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/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/kcov.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60 #include <linux/magic.h>
61 #include <linux/cgroup-defs.h>
63 #include <asm/processor.h>
65 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
68 * Extended scheduling parameters data structure.
70 * This is needed because the original struct sched_param can not be
71 * altered without introducing ABI issues with legacy applications
72 * (e.g., in sched_getparam()).
74 * However, the possibility of specifying more than just a priority for
75 * the tasks may be useful for a wide variety of application fields, e.g.,
76 * multimedia, streaming, automation and control, and many others.
78 * This variant (sched_attr) is meant at describing a so-called
79 * sporadic time-constrained task. In such model a task is specified by:
80 * - the activation period or minimum instance inter-arrival time;
81 * - the maximum (or average, depending on the actual scheduling
82 * discipline) computation time of all instances, a.k.a. runtime;
83 * - the deadline (relative to the actual activation time) of each
85 * Very briefly, a periodic (sporadic) task asks for the execution of
86 * some specific computation --which is typically called an instance--
87 * (at most) every period. Moreover, each instance typically lasts no more
88 * than the runtime and must be completed by time instant t equal to
89 * the instance activation time + the deadline.
91 * This is reflected by the actual fields of the sched_attr structure:
93 * @size size of the structure, for fwd/bwd compat.
95 * @sched_policy task's scheduling policy
96 * @sched_flags for customizing the scheduler behaviour
97 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
98 * @sched_priority task's static priority (SCHED_FIFO/RR)
99 * @sched_deadline representative of the task's deadline
100 * @sched_runtime representative of the task's runtime
101 * @sched_period representative of the task's period
103 * Given this task model, there are a multiplicity of scheduling algorithms
104 * and policies, that can be used to ensure all the tasks will make their
105 * timing constraints.
107 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
108 * only user of this new interface. More information about the algorithm
109 * available in the scheduling class file or in Documentation/.
117 /* SCHED_NORMAL, SCHED_BATCH */
120 /* SCHED_FIFO, SCHED_RR */
129 struct futex_pi_state
;
130 struct robust_list_head
;
133 struct perf_event_context
;
138 #define VMACACHE_BITS 2
139 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
140 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 * These are the constant used to fake the fixed-point load-average
144 * counting. Some notes:
145 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
146 * a load-average precision of 10 bits integer + 11 bits fractional
147 * - if you want to count load-averages more often, you need more
148 * precision, or rounding will get you. With 2-second counting freq,
149 * the EXP_n values would be 1981, 2034 and 2043 if still using only
152 extern unsigned long avenrun
[]; /* Load averages */
153 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
155 #define FSHIFT 11 /* nr of bits of precision */
156 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
157 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
158 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
159 #define EXP_5 2014 /* 1/exp(5sec/5min) */
160 #define EXP_15 2037 /* 1/exp(5sec/15min) */
162 #define CALC_LOAD(load,exp,n) \
164 load += n*(FIXED_1-exp); \
167 extern unsigned long total_forks
;
168 extern int nr_threads
;
169 DECLARE_PER_CPU(unsigned long, process_counts
);
170 extern int nr_processes(void);
171 extern unsigned long nr_running(void);
172 extern bool single_task_running(void);
173 extern unsigned long nr_iowait(void);
174 extern unsigned long nr_iowait_cpu(int cpu
);
175 extern void get_iowait_load(unsigned long *nr_waiters
, unsigned long *load
);
177 extern void calc_global_load(unsigned long ticks
);
179 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
180 extern void cpu_load_update_nohz_start(void);
181 extern void cpu_load_update_nohz_stop(void);
183 static inline void cpu_load_update_nohz_start(void) { }
184 static inline void cpu_load_update_nohz_stop(void) { }
187 extern void dump_cpu_task(int cpu
);
192 #ifdef CONFIG_SCHED_DEBUG
193 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
194 extern void proc_sched_set_task(struct task_struct
*p
);
198 * Task state bitmask. NOTE! These bits are also
199 * encoded in fs/proc/array.c: get_task_state().
201 * We have two separate sets of flags: task->state
202 * is about runnability, while task->exit_state are
203 * about the task exiting. Confusing, but this way
204 * modifying one set can't modify the other one by
207 #define TASK_RUNNING 0
208 #define TASK_INTERRUPTIBLE 1
209 #define TASK_UNINTERRUPTIBLE 2
210 #define __TASK_STOPPED 4
211 #define __TASK_TRACED 8
212 /* in tsk->exit_state */
214 #define EXIT_ZOMBIE 32
215 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
216 /* in tsk->state again */
218 #define TASK_WAKEKILL 128
219 #define TASK_WAKING 256
220 #define TASK_PARKED 512
221 #define TASK_NOLOAD 1024
222 #define TASK_STATE_MAX 2048
224 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
226 extern char ___assert_task_state
[1 - 2*!!(
227 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
229 /* Convenience macros for the sake of set_task_state */
230 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
231 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
232 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
234 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
236 /* Convenience macros for the sake of wake_up */
237 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
238 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
240 /* get_task_state() */
241 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
242 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
243 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
245 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
246 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
247 #define task_is_stopped_or_traced(task) \
248 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
249 #define task_contributes_to_load(task) \
250 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
251 (task->flags & PF_FROZEN) == 0 && \
252 (task->state & TASK_NOLOAD) == 0)
254 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
256 #define __set_task_state(tsk, state_value) \
258 (tsk)->task_state_change = _THIS_IP_; \
259 (tsk)->state = (state_value); \
261 #define set_task_state(tsk, state_value) \
263 (tsk)->task_state_change = _THIS_IP_; \
264 smp_store_mb((tsk)->state, (state_value)); \
268 * set_current_state() includes a barrier so that the write of current->state
269 * is correctly serialised wrt the caller's subsequent test of whether to
272 * set_current_state(TASK_UNINTERRUPTIBLE);
273 * if (do_i_need_to_sleep())
276 * If the caller does not need such serialisation then use __set_current_state()
278 #define __set_current_state(state_value) \
280 current->task_state_change = _THIS_IP_; \
281 current->state = (state_value); \
283 #define set_current_state(state_value) \
285 current->task_state_change = _THIS_IP_; \
286 smp_store_mb(current->state, (state_value)); \
291 #define __set_task_state(tsk, state_value) \
292 do { (tsk)->state = (state_value); } while (0)
293 #define set_task_state(tsk, state_value) \
294 smp_store_mb((tsk)->state, (state_value))
297 * set_current_state() includes a barrier so that the write of current->state
298 * is correctly serialised wrt the caller's subsequent test of whether to
301 * set_current_state(TASK_UNINTERRUPTIBLE);
302 * if (do_i_need_to_sleep())
305 * If the caller does not need such serialisation then use __set_current_state()
307 #define __set_current_state(state_value) \
308 do { current->state = (state_value); } while (0)
309 #define set_current_state(state_value) \
310 smp_store_mb(current->state, (state_value))
314 /* Task command name length */
315 #define TASK_COMM_LEN 16
317 #include <linux/spinlock.h>
320 * This serializes "schedule()" and also protects
321 * the run-queue from deletions/modifications (but
322 * _adding_ to the beginning of the run-queue has
325 extern rwlock_t tasklist_lock
;
326 extern spinlock_t mmlist_lock
;
330 #ifdef CONFIG_PROVE_RCU
331 extern int lockdep_tasklist_lock_is_held(void);
332 #endif /* #ifdef CONFIG_PROVE_RCU */
334 extern void sched_init(void);
335 extern void sched_init_smp(void);
336 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
337 extern void init_idle(struct task_struct
*idle
, int cpu
);
338 extern void init_idle_bootup_task(struct task_struct
*idle
);
340 extern cpumask_var_t cpu_isolated_map
;
342 extern int runqueue_is_locked(int cpu
);
344 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
345 extern void nohz_balance_enter_idle(int cpu
);
346 extern void set_cpu_sd_state_idle(void);
347 extern int get_nohz_timer_target(void);
349 static inline void nohz_balance_enter_idle(int cpu
) { }
350 static inline void set_cpu_sd_state_idle(void) { }
354 * Only dump TASK_* tasks. (0 for all tasks)
356 extern void show_state_filter(unsigned long state_filter
);
358 static inline void show_state(void)
360 show_state_filter(0);
363 extern void show_regs(struct pt_regs
*);
366 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
367 * task), SP is the stack pointer of the first frame that should be shown in the back
368 * trace (or NULL if the entire call-chain of the task should be shown).
370 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
372 extern void cpu_init (void);
373 extern void trap_init(void);
374 extern void update_process_times(int user
);
375 extern void scheduler_tick(void);
376 extern int sched_cpu_starting(unsigned int cpu
);
377 extern int sched_cpu_activate(unsigned int cpu
);
378 extern int sched_cpu_deactivate(unsigned int cpu
);
380 #ifdef CONFIG_HOTPLUG_CPU
381 extern int sched_cpu_dying(unsigned int cpu
);
383 # define sched_cpu_dying NULL
386 extern void sched_show_task(struct task_struct
*p
);
388 #ifdef CONFIG_LOCKUP_DETECTOR
389 extern void touch_softlockup_watchdog_sched(void);
390 extern void touch_softlockup_watchdog(void);
391 extern void touch_softlockup_watchdog_sync(void);
392 extern void touch_all_softlockup_watchdogs(void);
393 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
395 size_t *lenp
, loff_t
*ppos
);
396 extern unsigned int softlockup_panic
;
397 extern unsigned int hardlockup_panic
;
398 void lockup_detector_init(void);
400 static inline void touch_softlockup_watchdog_sched(void)
403 static inline void touch_softlockup_watchdog(void)
406 static inline void touch_softlockup_watchdog_sync(void)
409 static inline void touch_all_softlockup_watchdogs(void)
412 static inline void lockup_detector_init(void)
417 #ifdef CONFIG_DETECT_HUNG_TASK
418 void reset_hung_task_detector(void);
420 static inline void reset_hung_task_detector(void)
425 /* Attach to any functions which should be ignored in wchan output. */
426 #define __sched __attribute__((__section__(".sched.text")))
428 /* Linker adds these: start and end of __sched functions */
429 extern char __sched_text_start
[], __sched_text_end
[];
431 /* Is this address in the __sched functions? */
432 extern int in_sched_functions(unsigned long addr
);
434 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
435 extern signed long schedule_timeout(signed long timeout
);
436 extern signed long schedule_timeout_interruptible(signed long timeout
);
437 extern signed long schedule_timeout_killable(signed long timeout
);
438 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
439 extern signed long schedule_timeout_idle(signed long timeout
);
440 asmlinkage
void schedule(void);
441 extern void schedule_preempt_disabled(void);
443 extern long io_schedule_timeout(long timeout
);
445 static inline void io_schedule(void)
447 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT
);
451 struct user_namespace
;
454 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
456 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
457 unsigned long, unsigned long);
459 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
460 unsigned long len
, unsigned long pgoff
,
461 unsigned long flags
);
463 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
466 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
467 #define SUID_DUMP_USER 1 /* Dump as user of process */
468 #define SUID_DUMP_ROOT 2 /* Dump as root */
472 /* for SUID_DUMP_* above */
473 #define MMF_DUMPABLE_BITS 2
474 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
476 extern void set_dumpable(struct mm_struct
*mm
, int value
);
478 * This returns the actual value of the suid_dumpable flag. For things
479 * that are using this for checking for privilege transitions, it must
480 * test against SUID_DUMP_USER rather than treating it as a boolean
483 static inline int __get_dumpable(unsigned long mm_flags
)
485 return mm_flags
& MMF_DUMPABLE_MASK
;
488 static inline int get_dumpable(struct mm_struct
*mm
)
490 return __get_dumpable(mm
->flags
);
493 /* coredump filter bits */
494 #define MMF_DUMP_ANON_PRIVATE 2
495 #define MMF_DUMP_ANON_SHARED 3
496 #define MMF_DUMP_MAPPED_PRIVATE 4
497 #define MMF_DUMP_MAPPED_SHARED 5
498 #define MMF_DUMP_ELF_HEADERS 6
499 #define MMF_DUMP_HUGETLB_PRIVATE 7
500 #define MMF_DUMP_HUGETLB_SHARED 8
501 #define MMF_DUMP_DAX_PRIVATE 9
502 #define MMF_DUMP_DAX_SHARED 10
504 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
505 #define MMF_DUMP_FILTER_BITS 9
506 #define MMF_DUMP_FILTER_MASK \
507 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
508 #define MMF_DUMP_FILTER_DEFAULT \
509 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
510 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
512 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
513 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
515 # define MMF_DUMP_MASK_DEFAULT_ELF 0
517 /* leave room for more dump flags */
518 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
519 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
520 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
522 #define MMF_HAS_UPROBES 19 /* has uprobes */
523 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
525 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
527 struct sighand_struct
{
529 struct k_sigaction action
[_NSIG
];
531 wait_queue_head_t signalfd_wqh
;
534 struct pacct_struct
{
537 unsigned long ac_mem
;
538 cputime_t ac_utime
, ac_stime
;
539 unsigned long ac_minflt
, ac_majflt
;
550 * struct prev_cputime - snaphsot of system and user cputime
551 * @utime: time spent in user mode
552 * @stime: time spent in system mode
553 * @lock: protects the above two fields
555 * Stores previous user/system time values such that we can guarantee
558 struct prev_cputime
{
559 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
566 static inline void prev_cputime_init(struct prev_cputime
*prev
)
568 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
569 prev
->utime
= prev
->stime
= 0;
570 raw_spin_lock_init(&prev
->lock
);
575 * struct task_cputime - collected CPU time counts
576 * @utime: time spent in user mode, in &cputime_t units
577 * @stime: time spent in kernel mode, in &cputime_t units
578 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
580 * This structure groups together three kinds of CPU time that are tracked for
581 * threads and thread groups. Most things considering CPU time want to group
582 * these counts together and treat all three of them in parallel.
584 struct task_cputime
{
587 unsigned long long sum_exec_runtime
;
590 /* Alternate field names when used to cache expirations. */
591 #define virt_exp utime
592 #define prof_exp stime
593 #define sched_exp sum_exec_runtime
595 #define INIT_CPUTIME \
596 (struct task_cputime) { \
599 .sum_exec_runtime = 0, \
603 * This is the atomic variant of task_cputime, which can be used for
604 * storing and updating task_cputime statistics without locking.
606 struct task_cputime_atomic
{
609 atomic64_t sum_exec_runtime
;
612 #define INIT_CPUTIME_ATOMIC \
613 (struct task_cputime_atomic) { \
614 .utime = ATOMIC64_INIT(0), \
615 .stime = ATOMIC64_INIT(0), \
616 .sum_exec_runtime = ATOMIC64_INIT(0), \
619 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
622 * Disable preemption until the scheduler is running -- use an unconditional
623 * value so that it also works on !PREEMPT_COUNT kernels.
625 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
627 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
630 * Initial preempt_count value; reflects the preempt_count schedule invariant
631 * which states that during context switches:
633 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
635 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
636 * Note: See finish_task_switch().
638 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
641 * struct thread_group_cputimer - thread group interval timer counts
642 * @cputime_atomic: atomic thread group interval timers.
643 * @running: true when there are timers running and
644 * @cputime_atomic receives updates.
645 * @checking_timer: true when a thread in the group is in the
646 * process of checking for thread group timers.
648 * This structure contains the version of task_cputime, above, that is
649 * used for thread group CPU timer calculations.
651 struct thread_group_cputimer
{
652 struct task_cputime_atomic cputime_atomic
;
657 #include <linux/rwsem.h>
661 * NOTE! "signal_struct" does not have its own
662 * locking, because a shared signal_struct always
663 * implies a shared sighand_struct, so locking
664 * sighand_struct is always a proper superset of
665 * the locking of signal_struct.
667 struct signal_struct
{
671 struct list_head thread_head
;
673 wait_queue_head_t wait_chldexit
; /* for wait4() */
675 /* current thread group signal load-balancing target: */
676 struct task_struct
*curr_target
;
678 /* shared signal handling: */
679 struct sigpending shared_pending
;
681 /* thread group exit support */
684 * - notify group_exit_task when ->count is equal to notify_count
685 * - everyone except group_exit_task is stopped during signal delivery
686 * of fatal signals, group_exit_task processes the signal.
689 struct task_struct
*group_exit_task
;
691 /* thread group stop support, overloads group_exit_code too */
692 int group_stop_count
;
693 unsigned int flags
; /* see SIGNAL_* flags below */
696 * PR_SET_CHILD_SUBREAPER marks a process, like a service
697 * manager, to re-parent orphan (double-forking) child processes
698 * to this process instead of 'init'. The service manager is
699 * able to receive SIGCHLD signals and is able to investigate
700 * the process until it calls wait(). All children of this
701 * process will inherit a flag if they should look for a
702 * child_subreaper process at exit.
704 unsigned int is_child_subreaper
:1;
705 unsigned int has_child_subreaper
:1;
707 /* POSIX.1b Interval Timers */
709 struct list_head posix_timers
;
711 /* ITIMER_REAL timer for the process */
712 struct hrtimer real_timer
;
713 struct pid
*leader_pid
;
714 ktime_t it_real_incr
;
717 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
718 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
719 * values are defined to 0 and 1 respectively
721 struct cpu_itimer it
[2];
724 * Thread group totals for process CPU timers.
725 * See thread_group_cputimer(), et al, for details.
727 struct thread_group_cputimer cputimer
;
729 /* Earliest-expiration cache. */
730 struct task_cputime cputime_expires
;
732 #ifdef CONFIG_NO_HZ_FULL
733 atomic_t tick_dep_mask
;
736 struct list_head cpu_timers
[3];
738 struct pid
*tty_old_pgrp
;
740 /* boolean value for session group leader */
743 struct tty_struct
*tty
; /* NULL if no tty */
745 #ifdef CONFIG_SCHED_AUTOGROUP
746 struct autogroup
*autogroup
;
749 * Cumulative resource counters for dead threads in the group,
750 * and for reaped dead child processes forked by this group.
751 * Live threads maintain their own counters and add to these
752 * in __exit_signal, except for the group leader.
754 seqlock_t stats_lock
;
755 cputime_t utime
, stime
, cutime
, cstime
;
758 struct prev_cputime prev_cputime
;
759 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
760 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
761 unsigned long inblock
, oublock
, cinblock
, coublock
;
762 unsigned long maxrss
, cmaxrss
;
763 struct task_io_accounting ioac
;
766 * Cumulative ns of schedule CPU time fo dead threads in the
767 * group, not including a zombie group leader, (This only differs
768 * from jiffies_to_ns(utime + stime) if sched_clock uses something
769 * other than jiffies.)
771 unsigned long long sum_sched_runtime
;
774 * We don't bother to synchronize most readers of this at all,
775 * because there is no reader checking a limit that actually needs
776 * to get both rlim_cur and rlim_max atomically, and either one
777 * alone is a single word that can safely be read normally.
778 * getrlimit/setrlimit use task_lock(current->group_leader) to
779 * protect this instead of the siglock, because they really
780 * have no need to disable irqs.
782 struct rlimit rlim
[RLIM_NLIMITS
];
784 #ifdef CONFIG_BSD_PROCESS_ACCT
785 struct pacct_struct pacct
; /* per-process accounting information */
787 #ifdef CONFIG_TASKSTATS
788 struct taskstats
*stats
;
792 struct tty_audit_buf
*tty_audit_buf
;
795 oom_flags_t oom_flags
;
796 short oom_score_adj
; /* OOM kill score adjustment */
797 short oom_score_adj_min
; /* OOM kill score adjustment min value.
798 * Only settable by CAP_SYS_RESOURCE. */
800 struct mutex cred_guard_mutex
; /* guard against foreign influences on
801 * credential calculations
802 * (notably. ptrace) */
806 * Bits in flags field of signal_struct.
808 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
809 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
810 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
811 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
813 * Pending notifications to parent.
815 #define SIGNAL_CLD_STOPPED 0x00000010
816 #define SIGNAL_CLD_CONTINUED 0x00000020
817 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
819 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
821 /* If true, all threads except ->group_exit_task have pending SIGKILL */
822 static inline int signal_group_exit(const struct signal_struct
*sig
)
824 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
825 (sig
->group_exit_task
!= NULL
);
829 * Some day this will be a full-fledged user tracking system..
832 atomic_t __count
; /* reference count */
833 atomic_t processes
; /* How many processes does this user have? */
834 atomic_t sigpending
; /* How many pending signals does this user have? */
835 #ifdef CONFIG_INOTIFY_USER
836 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
837 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
839 #ifdef CONFIG_FANOTIFY
840 atomic_t fanotify_listeners
;
843 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
845 #ifdef CONFIG_POSIX_MQUEUE
846 /* protected by mq_lock */
847 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
849 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
850 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
851 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
854 struct key
*uid_keyring
; /* UID specific keyring */
855 struct key
*session_keyring
; /* UID's default session keyring */
858 /* Hash table maintenance information */
859 struct hlist_node uidhash_node
;
862 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
863 atomic_long_t locked_vm
;
867 extern int uids_sysfs_init(void);
869 extern struct user_struct
*find_user(kuid_t
);
871 extern struct user_struct root_user
;
872 #define INIT_USER (&root_user)
875 struct backing_dev_info
;
876 struct reclaim_state
;
878 #ifdef CONFIG_SCHED_INFO
880 /* cumulative counters */
881 unsigned long pcount
; /* # of times run on this cpu */
882 unsigned long long run_delay
; /* time spent waiting on a runqueue */
885 unsigned long long last_arrival
,/* when we last ran on a cpu */
886 last_queued
; /* when we were last queued to run */
888 #endif /* CONFIG_SCHED_INFO */
890 #ifdef CONFIG_TASK_DELAY_ACCT
891 struct task_delay_info
{
893 unsigned int flags
; /* Private per-task flags */
895 /* For each stat XXX, add following, aligned appropriately
897 * struct timespec XXX_start, XXX_end;
901 * Atomicity of updates to XXX_delay, XXX_count protected by
902 * single lock above (split into XXX_lock if contention is an issue).
906 * XXX_count is incremented on every XXX operation, the delay
907 * associated with the operation is added to XXX_delay.
908 * XXX_delay contains the accumulated delay time in nanoseconds.
910 u64 blkio_start
; /* Shared by blkio, swapin */
911 u64 blkio_delay
; /* wait for sync block io completion */
912 u64 swapin_delay
; /* wait for swapin block io completion */
913 u32 blkio_count
; /* total count of the number of sync block */
914 /* io operations performed */
915 u32 swapin_count
; /* total count of the number of swapin block */
916 /* io operations performed */
919 u64 freepages_delay
; /* wait for memory reclaim */
920 u32 freepages_count
; /* total count of memory reclaim */
922 #endif /* CONFIG_TASK_DELAY_ACCT */
924 static inline int sched_info_on(void)
926 #ifdef CONFIG_SCHEDSTATS
928 #elif defined(CONFIG_TASK_DELAY_ACCT)
929 extern int delayacct_on
;
936 #ifdef CONFIG_SCHEDSTATS
937 void force_schedstat_enabled(void);
948 * Integer metrics need fixed point arithmetic, e.g., sched/fair
949 * has a few: load, load_avg, util_avg, freq, and capacity.
951 * We define a basic fixed point arithmetic range, and then formalize
952 * all these metrics based on that basic range.
954 # define SCHED_FIXEDPOINT_SHIFT 10
955 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
958 * Increase resolution of cpu_capacity calculations
960 #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
961 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
964 * Wake-queues are lists of tasks with a pending wakeup, whose
965 * callers have already marked the task as woken internally,
966 * and can thus carry on. A common use case is being able to
967 * do the wakeups once the corresponding user lock as been
970 * We hold reference to each task in the list across the wakeup,
971 * thus guaranteeing that the memory is still valid by the time
972 * the actual wakeups are performed in wake_up_q().
974 * One per task suffices, because there's never a need for a task to be
975 * in two wake queues simultaneously; it is forbidden to abandon a task
976 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
977 * already in a wake queue, the wakeup will happen soon and the second
978 * waker can just skip it.
980 * The WAKE_Q macro declares and initializes the list head.
981 * wake_up_q() does NOT reinitialize the list; it's expected to be
982 * called near the end of a function, where the fact that the queue is
983 * not used again will be easy to see by inspection.
985 * Note that this can cause spurious wakeups. schedule() callers
986 * must ensure the call is done inside a loop, confirming that the
987 * wakeup condition has in fact occurred.
990 struct wake_q_node
*next
;
994 struct wake_q_node
*first
;
995 struct wake_q_node
**lastp
;
998 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1000 #define WAKE_Q(name) \
1001 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1003 extern void wake_q_add(struct wake_q_head
*head
,
1004 struct task_struct
*task
);
1005 extern void wake_up_q(struct wake_q_head
*head
);
1008 * sched-domains (multiprocessor balancing) declarations:
1011 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1012 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1013 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1014 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
1015 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
1016 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
1017 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
1018 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1019 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1020 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1021 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1022 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1023 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1024 #define SD_NUMA 0x4000 /* cross-node balancing */
1026 #ifdef CONFIG_SCHED_SMT
1027 static inline int cpu_smt_flags(void)
1029 return SD_SHARE_CPUCAPACITY
| SD_SHARE_PKG_RESOURCES
;
1033 #ifdef CONFIG_SCHED_MC
1034 static inline int cpu_core_flags(void)
1036 return SD_SHARE_PKG_RESOURCES
;
1041 static inline int cpu_numa_flags(void)
1047 struct sched_domain_attr
{
1048 int relax_domain_level
;
1051 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1052 .relax_domain_level = -1, \
1055 extern int sched_domain_level_max
;
1059 struct sched_domain
{
1060 /* These fields must be setup */
1061 struct sched_domain
*parent
; /* top domain must be null terminated */
1062 struct sched_domain
*child
; /* bottom domain must be null terminated */
1063 struct sched_group
*groups
; /* the balancing groups of the domain */
1064 unsigned long min_interval
; /* Minimum balance interval ms */
1065 unsigned long max_interval
; /* Maximum balance interval ms */
1066 unsigned int busy_factor
; /* less balancing by factor if busy */
1067 unsigned int imbalance_pct
; /* No balance until over watermark */
1068 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
1069 unsigned int busy_idx
;
1070 unsigned int idle_idx
;
1071 unsigned int newidle_idx
;
1072 unsigned int wake_idx
;
1073 unsigned int forkexec_idx
;
1074 unsigned int smt_gain
;
1076 int nohz_idle
; /* NOHZ IDLE status */
1077 int flags
; /* See SD_* */
1080 /* Runtime fields. */
1081 unsigned long last_balance
; /* init to jiffies. units in jiffies */
1082 unsigned int balance_interval
; /* initialise to 1. units in ms. */
1083 unsigned int nr_balance_failed
; /* initialise to 0 */
1085 /* idle_balance() stats */
1086 u64 max_newidle_lb_cost
;
1087 unsigned long next_decay_max_lb_cost
;
1089 #ifdef CONFIG_SCHEDSTATS
1090 /* load_balance() stats */
1091 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
1092 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
1093 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
1094 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
1095 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
1096 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
1097 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
1098 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
1100 /* Active load balancing */
1101 unsigned int alb_count
;
1102 unsigned int alb_failed
;
1103 unsigned int alb_pushed
;
1105 /* SD_BALANCE_EXEC stats */
1106 unsigned int sbe_count
;
1107 unsigned int sbe_balanced
;
1108 unsigned int sbe_pushed
;
1110 /* SD_BALANCE_FORK stats */
1111 unsigned int sbf_count
;
1112 unsigned int sbf_balanced
;
1113 unsigned int sbf_pushed
;
1115 /* try_to_wake_up() stats */
1116 unsigned int ttwu_wake_remote
;
1117 unsigned int ttwu_move_affine
;
1118 unsigned int ttwu_move_balance
;
1120 #ifdef CONFIG_SCHED_DEBUG
1124 void *private; /* used during construction */
1125 struct rcu_head rcu
; /* used during destruction */
1128 unsigned int span_weight
;
1130 * Span of all CPUs in this domain.
1132 * NOTE: this field is variable length. (Allocated dynamically
1133 * by attaching extra space to the end of the structure,
1134 * depending on how many CPUs the kernel has booted up with)
1136 unsigned long span
[0];
1139 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1141 return to_cpumask(sd
->span
);
1144 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1145 struct sched_domain_attr
*dattr_new
);
1147 /* Allocate an array of sched domains, for partition_sched_domains(). */
1148 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1149 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1151 bool cpus_share_cache(int this_cpu
, int that_cpu
);
1153 typedef const struct cpumask
*(*sched_domain_mask_f
)(int cpu
);
1154 typedef int (*sched_domain_flags_f
)(void);
1156 #define SDTL_OVERLAP 0x01
1159 struct sched_domain
**__percpu sd
;
1160 struct sched_group
**__percpu sg
;
1161 struct sched_group_capacity
**__percpu sgc
;
1164 struct sched_domain_topology_level
{
1165 sched_domain_mask_f mask
;
1166 sched_domain_flags_f sd_flags
;
1169 struct sd_data data
;
1170 #ifdef CONFIG_SCHED_DEBUG
1175 extern void set_sched_topology(struct sched_domain_topology_level
*tl
);
1176 extern void wake_up_if_idle(int cpu
);
1178 #ifdef CONFIG_SCHED_DEBUG
1179 # define SD_INIT_NAME(type) .name = #type
1181 # define SD_INIT_NAME(type)
1184 #else /* CONFIG_SMP */
1186 struct sched_domain_attr
;
1189 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1190 struct sched_domain_attr
*dattr_new
)
1194 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1199 #endif /* !CONFIG_SMP */
1202 struct io_context
; /* See blkdev.h */
1205 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1206 extern void prefetch_stack(struct task_struct
*t
);
1208 static inline void prefetch_stack(struct task_struct
*t
) { }
1211 struct audit_context
; /* See audit.c */
1213 struct pipe_inode_info
;
1214 struct uts_namespace
;
1216 struct load_weight
{
1217 unsigned long weight
;
1222 * The load_avg/util_avg accumulates an infinite geometric series
1223 * (see __update_load_avg() in kernel/sched/fair.c).
1225 * [load_avg definition]
1227 * load_avg = runnable% * scale_load_down(load)
1229 * where runnable% is the time ratio that a sched_entity is runnable.
1230 * For cfs_rq, it is the aggregated load_avg of all runnable and
1231 * blocked sched_entities.
1233 * load_avg may also take frequency scaling into account:
1235 * load_avg = runnable% * scale_load_down(load) * freq%
1237 * where freq% is the CPU frequency normalized to the highest frequency.
1239 * [util_avg definition]
1241 * util_avg = running% * SCHED_CAPACITY_SCALE
1243 * where running% is the time ratio that a sched_entity is running on
1244 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1245 * and blocked sched_entities.
1247 * util_avg may also factor frequency scaling and CPU capacity scaling:
1249 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1251 * where freq% is the same as above, and capacity% is the CPU capacity
1252 * normalized to the greatest capacity (due to uarch differences, etc).
1254 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1255 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1256 * we therefore scale them to as large a range as necessary. This is for
1257 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1261 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1262 * with the highest load (=88761), always runnable on a single cfs_rq,
1263 * and should not overflow as the number already hits PID_MAX_LIMIT.
1265 * For all other cases (including 32-bit kernels), struct load_weight's
1266 * weight will overflow first before we do, because:
1268 * Max(load_avg) <= Max(load.weight)
1270 * Then it is the load_weight's responsibility to consider overflow
1274 u64 last_update_time
, load_sum
;
1275 u32 util_sum
, period_contrib
;
1276 unsigned long load_avg
, util_avg
;
1279 #ifdef CONFIG_SCHEDSTATS
1280 struct sched_statistics
{
1290 s64 sum_sleep_runtime
;
1297 u64 nr_migrations_cold
;
1298 u64 nr_failed_migrations_affine
;
1299 u64 nr_failed_migrations_running
;
1300 u64 nr_failed_migrations_hot
;
1301 u64 nr_forced_migrations
;
1304 u64 nr_wakeups_sync
;
1305 u64 nr_wakeups_migrate
;
1306 u64 nr_wakeups_local
;
1307 u64 nr_wakeups_remote
;
1308 u64 nr_wakeups_affine
;
1309 u64 nr_wakeups_affine_attempts
;
1310 u64 nr_wakeups_passive
;
1311 u64 nr_wakeups_idle
;
1315 struct sched_entity
{
1316 struct load_weight load
; /* for load-balancing */
1317 struct rb_node run_node
;
1318 struct list_head group_node
;
1322 u64 sum_exec_runtime
;
1324 u64 prev_sum_exec_runtime
;
1328 #ifdef CONFIG_SCHEDSTATS
1329 struct sched_statistics statistics
;
1332 #ifdef CONFIG_FAIR_GROUP_SCHED
1334 struct sched_entity
*parent
;
1335 /* rq on which this entity is (to be) queued: */
1336 struct cfs_rq
*cfs_rq
;
1337 /* rq "owned" by this entity/group: */
1338 struct cfs_rq
*my_q
;
1343 * Per entity load average tracking.
1345 * Put into separate cache line so it does not
1346 * collide with read-mostly values above.
1348 struct sched_avg avg ____cacheline_aligned_in_smp
;
1352 struct sched_rt_entity
{
1353 struct list_head run_list
;
1354 unsigned long timeout
;
1355 unsigned long watchdog_stamp
;
1356 unsigned int time_slice
;
1357 unsigned short on_rq
;
1358 unsigned short on_list
;
1360 struct sched_rt_entity
*back
;
1361 #ifdef CONFIG_RT_GROUP_SCHED
1362 struct sched_rt_entity
*parent
;
1363 /* rq on which this entity is (to be) queued: */
1364 struct rt_rq
*rt_rq
;
1365 /* rq "owned" by this entity/group: */
1370 struct sched_dl_entity
{
1371 struct rb_node rb_node
;
1374 * Original scheduling parameters. Copied here from sched_attr
1375 * during sched_setattr(), they will remain the same until
1376 * the next sched_setattr().
1378 u64 dl_runtime
; /* maximum runtime for each instance */
1379 u64 dl_deadline
; /* relative deadline of each instance */
1380 u64 dl_period
; /* separation of two instances (period) */
1381 u64 dl_bw
; /* dl_runtime / dl_deadline */
1384 * Actual scheduling parameters. Initialized with the values above,
1385 * they are continously updated during task execution. Note that
1386 * the remaining runtime could be < 0 in case we are in overrun.
1388 s64 runtime
; /* remaining runtime for this instance */
1389 u64 deadline
; /* absolute deadline for this instance */
1390 unsigned int flags
; /* specifying the scheduler behaviour */
1395 * @dl_throttled tells if we exhausted the runtime. If so, the
1396 * task has to wait for a replenishment to be performed at the
1397 * next firing of dl_timer.
1399 * @dl_boosted tells if we are boosted due to DI. If so we are
1400 * outside bandwidth enforcement mechanism (but only until we
1401 * exit the critical section);
1403 * @dl_yielded tells if task gave up the cpu before consuming
1404 * all its available runtime during the last job.
1406 int dl_throttled
, dl_boosted
, dl_yielded
;
1409 * Bandwidth enforcement timer. Each -deadline task has its
1410 * own bandwidth to be enforced, thus we need one timer per task.
1412 struct hrtimer dl_timer
;
1420 u8 pad
; /* Otherwise the compiler can store garbage here. */
1422 u32 s
; /* Set of bits. */
1426 enum perf_event_task_context
{
1427 perf_invalid_context
= -1,
1428 perf_hw_context
= 0,
1430 perf_nr_task_contexts
,
1433 /* Track pages that require TLB flushes */
1434 struct tlbflush_unmap_batch
{
1436 * Each bit set is a CPU that potentially has a TLB entry for one of
1437 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1439 struct cpumask cpumask
;
1441 /* True if any bit in cpumask is set */
1442 bool flush_required
;
1445 * If true then the PTE was dirty when unmapped. The entry must be
1446 * flushed before IO is initiated or a stale TLB entry potentially
1447 * allows an update without redirtying the page.
1452 struct task_struct
{
1453 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1456 unsigned int flags
; /* per process flags, defined below */
1457 unsigned int ptrace
;
1460 struct llist_node wake_entry
;
1462 unsigned int wakee_flips
;
1463 unsigned long wakee_flip_decay_ts
;
1464 struct task_struct
*last_wakee
;
1470 int prio
, static_prio
, normal_prio
;
1471 unsigned int rt_priority
;
1472 const struct sched_class
*sched_class
;
1473 struct sched_entity se
;
1474 struct sched_rt_entity rt
;
1475 #ifdef CONFIG_CGROUP_SCHED
1476 struct task_group
*sched_task_group
;
1478 struct sched_dl_entity dl
;
1480 #ifdef CONFIG_PREEMPT_NOTIFIERS
1481 /* list of struct preempt_notifier: */
1482 struct hlist_head preempt_notifiers
;
1485 #ifdef CONFIG_BLK_DEV_IO_TRACE
1486 unsigned int btrace_seq
;
1489 unsigned int policy
;
1490 int nr_cpus_allowed
;
1491 cpumask_t cpus_allowed
;
1493 #ifdef CONFIG_PREEMPT_RCU
1494 int rcu_read_lock_nesting
;
1495 union rcu_special rcu_read_unlock_special
;
1496 struct list_head rcu_node_entry
;
1497 struct rcu_node
*rcu_blocked_node
;
1498 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1499 #ifdef CONFIG_TASKS_RCU
1500 unsigned long rcu_tasks_nvcsw
;
1501 bool rcu_tasks_holdout
;
1502 struct list_head rcu_tasks_holdout_list
;
1503 int rcu_tasks_idle_cpu
;
1504 #endif /* #ifdef CONFIG_TASKS_RCU */
1506 #ifdef CONFIG_SCHED_INFO
1507 struct sched_info sched_info
;
1510 struct list_head tasks
;
1512 struct plist_node pushable_tasks
;
1513 struct rb_node pushable_dl_tasks
;
1516 struct mm_struct
*mm
, *active_mm
;
1517 /* per-thread vma caching */
1518 u32 vmacache_seqnum
;
1519 struct vm_area_struct
*vmacache
[VMACACHE_SIZE
];
1520 #if defined(SPLIT_RSS_COUNTING)
1521 struct task_rss_stat rss_stat
;
1525 int exit_code
, exit_signal
;
1526 int pdeath_signal
; /* The signal sent when the parent dies */
1527 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1529 /* Used for emulating ABI behavior of previous Linux versions */
1530 unsigned int personality
;
1532 /* scheduler bits, serialized by scheduler locks */
1533 unsigned sched_reset_on_fork
:1;
1534 unsigned sched_contributes_to_load
:1;
1535 unsigned sched_migrated
:1;
1536 unsigned :0; /* force alignment to the next boundary */
1538 /* unserialized, strictly 'current' */
1539 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1540 unsigned in_iowait
:1;
1542 unsigned memcg_may_oom
:1;
1544 unsigned memcg_kmem_skip_account
:1;
1547 #ifdef CONFIG_COMPAT_BRK
1548 unsigned brk_randomized
:1;
1551 unsigned long atomic_flags
; /* Flags needing atomic access. */
1553 struct restart_block restart_block
;
1558 #ifdef CONFIG_CC_STACKPROTECTOR
1559 /* Canary value for the -fstack-protector gcc feature */
1560 unsigned long stack_canary
;
1563 * pointers to (original) parent process, youngest child, younger sibling,
1564 * older sibling, respectively. (p->father can be replaced with
1565 * p->real_parent->pid)
1567 struct task_struct __rcu
*real_parent
; /* real parent process */
1568 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1570 * children/sibling forms the list of my natural children
1572 struct list_head children
; /* list of my children */
1573 struct list_head sibling
; /* linkage in my parent's children list */
1574 struct task_struct
*group_leader
; /* threadgroup leader */
1577 * ptraced is the list of tasks this task is using ptrace on.
1578 * This includes both natural children and PTRACE_ATTACH targets.
1579 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1581 struct list_head ptraced
;
1582 struct list_head ptrace_entry
;
1584 /* PID/PID hash table linkage. */
1585 struct pid_link pids
[PIDTYPE_MAX
];
1586 struct list_head thread_group
;
1587 struct list_head thread_node
;
1589 struct completion
*vfork_done
; /* for vfork() */
1590 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1591 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1593 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1595 struct prev_cputime prev_cputime
;
1596 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1597 seqcount_t vtime_seqcount
;
1598 unsigned long long vtime_snap
;
1600 /* Task is sleeping or running in a CPU with VTIME inactive */
1602 /* Task runs in userspace in a CPU with VTIME active */
1604 /* Task runs in kernelspace in a CPU with VTIME active */
1606 } vtime_snap_whence
;
1609 #ifdef CONFIG_NO_HZ_FULL
1610 atomic_t tick_dep_mask
;
1612 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1613 u64 start_time
; /* monotonic time in nsec */
1614 u64 real_start_time
; /* boot based time in nsec */
1615 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1616 unsigned long min_flt
, maj_flt
;
1618 struct task_cputime cputime_expires
;
1619 struct list_head cpu_timers
[3];
1621 /* process credentials */
1622 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1623 * credentials (COW) */
1624 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1625 * credentials (COW) */
1626 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1627 - access with [gs]et_task_comm (which lock
1628 it with task_lock())
1629 - initialized normally by setup_new_exec */
1630 /* file system info */
1631 struct nameidata
*nameidata
;
1632 #ifdef CONFIG_SYSVIPC
1634 struct sysv_sem sysvsem
;
1635 struct sysv_shm sysvshm
;
1637 #ifdef CONFIG_DETECT_HUNG_TASK
1638 /* hung task detection */
1639 unsigned long last_switch_count
;
1641 /* filesystem information */
1642 struct fs_struct
*fs
;
1643 /* open file information */
1644 struct files_struct
*files
;
1646 struct nsproxy
*nsproxy
;
1647 /* signal handlers */
1648 struct signal_struct
*signal
;
1649 struct sighand_struct
*sighand
;
1651 sigset_t blocked
, real_blocked
;
1652 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1653 struct sigpending pending
;
1655 unsigned long sas_ss_sp
;
1657 unsigned sas_ss_flags
;
1659 struct callback_head
*task_works
;
1661 struct audit_context
*audit_context
;
1662 #ifdef CONFIG_AUDITSYSCALL
1664 unsigned int sessionid
;
1666 struct seccomp seccomp
;
1668 /* Thread group tracking */
1671 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1673 spinlock_t alloc_lock
;
1675 /* Protection of the PI data structures: */
1676 raw_spinlock_t pi_lock
;
1678 struct wake_q_node wake_q
;
1680 #ifdef CONFIG_RT_MUTEXES
1681 /* PI waiters blocked on a rt_mutex held by this task */
1682 struct rb_root pi_waiters
;
1683 struct rb_node
*pi_waiters_leftmost
;
1684 /* Deadlock detection and priority inheritance handling */
1685 struct rt_mutex_waiter
*pi_blocked_on
;
1688 #ifdef CONFIG_DEBUG_MUTEXES
1689 /* mutex deadlock detection */
1690 struct mutex_waiter
*blocked_on
;
1692 #ifdef CONFIG_TRACE_IRQFLAGS
1693 unsigned int irq_events
;
1694 unsigned long hardirq_enable_ip
;
1695 unsigned long hardirq_disable_ip
;
1696 unsigned int hardirq_enable_event
;
1697 unsigned int hardirq_disable_event
;
1698 int hardirqs_enabled
;
1699 int hardirq_context
;
1700 unsigned long softirq_disable_ip
;
1701 unsigned long softirq_enable_ip
;
1702 unsigned int softirq_disable_event
;
1703 unsigned int softirq_enable_event
;
1704 int softirqs_enabled
;
1705 int softirq_context
;
1707 #ifdef CONFIG_LOCKDEP
1708 # define MAX_LOCK_DEPTH 48UL
1711 unsigned int lockdep_recursion
;
1712 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1713 gfp_t lockdep_reclaim_gfp
;
1716 unsigned int in_ubsan
;
1719 /* journalling filesystem info */
1722 /* stacked block device info */
1723 struct bio_list
*bio_list
;
1726 /* stack plugging */
1727 struct blk_plug
*plug
;
1731 struct reclaim_state
*reclaim_state
;
1733 struct backing_dev_info
*backing_dev_info
;
1735 struct io_context
*io_context
;
1737 unsigned long ptrace_message
;
1738 siginfo_t
*last_siginfo
; /* For ptrace use. */
1739 struct task_io_accounting ioac
;
1740 #if defined(CONFIG_TASK_XACCT)
1741 u64 acct_rss_mem1
; /* accumulated rss usage */
1742 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1743 cputime_t acct_timexpd
; /* stime + utime since last update */
1745 #ifdef CONFIG_CPUSETS
1746 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1747 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1748 int cpuset_mem_spread_rotor
;
1749 int cpuset_slab_spread_rotor
;
1751 #ifdef CONFIG_CGROUPS
1752 /* Control Group info protected by css_set_lock */
1753 struct css_set __rcu
*cgroups
;
1754 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1755 struct list_head cg_list
;
1758 struct robust_list_head __user
*robust_list
;
1759 #ifdef CONFIG_COMPAT
1760 struct compat_robust_list_head __user
*compat_robust_list
;
1762 struct list_head pi_state_list
;
1763 struct futex_pi_state
*pi_state_cache
;
1765 #ifdef CONFIG_PERF_EVENTS
1766 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1767 struct mutex perf_event_mutex
;
1768 struct list_head perf_event_list
;
1770 #ifdef CONFIG_DEBUG_PREEMPT
1771 unsigned long preempt_disable_ip
;
1774 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1776 short pref_node_fork
;
1778 #ifdef CONFIG_NUMA_BALANCING
1780 unsigned int numa_scan_period
;
1781 unsigned int numa_scan_period_max
;
1782 int numa_preferred_nid
;
1783 unsigned long numa_migrate_retry
;
1784 u64 node_stamp
; /* migration stamp */
1785 u64 last_task_numa_placement
;
1786 u64 last_sum_exec_runtime
;
1787 struct callback_head numa_work
;
1789 struct list_head numa_entry
;
1790 struct numa_group
*numa_group
;
1793 * numa_faults is an array split into four regions:
1794 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1795 * in this precise order.
1797 * faults_memory: Exponential decaying average of faults on a per-node
1798 * basis. Scheduling placement decisions are made based on these
1799 * counts. The values remain static for the duration of a PTE scan.
1800 * faults_cpu: Track the nodes the process was running on when a NUMA
1801 * hinting fault was incurred.
1802 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1803 * during the current scan window. When the scan completes, the counts
1804 * in faults_memory and faults_cpu decay and these values are copied.
1806 unsigned long *numa_faults
;
1807 unsigned long total_numa_faults
;
1810 * numa_faults_locality tracks if faults recorded during the last
1811 * scan window were remote/local or failed to migrate. The task scan
1812 * period is adapted based on the locality of the faults with different
1813 * weights depending on whether they were shared or private faults
1815 unsigned long numa_faults_locality
[3];
1817 unsigned long numa_pages_migrated
;
1818 #endif /* CONFIG_NUMA_BALANCING */
1820 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1821 struct tlbflush_unmap_batch tlb_ubc
;
1824 struct rcu_head rcu
;
1827 * cache last used pipe for splice
1829 struct pipe_inode_info
*splice_pipe
;
1831 struct page_frag task_frag
;
1833 #ifdef CONFIG_TASK_DELAY_ACCT
1834 struct task_delay_info
*delays
;
1836 #ifdef CONFIG_FAULT_INJECTION
1840 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1841 * balance_dirty_pages() for some dirty throttling pause
1844 int nr_dirtied_pause
;
1845 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1847 #ifdef CONFIG_LATENCYTOP
1848 int latency_record_count
;
1849 struct latency_record latency_record
[LT_SAVECOUNT
];
1852 * time slack values; these are used to round up poll() and
1853 * select() etc timeout values. These are in nanoseconds.
1856 u64 default_timer_slack_ns
;
1859 unsigned int kasan_depth
;
1861 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1862 /* Index of current stored address in ret_stack */
1864 /* Stack of return addresses for return function tracing */
1865 struct ftrace_ret_stack
*ret_stack
;
1866 /* time stamp for last schedule */
1867 unsigned long long ftrace_timestamp
;
1869 * Number of functions that haven't been traced
1870 * because of depth overrun.
1872 atomic_t trace_overrun
;
1873 /* Pause for the tracing */
1874 atomic_t tracing_graph_pause
;
1876 #ifdef CONFIG_TRACING
1877 /* state flags for use by tracers */
1878 unsigned long trace
;
1879 /* bitmask and counter of trace recursion */
1880 unsigned long trace_recursion
;
1881 #endif /* CONFIG_TRACING */
1883 /* Coverage collection mode enabled for this task (0 if disabled). */
1884 enum kcov_mode kcov_mode
;
1885 /* Size of the kcov_area. */
1887 /* Buffer for coverage collection. */
1889 /* kcov desciptor wired with this task or NULL. */
1893 struct mem_cgroup
*memcg_in_oom
;
1894 gfp_t memcg_oom_gfp_mask
;
1895 int memcg_oom_order
;
1897 /* number of pages to reclaim on returning to userland */
1898 unsigned int memcg_nr_pages_over_high
;
1900 #ifdef CONFIG_UPROBES
1901 struct uprobe_task
*utask
;
1903 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1904 unsigned int sequential_io
;
1905 unsigned int sequential_io_avg
;
1907 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1908 unsigned long task_state_change
;
1910 int pagefault_disabled
;
1912 struct task_struct
*oom_reaper_list
;
1914 /* CPU-specific state of this task */
1915 struct thread_struct thread
;
1917 * WARNING: on x86, 'thread_struct' contains a variable-sized
1918 * structure. It *MUST* be at the end of 'task_struct'.
1920 * Do not put anything below here!
1924 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1925 extern int arch_task_struct_size __read_mostly
;
1927 # define arch_task_struct_size (sizeof(struct task_struct))
1930 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1931 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1933 static inline int tsk_nr_cpus_allowed(struct task_struct
*p
)
1935 return p
->nr_cpus_allowed
;
1938 #define TNF_MIGRATED 0x01
1939 #define TNF_NO_GROUP 0x02
1940 #define TNF_SHARED 0x04
1941 #define TNF_FAULT_LOCAL 0x08
1942 #define TNF_MIGRATE_FAIL 0x10
1944 #ifdef CONFIG_NUMA_BALANCING
1945 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
1946 extern pid_t
task_numa_group_id(struct task_struct
*p
);
1947 extern void set_numabalancing_state(bool enabled
);
1948 extern void task_numa_free(struct task_struct
*p
);
1949 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
1950 int src_nid
, int dst_cpu
);
1952 static inline void task_numa_fault(int last_node
, int node
, int pages
,
1956 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
1960 static inline void set_numabalancing_state(bool enabled
)
1963 static inline void task_numa_free(struct task_struct
*p
)
1966 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
1967 struct page
*page
, int src_nid
, int dst_cpu
)
1973 static inline struct pid
*task_pid(struct task_struct
*task
)
1975 return task
->pids
[PIDTYPE_PID
].pid
;
1978 static inline struct pid
*task_tgid(struct task_struct
*task
)
1980 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1984 * Without tasklist or rcu lock it is not safe to dereference
1985 * the result of task_pgrp/task_session even if task == current,
1986 * we can race with another thread doing sys_setsid/sys_setpgid.
1988 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1990 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1993 static inline struct pid
*task_session(struct task_struct
*task
)
1995 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1998 struct pid_namespace
;
2001 * the helpers to get the task's different pids as they are seen
2002 * from various namespaces
2004 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
2005 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2007 * task_xid_nr_ns() : id seen from the ns specified;
2009 * set_task_vxid() : assigns a virtual id to a task;
2011 * see also pid_nr() etc in include/linux/pid.h
2013 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
2014 struct pid_namespace
*ns
);
2016 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
2021 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
2022 struct pid_namespace
*ns
)
2024 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
2027 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
2029 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
2033 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
2038 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
2040 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
2042 return pid_vnr(task_tgid(tsk
));
2046 static inline int pid_alive(const struct task_struct
*p
);
2047 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
2053 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
2059 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
2061 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
2064 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
2065 struct pid_namespace
*ns
)
2067 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
2070 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
2072 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
2076 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
2077 struct pid_namespace
*ns
)
2079 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
2082 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
2084 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
2087 /* obsolete, do not use */
2088 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
2090 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
2094 * pid_alive - check that a task structure is not stale
2095 * @p: Task structure to be checked.
2097 * Test if a process is not yet dead (at most zombie state)
2098 * If pid_alive fails, then pointers within the task structure
2099 * can be stale and must not be dereferenced.
2101 * Return: 1 if the process is alive. 0 otherwise.
2103 static inline int pid_alive(const struct task_struct
*p
)
2105 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
2109 * is_global_init - check if a task structure is init. Since init
2110 * is free to have sub-threads we need to check tgid.
2111 * @tsk: Task structure to be checked.
2113 * Check if a task structure is the first user space task the kernel created.
2115 * Return: 1 if the task structure is init. 0 otherwise.
2117 static inline int is_global_init(struct task_struct
*tsk
)
2119 return task_tgid_nr(tsk
) == 1;
2122 extern struct pid
*cad_pid
;
2124 extern void free_task(struct task_struct
*tsk
);
2125 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2127 extern void __put_task_struct(struct task_struct
*t
);
2129 static inline void put_task_struct(struct task_struct
*t
)
2131 if (atomic_dec_and_test(&t
->usage
))
2132 __put_task_struct(t
);
2135 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2136 extern void task_cputime(struct task_struct
*t
,
2137 cputime_t
*utime
, cputime_t
*stime
);
2138 extern void task_cputime_scaled(struct task_struct
*t
,
2139 cputime_t
*utimescaled
, cputime_t
*stimescaled
);
2140 extern cputime_t
task_gtime(struct task_struct
*t
);
2142 static inline void task_cputime(struct task_struct
*t
,
2143 cputime_t
*utime
, cputime_t
*stime
)
2151 static inline void task_cputime_scaled(struct task_struct
*t
,
2152 cputime_t
*utimescaled
,
2153 cputime_t
*stimescaled
)
2156 *utimescaled
= t
->utimescaled
;
2158 *stimescaled
= t
->stimescaled
;
2161 static inline cputime_t
task_gtime(struct task_struct
*t
)
2166 extern void task_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2167 extern void thread_group_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2172 #define PF_EXITING 0x00000004 /* getting shut down */
2173 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2174 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2175 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2176 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2177 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2178 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2179 #define PF_DUMPCORE 0x00000200 /* dumped core */
2180 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2181 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2182 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2183 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2184 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2185 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2186 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2187 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2188 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2189 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2190 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2191 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2192 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2193 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2194 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2195 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2196 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2197 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2198 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2201 * Only the _current_ task can read/write to tsk->flags, but other
2202 * tasks can access tsk->flags in readonly mode for example
2203 * with tsk_used_math (like during threaded core dumping).
2204 * There is however an exception to this rule during ptrace
2205 * or during fork: the ptracer task is allowed to write to the
2206 * child->flags of its traced child (same goes for fork, the parent
2207 * can write to the child->flags), because we're guaranteed the
2208 * child is not running and in turn not changing child->flags
2209 * at the same time the parent does it.
2211 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2212 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2213 #define clear_used_math() clear_stopped_child_used_math(current)
2214 #define set_used_math() set_stopped_child_used_math(current)
2215 #define conditional_stopped_child_used_math(condition, child) \
2216 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2217 #define conditional_used_math(condition) \
2218 conditional_stopped_child_used_math(condition, current)
2219 #define copy_to_stopped_child_used_math(child) \
2220 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2221 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2222 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2223 #define used_math() tsk_used_math(current)
2225 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2226 * __GFP_FS is also cleared as it implies __GFP_IO.
2228 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
2230 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
2231 flags
&= ~(__GFP_IO
| __GFP_FS
);
2235 static inline unsigned int memalloc_noio_save(void)
2237 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
2238 current
->flags
|= PF_MEMALLOC_NOIO
;
2242 static inline void memalloc_noio_restore(unsigned int flags
)
2244 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
2247 /* Per-process atomic flags. */
2248 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2249 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2250 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2253 #define TASK_PFA_TEST(name, func) \
2254 static inline bool task_##func(struct task_struct *p) \
2255 { return test_bit(PFA_##name, &p->atomic_flags); }
2256 #define TASK_PFA_SET(name, func) \
2257 static inline void task_set_##func(struct task_struct *p) \
2258 { set_bit(PFA_##name, &p->atomic_flags); }
2259 #define TASK_PFA_CLEAR(name, func) \
2260 static inline void task_clear_##func(struct task_struct *p) \
2261 { clear_bit(PFA_##name, &p->atomic_flags); }
2263 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
2264 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
2266 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
2267 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
2268 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2270 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2271 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2272 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2275 * task->jobctl flags
2277 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2279 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2280 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2281 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2282 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2283 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2284 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2285 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2287 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2288 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2289 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2290 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2291 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2292 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2293 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2295 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2296 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2298 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2299 unsigned long mask
);
2300 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2301 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2302 unsigned long mask
);
2304 static inline void rcu_copy_process(struct task_struct
*p
)
2306 #ifdef CONFIG_PREEMPT_RCU
2307 p
->rcu_read_lock_nesting
= 0;
2308 p
->rcu_read_unlock_special
.s
= 0;
2309 p
->rcu_blocked_node
= NULL
;
2310 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2311 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2312 #ifdef CONFIG_TASKS_RCU
2313 p
->rcu_tasks_holdout
= false;
2314 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2315 p
->rcu_tasks_idle_cpu
= -1;
2316 #endif /* #ifdef CONFIG_TASKS_RCU */
2319 static inline void tsk_restore_flags(struct task_struct
*task
,
2320 unsigned long orig_flags
, unsigned long flags
)
2322 task
->flags
&= ~flags
;
2323 task
->flags
|= orig_flags
& flags
;
2326 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2327 const struct cpumask
*trial
);
2328 extern int task_can_attach(struct task_struct
*p
,
2329 const struct cpumask
*cs_cpus_allowed
);
2331 extern void do_set_cpus_allowed(struct task_struct
*p
,
2332 const struct cpumask
*new_mask
);
2334 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2335 const struct cpumask
*new_mask
);
2337 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2338 const struct cpumask
*new_mask
)
2341 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2342 const struct cpumask
*new_mask
)
2344 if (!cpumask_test_cpu(0, new_mask
))
2350 #ifdef CONFIG_NO_HZ_COMMON
2351 void calc_load_enter_idle(void);
2352 void calc_load_exit_idle(void);
2354 static inline void calc_load_enter_idle(void) { }
2355 static inline void calc_load_exit_idle(void) { }
2356 #endif /* CONFIG_NO_HZ_COMMON */
2359 * Do not use outside of architecture code which knows its limitations.
2361 * sched_clock() has no promise of monotonicity or bounded drift between
2362 * CPUs, use (which you should not) requires disabling IRQs.
2364 * Please use one of the three interfaces below.
2366 extern unsigned long long notrace
sched_clock(void);
2368 * See the comment in kernel/sched/clock.c
2370 extern u64
running_clock(void);
2371 extern u64
sched_clock_cpu(int cpu
);
2374 extern void sched_clock_init(void);
2376 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2377 static inline void sched_clock_tick(void)
2381 static inline void sched_clock_idle_sleep_event(void)
2385 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
2389 static inline u64
cpu_clock(int cpu
)
2391 return sched_clock();
2394 static inline u64
local_clock(void)
2396 return sched_clock();
2400 * Architectures can set this to 1 if they have specified
2401 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2402 * but then during bootup it turns out that sched_clock()
2403 * is reliable after all:
2405 extern int sched_clock_stable(void);
2406 extern void set_sched_clock_stable(void);
2407 extern void clear_sched_clock_stable(void);
2409 extern void sched_clock_tick(void);
2410 extern void sched_clock_idle_sleep_event(void);
2411 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2414 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2415 * time source that is monotonic per cpu argument and has bounded drift
2418 * ######################### BIG FAT WARNING ##########################
2419 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2420 * # go backwards !! #
2421 * ####################################################################
2423 static inline u64
cpu_clock(int cpu
)
2425 return sched_clock_cpu(cpu
);
2428 static inline u64
local_clock(void)
2430 return sched_clock_cpu(raw_smp_processor_id());
2434 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2436 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2437 * The reason for this explicit opt-in is not to have perf penalty with
2438 * slow sched_clocks.
2440 extern void enable_sched_clock_irqtime(void);
2441 extern void disable_sched_clock_irqtime(void);
2443 static inline void enable_sched_clock_irqtime(void) {}
2444 static inline void disable_sched_clock_irqtime(void) {}
2447 extern unsigned long long
2448 task_sched_runtime(struct task_struct
*task
);
2450 /* sched_exec is called by processes performing an exec */
2452 extern void sched_exec(void);
2454 #define sched_exec() {}
2457 extern void sched_clock_idle_sleep_event(void);
2458 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2460 #ifdef CONFIG_HOTPLUG_CPU
2461 extern void idle_task_exit(void);
2463 static inline void idle_task_exit(void) {}
2466 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2467 extern void wake_up_nohz_cpu(int cpu
);
2469 static inline void wake_up_nohz_cpu(int cpu
) { }
2472 #ifdef CONFIG_NO_HZ_FULL
2473 extern u64
scheduler_tick_max_deferment(void);
2476 #ifdef CONFIG_SCHED_AUTOGROUP
2477 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2478 extern void sched_autogroup_detach(struct task_struct
*p
);
2479 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2480 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2481 #ifdef CONFIG_PROC_FS
2482 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2483 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2486 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2487 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2488 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2489 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2492 extern int yield_to(struct task_struct
*p
, bool preempt
);
2493 extern void set_user_nice(struct task_struct
*p
, long nice
);
2494 extern int task_prio(const struct task_struct
*p
);
2496 * task_nice - return the nice value of a given task.
2497 * @p: the task in question.
2499 * Return: The nice value [ -20 ... 0 ... 19 ].
2501 static inline int task_nice(const struct task_struct
*p
)
2503 return PRIO_TO_NICE((p
)->static_prio
);
2505 extern int can_nice(const struct task_struct
*p
, const int nice
);
2506 extern int task_curr(const struct task_struct
*p
);
2507 extern int idle_cpu(int cpu
);
2508 extern int sched_setscheduler(struct task_struct
*, int,
2509 const struct sched_param
*);
2510 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2511 const struct sched_param
*);
2512 extern int sched_setattr(struct task_struct
*,
2513 const struct sched_attr
*);
2514 extern struct task_struct
*idle_task(int cpu
);
2516 * is_idle_task - is the specified task an idle task?
2517 * @p: the task in question.
2519 * Return: 1 if @p is an idle task. 0 otherwise.
2521 static inline bool is_idle_task(const struct task_struct
*p
)
2525 extern struct task_struct
*curr_task(int cpu
);
2526 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2530 union thread_union
{
2531 struct thread_info thread_info
;
2532 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2535 #ifndef __HAVE_ARCH_KSTACK_END
2536 static inline int kstack_end(void *addr
)
2538 /* Reliable end of stack detection:
2539 * Some APM bios versions misalign the stack
2541 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2545 extern union thread_union init_thread_union
;
2546 extern struct task_struct init_task
;
2548 extern struct mm_struct init_mm
;
2550 extern struct pid_namespace init_pid_ns
;
2553 * find a task by one of its numerical ids
2555 * find_task_by_pid_ns():
2556 * finds a task by its pid in the specified namespace
2557 * find_task_by_vpid():
2558 * finds a task by its virtual pid
2560 * see also find_vpid() etc in include/linux/pid.h
2563 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2564 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2565 struct pid_namespace
*ns
);
2567 /* per-UID process charging. */
2568 extern struct user_struct
* alloc_uid(kuid_t
);
2569 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2571 atomic_inc(&u
->__count
);
2574 extern void free_uid(struct user_struct
*);
2576 #include <asm/current.h>
2578 extern void xtime_update(unsigned long ticks
);
2580 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2581 extern int wake_up_process(struct task_struct
*tsk
);
2582 extern void wake_up_new_task(struct task_struct
*tsk
);
2584 extern void kick_process(struct task_struct
*tsk
);
2586 static inline void kick_process(struct task_struct
*tsk
) { }
2588 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2589 extern void sched_dead(struct task_struct
*p
);
2591 extern void proc_caches_init(void);
2592 extern void flush_signals(struct task_struct
*);
2593 extern void ignore_signals(struct task_struct
*);
2594 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2595 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2597 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2599 struct task_struct
*tsk
= current
;
2603 spin_lock_irq(&tsk
->sighand
->siglock
);
2604 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2605 spin_unlock_irq(&tsk
->sighand
->siglock
);
2610 static inline void kernel_signal_stop(void)
2612 spin_lock_irq(¤t
->sighand
->siglock
);
2613 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2614 __set_current_state(TASK_STOPPED
);
2615 spin_unlock_irq(¤t
->sighand
->siglock
);
2620 extern void release_task(struct task_struct
* p
);
2621 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2622 extern int force_sigsegv(int, struct task_struct
*);
2623 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2624 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2625 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2626 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2627 const struct cred
*, u32
);
2628 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2629 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2630 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2631 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2632 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2633 extern void force_sig(int, struct task_struct
*);
2634 extern int send_sig(int, struct task_struct
*, int);
2635 extern int zap_other_threads(struct task_struct
*p
);
2636 extern struct sigqueue
*sigqueue_alloc(void);
2637 extern void sigqueue_free(struct sigqueue
*);
2638 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2639 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2641 static inline void restore_saved_sigmask(void)
2643 if (test_and_clear_restore_sigmask())
2644 __set_current_blocked(¤t
->saved_sigmask
);
2647 static inline sigset_t
*sigmask_to_save(void)
2649 sigset_t
*res
= ¤t
->blocked
;
2650 if (unlikely(test_restore_sigmask()))
2651 res
= ¤t
->saved_sigmask
;
2655 static inline int kill_cad_pid(int sig
, int priv
)
2657 return kill_pid(cad_pid
, sig
, priv
);
2660 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2661 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2662 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2663 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2666 * True if we are on the alternate signal stack.
2668 static inline int on_sig_stack(unsigned long sp
)
2671 * If the signal stack is SS_AUTODISARM then, by construction, we
2672 * can't be on the signal stack unless user code deliberately set
2673 * SS_AUTODISARM when we were already on it.
2675 * This improves reliability: if user state gets corrupted such that
2676 * the stack pointer points very close to the end of the signal stack,
2677 * then this check will enable the signal to be handled anyway.
2679 if (current
->sas_ss_flags
& SS_AUTODISARM
)
2682 #ifdef CONFIG_STACK_GROWSUP
2683 return sp
>= current
->sas_ss_sp
&&
2684 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2686 return sp
> current
->sas_ss_sp
&&
2687 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2691 static inline int sas_ss_flags(unsigned long sp
)
2693 if (!current
->sas_ss_size
)
2696 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2699 static inline void sas_ss_reset(struct task_struct
*p
)
2703 p
->sas_ss_flags
= SS_DISABLE
;
2706 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2708 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2709 #ifdef CONFIG_STACK_GROWSUP
2710 return current
->sas_ss_sp
;
2712 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2718 * Routines for handling mm_structs
2720 extern struct mm_struct
* mm_alloc(void);
2722 /* mmdrop drops the mm and the page tables */
2723 extern void __mmdrop(struct mm_struct
*);
2724 static inline void mmdrop(struct mm_struct
* mm
)
2726 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2730 /* mmput gets rid of the mappings and all user-space */
2731 extern void mmput(struct mm_struct
*);
2732 /* Grab a reference to a task's mm, if it is not already going away */
2733 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2735 * Grab a reference to a task's mm, if it is not already going away
2736 * and ptrace_may_access with the mode parameter passed to it
2739 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2740 /* Remove the current tasks stale references to the old mm_struct */
2741 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2743 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2744 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2745 struct task_struct
*, unsigned long);
2747 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2748 struct task_struct
*);
2750 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2751 * via pt_regs, so ignore the tls argument passed via C. */
2752 static inline int copy_thread_tls(
2753 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2754 struct task_struct
*p
, unsigned long tls
)
2756 return copy_thread(clone_flags
, sp
, arg
, p
);
2759 extern void flush_thread(void);
2760 extern void exit_thread(void);
2762 extern void exit_files(struct task_struct
*);
2763 extern void __cleanup_sighand(struct sighand_struct
*);
2765 extern void exit_itimers(struct signal_struct
*);
2766 extern void flush_itimer_signals(void);
2768 extern void do_group_exit(int);
2770 extern int do_execve(struct filename
*,
2771 const char __user
* const __user
*,
2772 const char __user
* const __user
*);
2773 extern int do_execveat(int, struct filename
*,
2774 const char __user
* const __user
*,
2775 const char __user
* const __user
*,
2777 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
2778 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2779 struct task_struct
*fork_idle(int);
2780 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2782 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
2783 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
2785 __set_task_comm(tsk
, from
, false);
2787 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2790 void scheduler_ipi(void);
2791 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2793 static inline void scheduler_ipi(void) { }
2794 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2801 #define tasklist_empty() \
2802 list_empty(&init_task.tasks)
2804 #define next_task(p) \
2805 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2807 #define for_each_process(p) \
2808 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2810 extern bool current_is_single_threaded(void);
2813 * Careful: do_each_thread/while_each_thread is a double loop so
2814 * 'break' will not work as expected - use goto instead.
2816 #define do_each_thread(g, t) \
2817 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2819 #define while_each_thread(g, t) \
2820 while ((t = next_thread(t)) != g)
2822 #define __for_each_thread(signal, t) \
2823 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2825 #define for_each_thread(p, t) \
2826 __for_each_thread((p)->signal, t)
2828 /* Careful: this is a double loop, 'break' won't work as expected. */
2829 #define for_each_process_thread(p, t) \
2830 for_each_process(p) for_each_thread(p, t)
2832 static inline int get_nr_threads(struct task_struct
*tsk
)
2834 return tsk
->signal
->nr_threads
;
2837 static inline bool thread_group_leader(struct task_struct
*p
)
2839 return p
->exit_signal
>= 0;
2842 /* Do to the insanities of de_thread it is possible for a process
2843 * to have the pid of the thread group leader without actually being
2844 * the thread group leader. For iteration through the pids in proc
2845 * all we care about is that we have a task with the appropriate
2846 * pid, we don't actually care if we have the right task.
2848 static inline bool has_group_leader_pid(struct task_struct
*p
)
2850 return task_pid(p
) == p
->signal
->leader_pid
;
2854 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2856 return p1
->signal
== p2
->signal
;
2859 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2861 return list_entry_rcu(p
->thread_group
.next
,
2862 struct task_struct
, thread_group
);
2865 static inline int thread_group_empty(struct task_struct
*p
)
2867 return list_empty(&p
->thread_group
);
2870 #define delay_group_leader(p) \
2871 (thread_group_leader(p) && !thread_group_empty(p))
2874 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2875 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2876 * pins the final release of task.io_context. Also protects ->cpuset and
2877 * ->cgroup.subsys[]. And ->vfork_done.
2879 * Nests both inside and outside of read_lock(&tasklist_lock).
2880 * It must not be nested with write_lock_irq(&tasklist_lock),
2881 * neither inside nor outside.
2883 static inline void task_lock(struct task_struct
*p
)
2885 spin_lock(&p
->alloc_lock
);
2888 static inline void task_unlock(struct task_struct
*p
)
2890 spin_unlock(&p
->alloc_lock
);
2893 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2894 unsigned long *flags
);
2896 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2897 unsigned long *flags
)
2899 struct sighand_struct
*ret
;
2901 ret
= __lock_task_sighand(tsk
, flags
);
2902 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2906 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2907 unsigned long *flags
)
2909 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2913 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2914 * @tsk: task causing the changes
2916 * All operations which modify a threadgroup - a new thread joining the
2917 * group, death of a member thread (the assertion of PF_EXITING) and
2918 * exec(2) dethreading the process and replacing the leader - are wrapped
2919 * by threadgroup_change_{begin|end}(). This is to provide a place which
2920 * subsystems needing threadgroup stability can hook into for
2923 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
2926 cgroup_threadgroup_change_begin(tsk
);
2930 * threadgroup_change_end - mark the end of changes to a threadgroup
2931 * @tsk: task causing the changes
2933 * See threadgroup_change_begin().
2935 static inline void threadgroup_change_end(struct task_struct
*tsk
)
2937 cgroup_threadgroup_change_end(tsk
);
2940 #ifndef __HAVE_THREAD_FUNCTIONS
2942 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2943 #define task_stack_page(task) ((task)->stack)
2945 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2947 *task_thread_info(p
) = *task_thread_info(org
);
2948 task_thread_info(p
)->task
= p
;
2952 * Return the address of the last usable long on the stack.
2954 * When the stack grows down, this is just above the thread
2955 * info struct. Going any lower will corrupt the threadinfo.
2957 * When the stack grows up, this is the highest address.
2958 * Beyond that position, we corrupt data on the next page.
2960 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2962 #ifdef CONFIG_STACK_GROWSUP
2963 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
2965 return (unsigned long *)(task_thread_info(p
) + 1);
2970 #define task_stack_end_corrupted(task) \
2971 (*(end_of_stack(task)) != STACK_END_MAGIC)
2973 static inline int object_is_on_stack(void *obj
)
2975 void *stack
= task_stack_page(current
);
2977 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2980 extern void thread_info_cache_init(void);
2982 #ifdef CONFIG_DEBUG_STACK_USAGE
2983 static inline unsigned long stack_not_used(struct task_struct
*p
)
2985 unsigned long *n
= end_of_stack(p
);
2987 do { /* Skip over canary */
2988 # ifdef CONFIG_STACK_GROWSUP
2995 # ifdef CONFIG_STACK_GROWSUP
2996 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
2998 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
3002 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
3004 /* set thread flags in other task's structures
3005 * - see asm/thread_info.h for TIF_xxxx flags available
3007 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3009 set_ti_thread_flag(task_thread_info(tsk
), flag
);
3012 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3014 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3017 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3019 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
3022 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3024 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3027 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3029 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
3032 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
3034 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3037 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
3039 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3042 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
3044 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
3047 static inline int restart_syscall(void)
3049 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
3050 return -ERESTARTNOINTR
;
3053 static inline int signal_pending(struct task_struct
*p
)
3055 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
3058 static inline int __fatal_signal_pending(struct task_struct
*p
)
3060 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
3063 static inline int fatal_signal_pending(struct task_struct
*p
)
3065 return signal_pending(p
) && __fatal_signal_pending(p
);
3068 static inline int signal_pending_state(long state
, struct task_struct
*p
)
3070 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
3072 if (!signal_pending(p
))
3075 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
3079 * cond_resched() and cond_resched_lock(): latency reduction via
3080 * explicit rescheduling in places that are safe. The return
3081 * value indicates whether a reschedule was done in fact.
3082 * cond_resched_lock() will drop the spinlock before scheduling,
3083 * cond_resched_softirq() will enable bhs before scheduling.
3085 extern int _cond_resched(void);
3087 #define cond_resched() ({ \
3088 ___might_sleep(__FILE__, __LINE__, 0); \
3092 extern int __cond_resched_lock(spinlock_t
*lock
);
3094 #define cond_resched_lock(lock) ({ \
3095 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3096 __cond_resched_lock(lock); \
3099 extern int __cond_resched_softirq(void);
3101 #define cond_resched_softirq() ({ \
3102 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3103 __cond_resched_softirq(); \
3106 static inline void cond_resched_rcu(void)
3108 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3116 * Does a critical section need to be broken due to another
3117 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3118 * but a general need for low latency)
3120 static inline int spin_needbreak(spinlock_t
*lock
)
3122 #ifdef CONFIG_PREEMPT
3123 return spin_is_contended(lock
);
3130 * Idle thread specific functions to determine the need_resched
3133 #ifdef TIF_POLLING_NRFLAG
3134 static inline int tsk_is_polling(struct task_struct
*p
)
3136 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
3139 static inline void __current_set_polling(void)
3141 set_thread_flag(TIF_POLLING_NRFLAG
);
3144 static inline bool __must_check
current_set_polling_and_test(void)
3146 __current_set_polling();
3149 * Polling state must be visible before we test NEED_RESCHED,
3150 * paired by resched_curr()
3152 smp_mb__after_atomic();
3154 return unlikely(tif_need_resched());
3157 static inline void __current_clr_polling(void)
3159 clear_thread_flag(TIF_POLLING_NRFLAG
);
3162 static inline bool __must_check
current_clr_polling_and_test(void)
3164 __current_clr_polling();
3167 * Polling state must be visible before we test NEED_RESCHED,
3168 * paired by resched_curr()
3170 smp_mb__after_atomic();
3172 return unlikely(tif_need_resched());
3176 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
3177 static inline void __current_set_polling(void) { }
3178 static inline void __current_clr_polling(void) { }
3180 static inline bool __must_check
current_set_polling_and_test(void)
3182 return unlikely(tif_need_resched());
3184 static inline bool __must_check
current_clr_polling_and_test(void)
3186 return unlikely(tif_need_resched());
3190 static inline void current_clr_polling(void)
3192 __current_clr_polling();
3195 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3196 * Once the bit is cleared, we'll get IPIs with every new
3197 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3200 smp_mb(); /* paired with resched_curr() */
3202 preempt_fold_need_resched();
3205 static __always_inline
bool need_resched(void)
3207 return unlikely(tif_need_resched());
3211 * Thread group CPU time accounting.
3213 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
3214 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
3217 * Reevaluate whether the task has signals pending delivery.
3218 * Wake the task if so.
3219 * This is required every time the blocked sigset_t changes.
3220 * callers must hold sighand->siglock.
3222 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
3223 extern void recalc_sigpending(void);
3225 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
3227 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
3229 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
3231 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
3233 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
3237 * Wrappers for p->thread_info->cpu access. No-op on UP.
3241 static inline unsigned int task_cpu(const struct task_struct
*p
)
3243 return task_thread_info(p
)->cpu
;
3246 static inline int task_node(const struct task_struct
*p
)
3248 return cpu_to_node(task_cpu(p
));
3251 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
3255 static inline unsigned int task_cpu(const struct task_struct
*p
)
3260 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
3264 #endif /* CONFIG_SMP */
3266 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3267 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3269 #ifdef CONFIG_CGROUP_SCHED
3270 extern struct task_group root_task_group
;
3271 #endif /* CONFIG_CGROUP_SCHED */
3273 extern int task_can_switch_user(struct user_struct
*up
,
3274 struct task_struct
*tsk
);
3276 #ifdef CONFIG_TASK_XACCT
3277 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3279 tsk
->ioac
.rchar
+= amt
;
3282 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3284 tsk
->ioac
.wchar
+= amt
;
3287 static inline void inc_syscr(struct task_struct
*tsk
)
3292 static inline void inc_syscw(struct task_struct
*tsk
)
3297 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3301 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3305 static inline void inc_syscr(struct task_struct
*tsk
)
3309 static inline void inc_syscw(struct task_struct
*tsk
)
3314 #ifndef TASK_SIZE_OF
3315 #define TASK_SIZE_OF(tsk) TASK_SIZE
3319 extern void mm_update_next_owner(struct mm_struct
*mm
);
3321 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3324 #endif /* CONFIG_MEMCG */
3326 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3329 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3332 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3335 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3338 static inline unsigned long rlimit(unsigned int limit
)
3340 return task_rlimit(current
, limit
);
3343 static inline unsigned long rlimit_max(unsigned int limit
)
3345 return task_rlimit_max(current
, limit
);
3348 #ifdef CONFIG_CPU_FREQ
3349 struct update_util_data
{
3350 void (*func
)(struct update_util_data
*data
,
3351 u64 time
, unsigned long util
, unsigned long max
);
3354 void cpufreq_add_update_util_hook(int cpu
, struct update_util_data
*data
,
3355 void (*func
)(struct update_util_data
*data
, u64 time
,
3356 unsigned long util
, unsigned long max
));
3357 void cpufreq_remove_update_util_hook(int cpu
);
3358 #endif /* CONFIG_CPU_FREQ */