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1000 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1001 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1002 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1003 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1004 #define SD_NUMA 0x4000 /* cross-node balancing */
1006 #ifdef CONFIG_SCHED_SMT
1007 static inline int cpu_smt_flags(void)
1009 return SD_SHARE_CPUCAPACITY
| SD_SHARE_PKG_RESOURCES
;
1013 #ifdef CONFIG_SCHED_MC
1014 static inline int cpu_core_flags(void)
1016 return SD_SHARE_PKG_RESOURCES
;
1021 static inline int cpu_numa_flags(void)
1027 struct sched_domain_attr
{
1028 int relax_domain_level
;
1031 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1032 .relax_domain_level = -1, \
1035 extern int sched_domain_level_max
;
1039 struct sched_domain
{
1040 /* These fields must be setup */
1041 struct sched_domain
*parent
; /* top domain must be null terminated */
1042 struct sched_domain
*child
; /* bottom domain must be null terminated */
1043 struct sched_group
*groups
; /* the balancing groups of the domain */
1044 unsigned long min_interval
; /* Minimum balance interval ms */
1045 unsigned long max_interval
; /* Maximum balance interval ms */
1046 unsigned int busy_factor
; /* less balancing by factor if busy */
1047 unsigned int imbalance_pct
; /* No balance until over watermark */
1048 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
1049 unsigned int busy_idx
;
1050 unsigned int idle_idx
;
1051 unsigned int newidle_idx
;
1052 unsigned int wake_idx
;
1053 unsigned int forkexec_idx
;
1054 unsigned int smt_gain
;
1056 int nohz_idle
; /* NOHZ IDLE status */
1057 int flags
; /* See SD_* */
1060 /* Runtime fields. */
1061 unsigned long last_balance
; /* init to jiffies. units in jiffies */
1062 unsigned int balance_interval
; /* initialise to 1. units in ms. */
1063 unsigned int nr_balance_failed
; /* initialise to 0 */
1065 /* idle_balance() stats */
1066 u64 max_newidle_lb_cost
;
1067 unsigned long next_decay_max_lb_cost
;
1069 #ifdef CONFIG_SCHEDSTATS
1070 /* load_balance() stats */
1071 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
1072 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
1073 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
1074 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
1075 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
1076 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
1077 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
1078 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
1080 /* Active load balancing */
1081 unsigned int alb_count
;
1082 unsigned int alb_failed
;
1083 unsigned int alb_pushed
;
1085 /* SD_BALANCE_EXEC stats */
1086 unsigned int sbe_count
;
1087 unsigned int sbe_balanced
;
1088 unsigned int sbe_pushed
;
1090 /* SD_BALANCE_FORK stats */
1091 unsigned int sbf_count
;
1092 unsigned int sbf_balanced
;
1093 unsigned int sbf_pushed
;
1095 /* try_to_wake_up() stats */
1096 unsigned int ttwu_wake_remote
;
1097 unsigned int ttwu_move_affine
;
1098 unsigned int ttwu_move_balance
;
1100 #ifdef CONFIG_SCHED_DEBUG
1104 void *private; /* used during construction */
1105 struct rcu_head rcu
; /* used during destruction */
1108 unsigned int span_weight
;
1110 * Span of all CPUs in this domain.
1112 * NOTE: this field is variable length. (Allocated dynamically
1113 * by attaching extra space to the end of the structure,
1114 * depending on how many CPUs the kernel has booted up with)
1116 unsigned long span
[0];
1119 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1121 return to_cpumask(sd
->span
);
1124 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1125 struct sched_domain_attr
*dattr_new
);
1127 /* Allocate an array of sched domains, for partition_sched_domains(). */
1128 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1129 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1131 bool cpus_share_cache(int this_cpu
, int that_cpu
);
1133 typedef const struct cpumask
*(*sched_domain_mask_f
)(int cpu
);
1134 typedef int (*sched_domain_flags_f
)(void);
1136 #define SDTL_OVERLAP 0x01
1139 struct sched_domain
**__percpu sd
;
1140 struct sched_group
**__percpu sg
;
1141 struct sched_group_capacity
**__percpu sgc
;
1144 struct sched_domain_topology_level
{
1145 sched_domain_mask_f mask
;
1146 sched_domain_flags_f sd_flags
;
1149 struct sd_data data
;
1150 #ifdef CONFIG_SCHED_DEBUG
1155 extern void set_sched_topology(struct sched_domain_topology_level
*tl
);
1156 extern void wake_up_if_idle(int cpu
);
1158 #ifdef CONFIG_SCHED_DEBUG
1159 # define SD_INIT_NAME(type) .name = #type
1161 # define SD_INIT_NAME(type)
1164 #else /* CONFIG_SMP */
1166 struct sched_domain_attr
;
1169 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1170 struct sched_domain_attr
*dattr_new
)
1174 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1179 #endif /* !CONFIG_SMP */
1182 struct io_context
; /* See blkdev.h */
1185 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1186 extern void prefetch_stack(struct task_struct
*t
);
1188 static inline void prefetch_stack(struct task_struct
*t
) { }
1191 struct audit_context
; /* See audit.c */
1193 struct pipe_inode_info
;
1194 struct uts_namespace
;
1196 struct load_weight
{
1197 unsigned long weight
;
1202 * The load_avg/util_avg accumulates an infinite geometric series.
1203 * 1) load_avg factors frequency scaling into the amount of time that a
1204 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1205 * aggregated such weights of all runnable and blocked sched_entities.
1206 * 2) util_avg factors frequency and cpu scaling into the amount of time
1207 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1208 * For cfs_rq, it is the aggregated such times of all runnable and
1209 * blocked sched_entities.
1210 * The 64 bit load_sum can:
1211 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1212 * the highest weight (=88761) always runnable, we should not overflow
1213 * 2) for entity, support any load.weight always runnable
1216 u64 last_update_time
, load_sum
;
1217 u32 util_sum
, period_contrib
;
1218 unsigned long load_avg
, util_avg
;
1221 #ifdef CONFIG_SCHEDSTATS
1222 struct sched_statistics
{
1232 s64 sum_sleep_runtime
;
1239 u64 nr_migrations_cold
;
1240 u64 nr_failed_migrations_affine
;
1241 u64 nr_failed_migrations_running
;
1242 u64 nr_failed_migrations_hot
;
1243 u64 nr_forced_migrations
;
1246 u64 nr_wakeups_sync
;
1247 u64 nr_wakeups_migrate
;
1248 u64 nr_wakeups_local
;
1249 u64 nr_wakeups_remote
;
1250 u64 nr_wakeups_affine
;
1251 u64 nr_wakeups_affine_attempts
;
1252 u64 nr_wakeups_passive
;
1253 u64 nr_wakeups_idle
;
1257 struct sched_entity
{
1258 struct load_weight load
; /* for load-balancing */
1259 struct rb_node run_node
;
1260 struct list_head group_node
;
1264 u64 sum_exec_runtime
;
1266 u64 prev_sum_exec_runtime
;
1270 #ifdef CONFIG_SCHEDSTATS
1271 struct sched_statistics statistics
;
1274 #ifdef CONFIG_FAIR_GROUP_SCHED
1276 struct sched_entity
*parent
;
1277 /* rq on which this entity is (to be) queued: */
1278 struct cfs_rq
*cfs_rq
;
1279 /* rq "owned" by this entity/group: */
1280 struct cfs_rq
*my_q
;
1285 * Per entity load average tracking.
1287 * Put into separate cache line so it does not
1288 * collide with read-mostly values above.
1290 struct sched_avg avg ____cacheline_aligned_in_smp
;
1294 struct sched_rt_entity
{
1295 struct list_head run_list
;
1296 unsigned long timeout
;
1297 unsigned long watchdog_stamp
;
1298 unsigned int time_slice
;
1299 unsigned short on_rq
;
1300 unsigned short on_list
;
1302 struct sched_rt_entity
*back
;
1303 #ifdef CONFIG_RT_GROUP_SCHED
1304 struct sched_rt_entity
*parent
;
1305 /* rq on which this entity is (to be) queued: */
1306 struct rt_rq
*rt_rq
;
1307 /* rq "owned" by this entity/group: */
1312 struct sched_dl_entity
{
1313 struct rb_node rb_node
;
1316 * Original scheduling parameters. Copied here from sched_attr
1317 * during sched_setattr(), they will remain the same until
1318 * the next sched_setattr().
1320 u64 dl_runtime
; /* maximum runtime for each instance */
1321 u64 dl_deadline
; /* relative deadline of each instance */
1322 u64 dl_period
; /* separation of two instances (period) */
1323 u64 dl_bw
; /* dl_runtime / dl_deadline */
1326 * Actual scheduling parameters. Initialized with the values above,
1327 * they are continously updated during task execution. Note that
1328 * the remaining runtime could be < 0 in case we are in overrun.
1330 s64 runtime
; /* remaining runtime for this instance */
1331 u64 deadline
; /* absolute deadline for this instance */
1332 unsigned int flags
; /* specifying the scheduler behaviour */
1337 * @dl_throttled tells if we exhausted the runtime. If so, the
1338 * task has to wait for a replenishment to be performed at the
1339 * next firing of dl_timer.
1341 * @dl_boosted tells if we are boosted due to DI. If so we are
1342 * outside bandwidth enforcement mechanism (but only until we
1343 * exit the critical section);
1345 * @dl_yielded tells if task gave up the cpu before consuming
1346 * all its available runtime during the last job.
1348 int dl_throttled
, dl_boosted
, dl_yielded
;
1351 * Bandwidth enforcement timer. Each -deadline task has its
1352 * own bandwidth to be enforced, thus we need one timer per task.
1354 struct hrtimer dl_timer
;
1362 u8 pad
; /* Otherwise the compiler can store garbage here. */
1364 u32 s
; /* Set of bits. */
1368 enum perf_event_task_context
{
1369 perf_invalid_context
= -1,
1370 perf_hw_context
= 0,
1372 perf_nr_task_contexts
,
1375 /* Track pages that require TLB flushes */
1376 struct tlbflush_unmap_batch
{
1378 * Each bit set is a CPU that potentially has a TLB entry for one of
1379 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1381 struct cpumask cpumask
;
1383 /* True if any bit in cpumask is set */
1384 bool flush_required
;
1387 * If true then the PTE was dirty when unmapped. The entry must be
1388 * flushed before IO is initiated or a stale TLB entry potentially
1389 * allows an update without redirtying the page.
1394 struct task_struct
{
1395 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1398 unsigned int flags
; /* per process flags, defined below */
1399 unsigned int ptrace
;
1402 struct llist_node wake_entry
;
1404 unsigned int wakee_flips
;
1405 unsigned long wakee_flip_decay_ts
;
1406 struct task_struct
*last_wakee
;
1412 int prio
, static_prio
, normal_prio
;
1413 unsigned int rt_priority
;
1414 const struct sched_class
*sched_class
;
1415 struct sched_entity se
;
1416 struct sched_rt_entity rt
;
1417 #ifdef CONFIG_CGROUP_SCHED
1418 struct task_group
*sched_task_group
;
1420 struct sched_dl_entity dl
;
1422 #ifdef CONFIG_PREEMPT_NOTIFIERS
1423 /* list of struct preempt_notifier: */
1424 struct hlist_head preempt_notifiers
;
1427 #ifdef CONFIG_BLK_DEV_IO_TRACE
1428 unsigned int btrace_seq
;
1431 unsigned int policy
;
1432 int nr_cpus_allowed
;
1433 cpumask_t cpus_allowed
;
1435 #ifdef CONFIG_PREEMPT_RCU
1436 int rcu_read_lock_nesting
;
1437 union rcu_special rcu_read_unlock_special
;
1438 struct list_head rcu_node_entry
;
1439 struct rcu_node
*rcu_blocked_node
;
1440 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1441 #ifdef CONFIG_TASKS_RCU
1442 unsigned long rcu_tasks_nvcsw
;
1443 bool rcu_tasks_holdout
;
1444 struct list_head rcu_tasks_holdout_list
;
1445 int rcu_tasks_idle_cpu
;
1446 #endif /* #ifdef CONFIG_TASKS_RCU */
1448 #ifdef CONFIG_SCHED_INFO
1449 struct sched_info sched_info
;
1452 struct list_head tasks
;
1454 struct plist_node pushable_tasks
;
1455 struct rb_node pushable_dl_tasks
;
1458 struct mm_struct
*mm
, *active_mm
;
1459 /* per-thread vma caching */
1460 u32 vmacache_seqnum
;
1461 struct vm_area_struct
*vmacache
[VMACACHE_SIZE
];
1462 #if defined(SPLIT_RSS_COUNTING)
1463 struct task_rss_stat rss_stat
;
1467 int exit_code
, exit_signal
;
1468 int pdeath_signal
; /* The signal sent when the parent dies */
1469 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1471 /* Used for emulating ABI behavior of previous Linux versions */
1472 unsigned int personality
;
1474 /* scheduler bits, serialized by scheduler locks */
1475 unsigned sched_reset_on_fork
:1;
1476 unsigned sched_contributes_to_load
:1;
1477 unsigned sched_migrated
:1;
1478 unsigned :0; /* force alignment to the next boundary */
1480 /* unserialized, strictly 'current' */
1481 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1482 unsigned in_iowait
:1;
1484 unsigned memcg_may_oom
:1;
1486 unsigned memcg_kmem_skip_account
:1;
1489 #ifdef CONFIG_COMPAT_BRK
1490 unsigned brk_randomized
:1;
1493 unsigned long atomic_flags
; /* Flags needing atomic access. */
1495 struct restart_block restart_block
;
1500 #ifdef CONFIG_CC_STACKPROTECTOR
1501 /* Canary value for the -fstack-protector gcc feature */
1502 unsigned long stack_canary
;
1505 * pointers to (original) parent process, youngest child, younger sibling,
1506 * older sibling, respectively. (p->father can be replaced with
1507 * p->real_parent->pid)
1509 struct task_struct __rcu
*real_parent
; /* real parent process */
1510 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1512 * children/sibling forms the list of my natural children
1514 struct list_head children
; /* list of my children */
1515 struct list_head sibling
; /* linkage in my parent's children list */
1516 struct task_struct
*group_leader
; /* threadgroup leader */
1519 * ptraced is the list of tasks this task is using ptrace on.
1520 * This includes both natural children and PTRACE_ATTACH targets.
1521 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1523 struct list_head ptraced
;
1524 struct list_head ptrace_entry
;
1526 /* PID/PID hash table linkage. */
1527 struct pid_link pids
[PIDTYPE_MAX
];
1528 struct list_head thread_group
;
1529 struct list_head thread_node
;
1531 struct completion
*vfork_done
; /* for vfork() */
1532 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1533 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1535 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1537 struct prev_cputime prev_cputime
;
1538 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1539 seqcount_t vtime_seqcount
;
1540 unsigned long long vtime_snap
;
1542 /* Task is sleeping or running in a CPU with VTIME inactive */
1544 /* Task runs in userspace in a CPU with VTIME active */
1546 /* Task runs in kernelspace in a CPU with VTIME active */
1548 } vtime_snap_whence
;
1551 #ifdef CONFIG_NO_HZ_FULL
1552 atomic_t tick_dep_mask
;
1554 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1555 u64 start_time
; /* monotonic time in nsec */
1556 u64 real_start_time
; /* boot based time in nsec */
1557 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1558 unsigned long min_flt
, maj_flt
;
1560 struct task_cputime cputime_expires
;
1561 struct list_head cpu_timers
[3];
1563 /* process credentials */
1564 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1565 * credentials (COW) */
1566 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1567 * credentials (COW) */
1568 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1569 - access with [gs]et_task_comm (which lock
1570 it with task_lock())
1571 - initialized normally by setup_new_exec */
1572 /* file system info */
1573 struct nameidata
*nameidata
;
1574 #ifdef CONFIG_SYSVIPC
1576 struct sysv_sem sysvsem
;
1577 struct sysv_shm sysvshm
;
1579 #ifdef CONFIG_DETECT_HUNG_TASK
1580 /* hung task detection */
1581 unsigned long last_switch_count
;
1583 /* filesystem information */
1584 struct fs_struct
*fs
;
1585 /* open file information */
1586 struct files_struct
*files
;
1588 struct nsproxy
*nsproxy
;
1589 /* signal handlers */
1590 struct signal_struct
*signal
;
1591 struct sighand_struct
*sighand
;
1593 sigset_t blocked
, real_blocked
;
1594 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1595 struct sigpending pending
;
1597 unsigned long sas_ss_sp
;
1600 struct callback_head
*task_works
;
1602 struct audit_context
*audit_context
;
1603 #ifdef CONFIG_AUDITSYSCALL
1605 unsigned int sessionid
;
1607 struct seccomp seccomp
;
1609 /* Thread group tracking */
1612 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1614 spinlock_t alloc_lock
;
1616 /* Protection of the PI data structures: */
1617 raw_spinlock_t pi_lock
;
1619 struct wake_q_node wake_q
;
1621 #ifdef CONFIG_RT_MUTEXES
1622 /* PI waiters blocked on a rt_mutex held by this task */
1623 struct rb_root pi_waiters
;
1624 struct rb_node
*pi_waiters_leftmost
;
1625 /* Deadlock detection and priority inheritance handling */
1626 struct rt_mutex_waiter
*pi_blocked_on
;
1629 #ifdef CONFIG_DEBUG_MUTEXES
1630 /* mutex deadlock detection */
1631 struct mutex_waiter
*blocked_on
;
1633 #ifdef CONFIG_TRACE_IRQFLAGS
1634 unsigned int irq_events
;
1635 unsigned long hardirq_enable_ip
;
1636 unsigned long hardirq_disable_ip
;
1637 unsigned int hardirq_enable_event
;
1638 unsigned int hardirq_disable_event
;
1639 int hardirqs_enabled
;
1640 int hardirq_context
;
1641 unsigned long softirq_disable_ip
;
1642 unsigned long softirq_enable_ip
;
1643 unsigned int softirq_disable_event
;
1644 unsigned int softirq_enable_event
;
1645 int softirqs_enabled
;
1646 int softirq_context
;
1648 #ifdef CONFIG_LOCKDEP
1649 # define MAX_LOCK_DEPTH 48UL
1652 unsigned int lockdep_recursion
;
1653 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1654 gfp_t lockdep_reclaim_gfp
;
1657 unsigned int in_ubsan
;
1660 /* journalling filesystem info */
1663 /* stacked block device info */
1664 struct bio_list
*bio_list
;
1667 /* stack plugging */
1668 struct blk_plug
*plug
;
1672 struct reclaim_state
*reclaim_state
;
1674 struct backing_dev_info
*backing_dev_info
;
1676 struct io_context
*io_context
;
1678 unsigned long ptrace_message
;
1679 siginfo_t
*last_siginfo
; /* For ptrace use. */
1680 struct task_io_accounting ioac
;
1681 #if defined(CONFIG_TASK_XACCT)
1682 u64 acct_rss_mem1
; /* accumulated rss usage */
1683 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1684 cputime_t acct_timexpd
; /* stime + utime since last update */
1686 #ifdef CONFIG_CPUSETS
1687 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1688 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1689 int cpuset_mem_spread_rotor
;
1690 int cpuset_slab_spread_rotor
;
1692 #ifdef CONFIG_CGROUPS
1693 /* Control Group info protected by css_set_lock */
1694 struct css_set __rcu
*cgroups
;
1695 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1696 struct list_head cg_list
;
1699 struct robust_list_head __user
*robust_list
;
1700 #ifdef CONFIG_COMPAT
1701 struct compat_robust_list_head __user
*compat_robust_list
;
1703 struct list_head pi_state_list
;
1704 struct futex_pi_state
*pi_state_cache
;
1706 #ifdef CONFIG_PERF_EVENTS
1707 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1708 struct mutex perf_event_mutex
;
1709 struct list_head perf_event_list
;
1711 #ifdef CONFIG_DEBUG_PREEMPT
1712 unsigned long preempt_disable_ip
;
1715 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1717 short pref_node_fork
;
1719 #ifdef CONFIG_NUMA_BALANCING
1721 unsigned int numa_scan_period
;
1722 unsigned int numa_scan_period_max
;
1723 int numa_preferred_nid
;
1724 unsigned long numa_migrate_retry
;
1725 u64 node_stamp
; /* migration stamp */
1726 u64 last_task_numa_placement
;
1727 u64 last_sum_exec_runtime
;
1728 struct callback_head numa_work
;
1730 struct list_head numa_entry
;
1731 struct numa_group
*numa_group
;
1734 * numa_faults is an array split into four regions:
1735 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1736 * in this precise order.
1738 * faults_memory: Exponential decaying average of faults on a per-node
1739 * basis. Scheduling placement decisions are made based on these
1740 * counts. The values remain static for the duration of a PTE scan.
1741 * faults_cpu: Track the nodes the process was running on when a NUMA
1742 * hinting fault was incurred.
1743 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1744 * during the current scan window. When the scan completes, the counts
1745 * in faults_memory and faults_cpu decay and these values are copied.
1747 unsigned long *numa_faults
;
1748 unsigned long total_numa_faults
;
1751 * numa_faults_locality tracks if faults recorded during the last
1752 * scan window were remote/local or failed to migrate. The task scan
1753 * period is adapted based on the locality of the faults with different
1754 * weights depending on whether they were shared or private faults
1756 unsigned long numa_faults_locality
[3];
1758 unsigned long numa_pages_migrated
;
1759 #endif /* CONFIG_NUMA_BALANCING */
1761 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1762 struct tlbflush_unmap_batch tlb_ubc
;
1765 struct rcu_head rcu
;
1768 * cache last used pipe for splice
1770 struct pipe_inode_info
*splice_pipe
;
1772 struct page_frag task_frag
;
1774 #ifdef CONFIG_TASK_DELAY_ACCT
1775 struct task_delay_info
*delays
;
1777 #ifdef CONFIG_FAULT_INJECTION
1781 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1782 * balance_dirty_pages() for some dirty throttling pause
1785 int nr_dirtied_pause
;
1786 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1788 #ifdef CONFIG_LATENCYTOP
1789 int latency_record_count
;
1790 struct latency_record latency_record
[LT_SAVECOUNT
];
1793 * time slack values; these are used to round up poll() and
1794 * select() etc timeout values. These are in nanoseconds.
1797 u64 default_timer_slack_ns
;
1800 unsigned int kasan_depth
;
1802 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1803 /* Index of current stored address in ret_stack */
1805 /* Stack of return addresses for return function tracing */
1806 struct ftrace_ret_stack
*ret_stack
;
1807 /* time stamp for last schedule */
1808 unsigned long long ftrace_timestamp
;
1810 * Number of functions that haven't been traced
1811 * because of depth overrun.
1813 atomic_t trace_overrun
;
1814 /* Pause for the tracing */
1815 atomic_t tracing_graph_pause
;
1817 #ifdef CONFIG_TRACING
1818 /* state flags for use by tracers */
1819 unsigned long trace
;
1820 /* bitmask and counter of trace recursion */
1821 unsigned long trace_recursion
;
1822 #endif /* CONFIG_TRACING */
1824 /* Coverage collection mode enabled for this task (0 if disabled). */
1825 enum kcov_mode kcov_mode
;
1826 /* Size of the kcov_area. */
1828 /* Buffer for coverage collection. */
1830 /* kcov desciptor wired with this task or NULL. */
1834 struct mem_cgroup
*memcg_in_oom
;
1835 gfp_t memcg_oom_gfp_mask
;
1836 int memcg_oom_order
;
1838 /* number of pages to reclaim on returning to userland */
1839 unsigned int memcg_nr_pages_over_high
;
1841 #ifdef CONFIG_UPROBES
1842 struct uprobe_task
*utask
;
1844 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1845 unsigned int sequential_io
;
1846 unsigned int sequential_io_avg
;
1848 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1849 unsigned long task_state_change
;
1851 int pagefault_disabled
;
1853 struct task_struct
*oom_reaper_list
;
1855 /* CPU-specific state of this task */
1856 struct thread_struct thread
;
1858 * WARNING: on x86, 'thread_struct' contains a variable-sized
1859 * structure. It *MUST* be at the end of 'task_struct'.
1861 * Do not put anything below here!
1865 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1866 extern int arch_task_struct_size __read_mostly
;
1868 # define arch_task_struct_size (sizeof(struct task_struct))
1871 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1872 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1874 #define TNF_MIGRATED 0x01
1875 #define TNF_NO_GROUP 0x02
1876 #define TNF_SHARED 0x04
1877 #define TNF_FAULT_LOCAL 0x08
1878 #define TNF_MIGRATE_FAIL 0x10
1880 #ifdef CONFIG_NUMA_BALANCING
1881 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
1882 extern pid_t
task_numa_group_id(struct task_struct
*p
);
1883 extern void set_numabalancing_state(bool enabled
);
1884 extern void task_numa_free(struct task_struct
*p
);
1885 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
1886 int src_nid
, int dst_cpu
);
1888 static inline void task_numa_fault(int last_node
, int node
, int pages
,
1892 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
1896 static inline void set_numabalancing_state(bool enabled
)
1899 static inline void task_numa_free(struct task_struct
*p
)
1902 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
1903 struct page
*page
, int src_nid
, int dst_cpu
)
1909 static inline struct pid
*task_pid(struct task_struct
*task
)
1911 return task
->pids
[PIDTYPE_PID
].pid
;
1914 static inline struct pid
*task_tgid(struct task_struct
*task
)
1916 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1920 * Without tasklist or rcu lock it is not safe to dereference
1921 * the result of task_pgrp/task_session even if task == current,
1922 * we can race with another thread doing sys_setsid/sys_setpgid.
1924 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1926 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1929 static inline struct pid
*task_session(struct task_struct
*task
)
1931 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1934 struct pid_namespace
;
1937 * the helpers to get the task's different pids as they are seen
1938 * from various namespaces
1940 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1941 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1943 * task_xid_nr_ns() : id seen from the ns specified;
1945 * set_task_vxid() : assigns a virtual id to a task;
1947 * see also pid_nr() etc in include/linux/pid.h
1949 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1950 struct pid_namespace
*ns
);
1952 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1957 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1958 struct pid_namespace
*ns
)
1960 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1963 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1965 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1969 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1974 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1976 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1978 return pid_vnr(task_tgid(tsk
));
1982 static inline int pid_alive(const struct task_struct
*p
);
1983 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1989 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1995 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1997 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
2000 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
2001 struct pid_namespace
*ns
)
2003 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
2006 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
2008 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
2012 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
2013 struct pid_namespace
*ns
)
2015 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
2018 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
2020 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
2023 /* obsolete, do not use */
2024 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
2026 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
2030 * pid_alive - check that a task structure is not stale
2031 * @p: Task structure to be checked.
2033 * Test if a process is not yet dead (at most zombie state)
2034 * If pid_alive fails, then pointers within the task structure
2035 * can be stale and must not be dereferenced.
2037 * Return: 1 if the process is alive. 0 otherwise.
2039 static inline int pid_alive(const struct task_struct
*p
)
2041 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
2045 * is_global_init - check if a task structure is init. Since init
2046 * is free to have sub-threads we need to check tgid.
2047 * @tsk: Task structure to be checked.
2049 * Check if a task structure is the first user space task the kernel created.
2051 * Return: 1 if the task structure is init. 0 otherwise.
2053 static inline int is_global_init(struct task_struct
*tsk
)
2055 return task_tgid_nr(tsk
) == 1;
2058 extern struct pid
*cad_pid
;
2060 extern void free_task(struct task_struct
*tsk
);
2061 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2063 extern void __put_task_struct(struct task_struct
*t
);
2065 static inline void put_task_struct(struct task_struct
*t
)
2067 if (atomic_dec_and_test(&t
->usage
))
2068 __put_task_struct(t
);
2071 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2072 extern void task_cputime(struct task_struct
*t
,
2073 cputime_t
*utime
, cputime_t
*stime
);
2074 extern void task_cputime_scaled(struct task_struct
*t
,
2075 cputime_t
*utimescaled
, cputime_t
*stimescaled
);
2076 extern cputime_t
task_gtime(struct task_struct
*t
);
2078 static inline void task_cputime(struct task_struct
*t
,
2079 cputime_t
*utime
, cputime_t
*stime
)
2087 static inline void task_cputime_scaled(struct task_struct
*t
,
2088 cputime_t
*utimescaled
,
2089 cputime_t
*stimescaled
)
2092 *utimescaled
= t
->utimescaled
;
2094 *stimescaled
= t
->stimescaled
;
2097 static inline cputime_t
task_gtime(struct task_struct
*t
)
2102 extern void task_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2103 extern void thread_group_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2108 #define PF_EXITING 0x00000004 /* getting shut down */
2109 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2110 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2111 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2112 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2113 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2114 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2115 #define PF_DUMPCORE 0x00000200 /* dumped core */
2116 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2117 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2118 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2119 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2120 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2121 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2122 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2123 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2124 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2125 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2126 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2127 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2128 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2129 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2130 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2131 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2132 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2133 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2134 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2137 * Only the _current_ task can read/write to tsk->flags, but other
2138 * tasks can access tsk->flags in readonly mode for example
2139 * with tsk_used_math (like during threaded core dumping).
2140 * There is however an exception to this rule during ptrace
2141 * or during fork: the ptracer task is allowed to write to the
2142 * child->flags of its traced child (same goes for fork, the parent
2143 * can write to the child->flags), because we're guaranteed the
2144 * child is not running and in turn not changing child->flags
2145 * at the same time the parent does it.
2147 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2148 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2149 #define clear_used_math() clear_stopped_child_used_math(current)
2150 #define set_used_math() set_stopped_child_used_math(current)
2151 #define conditional_stopped_child_used_math(condition, child) \
2152 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2153 #define conditional_used_math(condition) \
2154 conditional_stopped_child_used_math(condition, current)
2155 #define copy_to_stopped_child_used_math(child) \
2156 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2157 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2158 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2159 #define used_math() tsk_used_math(current)
2161 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2162 * __GFP_FS is also cleared as it implies __GFP_IO.
2164 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
2166 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
2167 flags
&= ~(__GFP_IO
| __GFP_FS
);
2171 static inline unsigned int memalloc_noio_save(void)
2173 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
2174 current
->flags
|= PF_MEMALLOC_NOIO
;
2178 static inline void memalloc_noio_restore(unsigned int flags
)
2180 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
2183 /* Per-process atomic flags. */
2184 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2185 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2186 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2189 #define TASK_PFA_TEST(name, func) \
2190 static inline bool task_##func(struct task_struct *p) \
2191 { return test_bit(PFA_##name, &p->atomic_flags); }
2192 #define TASK_PFA_SET(name, func) \
2193 static inline void task_set_##func(struct task_struct *p) \
2194 { set_bit(PFA_##name, &p->atomic_flags); }
2195 #define TASK_PFA_CLEAR(name, func) \
2196 static inline void task_clear_##func(struct task_struct *p) \
2197 { clear_bit(PFA_##name, &p->atomic_flags); }
2199 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
2200 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
2202 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
2203 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
2204 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2206 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2207 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2208 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2211 * task->jobctl flags
2213 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2215 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2216 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2217 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2218 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2219 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2220 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2221 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2223 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2224 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2225 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2226 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2227 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2228 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2229 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2231 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2232 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2234 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2235 unsigned long mask
);
2236 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2237 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2238 unsigned long mask
);
2240 static inline void rcu_copy_process(struct task_struct
*p
)
2242 #ifdef CONFIG_PREEMPT_RCU
2243 p
->rcu_read_lock_nesting
= 0;
2244 p
->rcu_read_unlock_special
.s
= 0;
2245 p
->rcu_blocked_node
= NULL
;
2246 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2247 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2248 #ifdef CONFIG_TASKS_RCU
2249 p
->rcu_tasks_holdout
= false;
2250 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2251 p
->rcu_tasks_idle_cpu
= -1;
2252 #endif /* #ifdef CONFIG_TASKS_RCU */
2255 static inline void tsk_restore_flags(struct task_struct
*task
,
2256 unsigned long orig_flags
, unsigned long flags
)
2258 task
->flags
&= ~flags
;
2259 task
->flags
|= orig_flags
& flags
;
2262 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2263 const struct cpumask
*trial
);
2264 extern int task_can_attach(struct task_struct
*p
,
2265 const struct cpumask
*cs_cpus_allowed
);
2267 extern void do_set_cpus_allowed(struct task_struct
*p
,
2268 const struct cpumask
*new_mask
);
2270 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2271 const struct cpumask
*new_mask
);
2273 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2274 const struct cpumask
*new_mask
)
2277 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2278 const struct cpumask
*new_mask
)
2280 if (!cpumask_test_cpu(0, new_mask
))
2286 #ifdef CONFIG_NO_HZ_COMMON
2287 void calc_load_enter_idle(void);
2288 void calc_load_exit_idle(void);
2290 static inline void calc_load_enter_idle(void) { }
2291 static inline void calc_load_exit_idle(void) { }
2292 #endif /* CONFIG_NO_HZ_COMMON */
2295 * Do not use outside of architecture code which knows its limitations.
2297 * sched_clock() has no promise of monotonicity or bounded drift between
2298 * CPUs, use (which you should not) requires disabling IRQs.
2300 * Please use one of the three interfaces below.
2302 extern unsigned long long notrace
sched_clock(void);
2304 * See the comment in kernel/sched/clock.c
2306 extern u64
cpu_clock(int cpu
);
2307 extern u64
local_clock(void);
2308 extern u64
running_clock(void);
2309 extern u64
sched_clock_cpu(int cpu
);
2312 extern void sched_clock_init(void);
2314 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2315 static inline void sched_clock_tick(void)
2319 static inline void sched_clock_idle_sleep_event(void)
2323 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
2328 * Architectures can set this to 1 if they have specified
2329 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2330 * but then during bootup it turns out that sched_clock()
2331 * is reliable after all:
2333 extern int sched_clock_stable(void);
2334 extern void set_sched_clock_stable(void);
2335 extern void clear_sched_clock_stable(void);
2337 extern void sched_clock_tick(void);
2338 extern void sched_clock_idle_sleep_event(void);
2339 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2342 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2344 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2345 * The reason for this explicit opt-in is not to have perf penalty with
2346 * slow sched_clocks.
2348 extern void enable_sched_clock_irqtime(void);
2349 extern void disable_sched_clock_irqtime(void);
2351 static inline void enable_sched_clock_irqtime(void) {}
2352 static inline void disable_sched_clock_irqtime(void) {}
2355 extern unsigned long long
2356 task_sched_runtime(struct task_struct
*task
);
2358 /* sched_exec is called by processes performing an exec */
2360 extern void sched_exec(void);
2362 #define sched_exec() {}
2365 extern void sched_clock_idle_sleep_event(void);
2366 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2368 #ifdef CONFIG_HOTPLUG_CPU
2369 extern void idle_task_exit(void);
2371 static inline void idle_task_exit(void) {}
2374 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2375 extern void wake_up_nohz_cpu(int cpu
);
2377 static inline void wake_up_nohz_cpu(int cpu
) { }
2380 #ifdef CONFIG_NO_HZ_FULL
2381 extern u64
scheduler_tick_max_deferment(void);
2384 #ifdef CONFIG_SCHED_AUTOGROUP
2385 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2386 extern void sched_autogroup_detach(struct task_struct
*p
);
2387 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2388 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2389 #ifdef CONFIG_PROC_FS
2390 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2391 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2394 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2395 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2396 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2397 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2400 extern int yield_to(struct task_struct
*p
, bool preempt
);
2401 extern void set_user_nice(struct task_struct
*p
, long nice
);
2402 extern int task_prio(const struct task_struct
*p
);
2404 * task_nice - return the nice value of a given task.
2405 * @p: the task in question.
2407 * Return: The nice value [ -20 ... 0 ... 19 ].
2409 static inline int task_nice(const struct task_struct
*p
)
2411 return PRIO_TO_NICE((p
)->static_prio
);
2413 extern int can_nice(const struct task_struct
*p
, const int nice
);
2414 extern int task_curr(const struct task_struct
*p
);
2415 extern int idle_cpu(int cpu
);
2416 extern int sched_setscheduler(struct task_struct
*, int,
2417 const struct sched_param
*);
2418 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2419 const struct sched_param
*);
2420 extern int sched_setattr(struct task_struct
*,
2421 const struct sched_attr
*);
2422 extern struct task_struct
*idle_task(int cpu
);
2424 * is_idle_task - is the specified task an idle task?
2425 * @p: the task in question.
2427 * Return: 1 if @p is an idle task. 0 otherwise.
2429 static inline bool is_idle_task(const struct task_struct
*p
)
2433 extern struct task_struct
*curr_task(int cpu
);
2434 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2438 union thread_union
{
2439 struct thread_info thread_info
;
2440 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2443 #ifndef __HAVE_ARCH_KSTACK_END
2444 static inline int kstack_end(void *addr
)
2446 /* Reliable end of stack detection:
2447 * Some APM bios versions misalign the stack
2449 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2453 extern union thread_union init_thread_union
;
2454 extern struct task_struct init_task
;
2456 extern struct mm_struct init_mm
;
2458 extern struct pid_namespace init_pid_ns
;
2461 * find a task by one of its numerical ids
2463 * find_task_by_pid_ns():
2464 * finds a task by its pid in the specified namespace
2465 * find_task_by_vpid():
2466 * finds a task by its virtual pid
2468 * see also find_vpid() etc in include/linux/pid.h
2471 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2472 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2473 struct pid_namespace
*ns
);
2475 /* per-UID process charging. */
2476 extern struct user_struct
* alloc_uid(kuid_t
);
2477 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2479 atomic_inc(&u
->__count
);
2482 extern void free_uid(struct user_struct
*);
2484 #include <asm/current.h>
2486 extern void xtime_update(unsigned long ticks
);
2488 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2489 extern int wake_up_process(struct task_struct
*tsk
);
2490 extern void wake_up_new_task(struct task_struct
*tsk
);
2492 extern void kick_process(struct task_struct
*tsk
);
2494 static inline void kick_process(struct task_struct
*tsk
) { }
2496 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2497 extern void sched_dead(struct task_struct
*p
);
2499 extern void proc_caches_init(void);
2500 extern void flush_signals(struct task_struct
*);
2501 extern void ignore_signals(struct task_struct
*);
2502 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2503 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2505 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2507 struct task_struct
*tsk
= current
;
2511 spin_lock_irq(&tsk
->sighand
->siglock
);
2512 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2513 spin_unlock_irq(&tsk
->sighand
->siglock
);
2518 static inline void kernel_signal_stop(void)
2520 spin_lock_irq(¤t
->sighand
->siglock
);
2521 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2522 __set_current_state(TASK_STOPPED
);
2523 spin_unlock_irq(¤t
->sighand
->siglock
);
2528 extern void release_task(struct task_struct
* p
);
2529 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2530 extern int force_sigsegv(int, struct task_struct
*);
2531 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2532 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2533 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2534 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2535 const struct cred
*, u32
);
2536 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2537 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2538 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2539 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2540 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2541 extern void force_sig(int, struct task_struct
*);
2542 extern int send_sig(int, struct task_struct
*, int);
2543 extern int zap_other_threads(struct task_struct
*p
);
2544 extern struct sigqueue
*sigqueue_alloc(void);
2545 extern void sigqueue_free(struct sigqueue
*);
2546 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2547 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2549 static inline void restore_saved_sigmask(void)
2551 if (test_and_clear_restore_sigmask())
2552 __set_current_blocked(¤t
->saved_sigmask
);
2555 static inline sigset_t
*sigmask_to_save(void)
2557 sigset_t
*res
= ¤t
->blocked
;
2558 if (unlikely(test_restore_sigmask()))
2559 res
= ¤t
->saved_sigmask
;
2563 static inline int kill_cad_pid(int sig
, int priv
)
2565 return kill_pid(cad_pid
, sig
, priv
);
2568 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2569 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2570 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2571 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2574 * True if we are on the alternate signal stack.
2576 static inline int on_sig_stack(unsigned long sp
)
2578 #ifdef CONFIG_STACK_GROWSUP
2579 return sp
>= current
->sas_ss_sp
&&
2580 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2582 return sp
> current
->sas_ss_sp
&&
2583 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2587 static inline int sas_ss_flags(unsigned long sp
)
2589 if (!current
->sas_ss_size
)
2592 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2595 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2597 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2598 #ifdef CONFIG_STACK_GROWSUP
2599 return current
->sas_ss_sp
;
2601 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2607 * Routines for handling mm_structs
2609 extern struct mm_struct
* mm_alloc(void);
2611 /* mmdrop drops the mm and the page tables */
2612 extern void __mmdrop(struct mm_struct
*);
2613 static inline void mmdrop(struct mm_struct
* mm
)
2615 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2619 /* mmput gets rid of the mappings and all user-space */
2620 extern void mmput(struct mm_struct
*);
2621 /* Grab a reference to a task's mm, if it is not already going away */
2622 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2624 * Grab a reference to a task's mm, if it is not already going away
2625 * and ptrace_may_access with the mode parameter passed to it
2628 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2629 /* Remove the current tasks stale references to the old mm_struct */
2630 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2632 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2633 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2634 struct task_struct
*, unsigned long);
2636 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2637 struct task_struct
*);
2639 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2640 * via pt_regs, so ignore the tls argument passed via C. */
2641 static inline int copy_thread_tls(
2642 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2643 struct task_struct
*p
, unsigned long tls
)
2645 return copy_thread(clone_flags
, sp
, arg
, p
);
2648 extern void flush_thread(void);
2649 extern void exit_thread(void);
2651 extern void exit_files(struct task_struct
*);
2652 extern void __cleanup_sighand(struct sighand_struct
*);
2654 extern void exit_itimers(struct signal_struct
*);
2655 extern void flush_itimer_signals(void);
2657 extern void do_group_exit(int);
2659 extern int do_execve(struct filename
*,
2660 const char __user
* const __user
*,
2661 const char __user
* const __user
*);
2662 extern int do_execveat(int, struct filename
*,
2663 const char __user
* const __user
*,
2664 const char __user
* const __user
*,
2666 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
2667 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2668 struct task_struct
*fork_idle(int);
2669 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2671 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
2672 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
2674 __set_task_comm(tsk
, from
, false);
2676 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2679 void scheduler_ipi(void);
2680 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2682 static inline void scheduler_ipi(void) { }
2683 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2690 #define tasklist_empty() \
2691 list_empty(&init_task.tasks)
2693 #define next_task(p) \
2694 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2696 #define for_each_process(p) \
2697 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2699 extern bool current_is_single_threaded(void);
2702 * Careful: do_each_thread/while_each_thread is a double loop so
2703 * 'break' will not work as expected - use goto instead.
2705 #define do_each_thread(g, t) \
2706 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2708 #define while_each_thread(g, t) \
2709 while ((t = next_thread(t)) != g)
2711 #define __for_each_thread(signal, t) \
2712 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2714 #define for_each_thread(p, t) \
2715 __for_each_thread((p)->signal, t)
2717 /* Careful: this is a double loop, 'break' won't work as expected. */
2718 #define for_each_process_thread(p, t) \
2719 for_each_process(p) for_each_thread(p, t)
2721 static inline int get_nr_threads(struct task_struct
*tsk
)
2723 return tsk
->signal
->nr_threads
;
2726 static inline bool thread_group_leader(struct task_struct
*p
)
2728 return p
->exit_signal
>= 0;
2731 /* Do to the insanities of de_thread it is possible for a process
2732 * to have the pid of the thread group leader without actually being
2733 * the thread group leader. For iteration through the pids in proc
2734 * all we care about is that we have a task with the appropriate
2735 * pid, we don't actually care if we have the right task.
2737 static inline bool has_group_leader_pid(struct task_struct
*p
)
2739 return task_pid(p
) == p
->signal
->leader_pid
;
2743 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2745 return p1
->signal
== p2
->signal
;
2748 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2750 return list_entry_rcu(p
->thread_group
.next
,
2751 struct task_struct
, thread_group
);
2754 static inline int thread_group_empty(struct task_struct
*p
)
2756 return list_empty(&p
->thread_group
);
2759 #define delay_group_leader(p) \
2760 (thread_group_leader(p) && !thread_group_empty(p))
2763 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2764 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2765 * pins the final release of task.io_context. Also protects ->cpuset and
2766 * ->cgroup.subsys[]. And ->vfork_done.
2768 * Nests both inside and outside of read_lock(&tasklist_lock).
2769 * It must not be nested with write_lock_irq(&tasklist_lock),
2770 * neither inside nor outside.
2772 static inline void task_lock(struct task_struct
*p
)
2774 spin_lock(&p
->alloc_lock
);
2777 static inline void task_unlock(struct task_struct
*p
)
2779 spin_unlock(&p
->alloc_lock
);
2782 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2783 unsigned long *flags
);
2785 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2786 unsigned long *flags
)
2788 struct sighand_struct
*ret
;
2790 ret
= __lock_task_sighand(tsk
, flags
);
2791 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2795 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2796 unsigned long *flags
)
2798 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2802 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2803 * @tsk: task causing the changes
2805 * All operations which modify a threadgroup - a new thread joining the
2806 * group, death of a member thread (the assertion of PF_EXITING) and
2807 * exec(2) dethreading the process and replacing the leader - are wrapped
2808 * by threadgroup_change_{begin|end}(). This is to provide a place which
2809 * subsystems needing threadgroup stability can hook into for
2812 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
2815 cgroup_threadgroup_change_begin(tsk
);
2819 * threadgroup_change_end - mark the end of changes to a threadgroup
2820 * @tsk: task causing the changes
2822 * See threadgroup_change_begin().
2824 static inline void threadgroup_change_end(struct task_struct
*tsk
)
2826 cgroup_threadgroup_change_end(tsk
);
2829 #ifndef __HAVE_THREAD_FUNCTIONS
2831 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2832 #define task_stack_page(task) ((task)->stack)
2834 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2836 *task_thread_info(p
) = *task_thread_info(org
);
2837 task_thread_info(p
)->task
= p
;
2841 * Return the address of the last usable long on the stack.
2843 * When the stack grows down, this is just above the thread
2844 * info struct. Going any lower will corrupt the threadinfo.
2846 * When the stack grows up, this is the highest address.
2847 * Beyond that position, we corrupt data on the next page.
2849 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2851 #ifdef CONFIG_STACK_GROWSUP
2852 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
2854 return (unsigned long *)(task_thread_info(p
) + 1);
2859 #define task_stack_end_corrupted(task) \
2860 (*(end_of_stack(task)) != STACK_END_MAGIC)
2862 static inline int object_is_on_stack(void *obj
)
2864 void *stack
= task_stack_page(current
);
2866 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2869 extern void thread_info_cache_init(void);
2871 #ifdef CONFIG_DEBUG_STACK_USAGE
2872 static inline unsigned long stack_not_used(struct task_struct
*p
)
2874 unsigned long *n
= end_of_stack(p
);
2876 do { /* Skip over canary */
2877 # ifdef CONFIG_STACK_GROWSUP
2884 # ifdef CONFIG_STACK_GROWSUP
2885 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
2887 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2891 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
2893 /* set thread flags in other task's structures
2894 * - see asm/thread_info.h for TIF_xxxx flags available
2896 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2898 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2901 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2903 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2906 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2908 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2911 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2913 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2916 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2918 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2921 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2923 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2926 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2928 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2931 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2933 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2936 static inline int restart_syscall(void)
2938 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2939 return -ERESTARTNOINTR
;
2942 static inline int signal_pending(struct task_struct
*p
)
2944 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2947 static inline int __fatal_signal_pending(struct task_struct
*p
)
2949 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2952 static inline int fatal_signal_pending(struct task_struct
*p
)
2954 return signal_pending(p
) && __fatal_signal_pending(p
);
2957 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2959 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2961 if (!signal_pending(p
))
2964 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2968 * cond_resched() and cond_resched_lock(): latency reduction via
2969 * explicit rescheduling in places that are safe. The return
2970 * value indicates whether a reschedule was done in fact.
2971 * cond_resched_lock() will drop the spinlock before scheduling,
2972 * cond_resched_softirq() will enable bhs before scheduling.
2974 extern int _cond_resched(void);
2976 #define cond_resched() ({ \
2977 ___might_sleep(__FILE__, __LINE__, 0); \
2981 extern int __cond_resched_lock(spinlock_t
*lock
);
2983 #define cond_resched_lock(lock) ({ \
2984 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2985 __cond_resched_lock(lock); \
2988 extern int __cond_resched_softirq(void);
2990 #define cond_resched_softirq() ({ \
2991 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2992 __cond_resched_softirq(); \
2995 static inline void cond_resched_rcu(void)
2997 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3005 * Does a critical section need to be broken due to another
3006 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3007 * but a general need for low latency)
3009 static inline int spin_needbreak(spinlock_t
*lock
)
3011 #ifdef CONFIG_PREEMPT
3012 return spin_is_contended(lock
);
3019 * Idle thread specific functions to determine the need_resched
3022 #ifdef TIF_POLLING_NRFLAG
3023 static inline int tsk_is_polling(struct task_struct
*p
)
3025 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
3028 static inline void __current_set_polling(void)
3030 set_thread_flag(TIF_POLLING_NRFLAG
);
3033 static inline bool __must_check
current_set_polling_and_test(void)
3035 __current_set_polling();
3038 * Polling state must be visible before we test NEED_RESCHED,
3039 * paired by resched_curr()
3041 smp_mb__after_atomic();
3043 return unlikely(tif_need_resched());
3046 static inline void __current_clr_polling(void)
3048 clear_thread_flag(TIF_POLLING_NRFLAG
);
3051 static inline bool __must_check
current_clr_polling_and_test(void)
3053 __current_clr_polling();
3056 * Polling state must be visible before we test NEED_RESCHED,
3057 * paired by resched_curr()
3059 smp_mb__after_atomic();
3061 return unlikely(tif_need_resched());
3065 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
3066 static inline void __current_set_polling(void) { }
3067 static inline void __current_clr_polling(void) { }
3069 static inline bool __must_check
current_set_polling_and_test(void)
3071 return unlikely(tif_need_resched());
3073 static inline bool __must_check
current_clr_polling_and_test(void)
3075 return unlikely(tif_need_resched());
3079 static inline void current_clr_polling(void)
3081 __current_clr_polling();
3084 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3085 * Once the bit is cleared, we'll get IPIs with every new
3086 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3089 smp_mb(); /* paired with resched_curr() */
3091 preempt_fold_need_resched();
3094 static __always_inline
bool need_resched(void)
3096 return unlikely(tif_need_resched());
3100 * Thread group CPU time accounting.
3102 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
3103 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
3106 * Reevaluate whether the task has signals pending delivery.
3107 * Wake the task if so.
3108 * This is required every time the blocked sigset_t changes.
3109 * callers must hold sighand->siglock.
3111 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
3112 extern void recalc_sigpending(void);
3114 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
3116 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
3118 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
3120 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
3122 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
3126 * Wrappers for p->thread_info->cpu access. No-op on UP.
3130 static inline unsigned int task_cpu(const struct task_struct
*p
)
3132 return task_thread_info(p
)->cpu
;
3135 static inline int task_node(const struct task_struct
*p
)
3137 return cpu_to_node(task_cpu(p
));
3140 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
3144 static inline unsigned int task_cpu(const struct task_struct
*p
)
3149 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
3153 #endif /* CONFIG_SMP */
3155 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3156 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3158 #ifdef CONFIG_CGROUP_SCHED
3159 extern struct task_group root_task_group
;
3160 #endif /* CONFIG_CGROUP_SCHED */
3162 extern int task_can_switch_user(struct user_struct
*up
,
3163 struct task_struct
*tsk
);
3165 #ifdef CONFIG_TASK_XACCT
3166 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3168 tsk
->ioac
.rchar
+= amt
;
3171 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3173 tsk
->ioac
.wchar
+= amt
;
3176 static inline void inc_syscr(struct task_struct
*tsk
)
3181 static inline void inc_syscw(struct task_struct
*tsk
)
3186 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3190 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3194 static inline void inc_syscr(struct task_struct
*tsk
)
3198 static inline void inc_syscw(struct task_struct
*tsk
)
3203 #ifndef TASK_SIZE_OF
3204 #define TASK_SIZE_OF(tsk) TASK_SIZE
3208 extern void mm_update_next_owner(struct mm_struct
*mm
);
3210 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3213 #endif /* CONFIG_MEMCG */
3215 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3218 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3221 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3224 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3227 static inline unsigned long rlimit(unsigned int limit
)
3229 return task_rlimit(current
, limit
);
3232 static inline unsigned long rlimit_max(unsigned int limit
)
3234 return task_rlimit_max(current
, limit
);
3237 #ifdef CONFIG_CPU_FREQ
3238 struct update_util_data
{
3239 void (*func
)(struct update_util_data
*data
,
3240 u64 time
, unsigned long util
, unsigned long max
);
3243 void cpufreq_set_update_util_data(int cpu
, struct update_util_data
*data
);
3244 #endif /* CONFIG_CPU_FREQ */