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_NEW 2048
223 #define TASK_STATE_MAX 4096
225 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
227 extern char ___assert_task_state
[1 - 2*!!(
228 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
230 /* Convenience macros for the sake of set_task_state */
231 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
232 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
233 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
235 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
237 /* Convenience macros for the sake of wake_up */
238 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
239 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
241 /* get_task_state() */
242 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
243 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
244 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
246 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
247 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
248 #define task_is_stopped_or_traced(task) \
249 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
250 #define task_contributes_to_load(task) \
251 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
252 (task->flags & PF_FROZEN) == 0 && \
253 (task->state & TASK_NOLOAD) == 0)
255 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
257 #define __set_task_state(tsk, state_value) \
259 (tsk)->task_state_change = _THIS_IP_; \
260 (tsk)->state = (state_value); \
262 #define set_task_state(tsk, state_value) \
264 (tsk)->task_state_change = _THIS_IP_; \
265 smp_store_mb((tsk)->state, (state_value)); \
269 * set_current_state() includes a barrier so that the write of current->state
270 * is correctly serialised wrt the caller's subsequent test of whether to
273 * set_current_state(TASK_UNINTERRUPTIBLE);
274 * if (do_i_need_to_sleep())
277 * If the caller does not need such serialisation then use __set_current_state()
279 #define __set_current_state(state_value) \
281 current->task_state_change = _THIS_IP_; \
282 current->state = (state_value); \
284 #define set_current_state(state_value) \
286 current->task_state_change = _THIS_IP_; \
287 smp_store_mb(current->state, (state_value)); \
292 #define __set_task_state(tsk, state_value) \
293 do { (tsk)->state = (state_value); } while (0)
294 #define set_task_state(tsk, state_value) \
295 smp_store_mb((tsk)->state, (state_value))
298 * set_current_state() includes a barrier so that the write of current->state
299 * is correctly serialised wrt the caller's subsequent test of whether to
302 * set_current_state(TASK_UNINTERRUPTIBLE);
303 * if (do_i_need_to_sleep())
306 * If the caller does not need such serialisation then use __set_current_state()
308 #define __set_current_state(state_value) \
309 do { current->state = (state_value); } while (0)
310 #define set_current_state(state_value) \
311 smp_store_mb(current->state, (state_value))
315 /* Task command name length */
316 #define TASK_COMM_LEN 16
318 #include <linux/spinlock.h>
321 * This serializes "schedule()" and also protects
322 * the run-queue from deletions/modifications (but
323 * _adding_ to the beginning of the run-queue has
326 extern rwlock_t tasklist_lock
;
327 extern spinlock_t mmlist_lock
;
331 #ifdef CONFIG_PROVE_RCU
332 extern int lockdep_tasklist_lock_is_held(void);
333 #endif /* #ifdef CONFIG_PROVE_RCU */
335 extern void sched_init(void);
336 extern void sched_init_smp(void);
337 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
338 extern void init_idle(struct task_struct
*idle
, int cpu
);
339 extern void init_idle_bootup_task(struct task_struct
*idle
);
341 extern cpumask_var_t cpu_isolated_map
;
343 extern int runqueue_is_locked(int cpu
);
345 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
346 extern void nohz_balance_enter_idle(int cpu
);
347 extern void set_cpu_sd_state_idle(void);
348 extern int get_nohz_timer_target(void);
350 static inline void nohz_balance_enter_idle(int cpu
) { }
351 static inline void set_cpu_sd_state_idle(void) { }
355 * Only dump TASK_* tasks. (0 for all tasks)
357 extern void show_state_filter(unsigned long state_filter
);
359 static inline void show_state(void)
361 show_state_filter(0);
364 extern void show_regs(struct pt_regs
*);
367 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
368 * task), SP is the stack pointer of the first frame that should be shown in the back
369 * trace (or NULL if the entire call-chain of the task should be shown).
371 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
373 extern void cpu_init (void);
374 extern void trap_init(void);
375 extern void update_process_times(int user
);
376 extern void scheduler_tick(void);
377 extern int sched_cpu_starting(unsigned int cpu
);
378 extern int sched_cpu_activate(unsigned int cpu
);
379 extern int sched_cpu_deactivate(unsigned int cpu
);
381 #ifdef CONFIG_HOTPLUG_CPU
382 extern int sched_cpu_dying(unsigned int cpu
);
384 # define sched_cpu_dying NULL
387 extern void sched_show_task(struct task_struct
*p
);
389 #ifdef CONFIG_LOCKUP_DETECTOR
390 extern void touch_softlockup_watchdog_sched(void);
391 extern void touch_softlockup_watchdog(void);
392 extern void touch_softlockup_watchdog_sync(void);
393 extern void touch_all_softlockup_watchdogs(void);
394 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
396 size_t *lenp
, loff_t
*ppos
);
397 extern unsigned int softlockup_panic
;
398 extern unsigned int hardlockup_panic
;
399 void lockup_detector_init(void);
401 static inline void touch_softlockup_watchdog_sched(void)
404 static inline void touch_softlockup_watchdog(void)
407 static inline void touch_softlockup_watchdog_sync(void)
410 static inline void touch_all_softlockup_watchdogs(void)
413 static inline void lockup_detector_init(void)
418 #ifdef CONFIG_DETECT_HUNG_TASK
419 void reset_hung_task_detector(void);
421 static inline void reset_hung_task_detector(void)
426 /* Attach to any functions which should be ignored in wchan output. */
427 #define __sched __attribute__((__section__(".sched.text")))
429 /* Linker adds these: start and end of __sched functions */
430 extern char __sched_text_start
[], __sched_text_end
[];
432 /* Is this address in the __sched functions? */
433 extern int in_sched_functions(unsigned long addr
);
435 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
436 extern signed long schedule_timeout(signed long timeout
);
437 extern signed long schedule_timeout_interruptible(signed long timeout
);
438 extern signed long schedule_timeout_killable(signed long timeout
);
439 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
440 extern signed long schedule_timeout_idle(signed long timeout
);
441 asmlinkage
void schedule(void);
442 extern void schedule_preempt_disabled(void);
444 extern long io_schedule_timeout(long timeout
);
446 static inline void io_schedule(void)
448 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT
);
452 struct user_namespace
;
455 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
457 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
458 unsigned long, unsigned long);
460 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
461 unsigned long len
, unsigned long pgoff
,
462 unsigned long flags
);
464 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
467 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
468 #define SUID_DUMP_USER 1 /* Dump as user of process */
469 #define SUID_DUMP_ROOT 2 /* Dump as root */
473 /* for SUID_DUMP_* above */
474 #define MMF_DUMPABLE_BITS 2
475 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
477 extern void set_dumpable(struct mm_struct
*mm
, int value
);
479 * This returns the actual value of the suid_dumpable flag. For things
480 * that are using this for checking for privilege transitions, it must
481 * test against SUID_DUMP_USER rather than treating it as a boolean
484 static inline int __get_dumpable(unsigned long mm_flags
)
486 return mm_flags
& MMF_DUMPABLE_MASK
;
489 static inline int get_dumpable(struct mm_struct
*mm
)
491 return __get_dumpable(mm
->flags
);
494 /* coredump filter bits */
495 #define MMF_DUMP_ANON_PRIVATE 2
496 #define MMF_DUMP_ANON_SHARED 3
497 #define MMF_DUMP_MAPPED_PRIVATE 4
498 #define MMF_DUMP_MAPPED_SHARED 5
499 #define MMF_DUMP_ELF_HEADERS 6
500 #define MMF_DUMP_HUGETLB_PRIVATE 7
501 #define MMF_DUMP_HUGETLB_SHARED 8
502 #define MMF_DUMP_DAX_PRIVATE 9
503 #define MMF_DUMP_DAX_SHARED 10
505 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
506 #define MMF_DUMP_FILTER_BITS 9
507 #define MMF_DUMP_FILTER_MASK \
508 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
509 #define MMF_DUMP_FILTER_DEFAULT \
510 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
511 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
513 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
514 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
516 # define MMF_DUMP_MASK_DEFAULT_ELF 0
518 /* leave room for more dump flags */
519 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
520 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
521 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
523 #define MMF_HAS_UPROBES 19 /* has uprobes */
524 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
525 #define MMF_OOM_REAPED 21 /* mm has been already reaped */
526 #define MMF_OOM_NOT_REAPABLE 22 /* mm couldn't be reaped */
528 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
530 struct sighand_struct
{
532 struct k_sigaction action
[_NSIG
];
534 wait_queue_head_t signalfd_wqh
;
537 struct pacct_struct
{
540 unsigned long ac_mem
;
541 cputime_t ac_utime
, ac_stime
;
542 unsigned long ac_minflt
, ac_majflt
;
553 * struct prev_cputime - snaphsot of system and user cputime
554 * @utime: time spent in user mode
555 * @stime: time spent in system mode
556 * @lock: protects the above two fields
558 * Stores previous user/system time values such that we can guarantee
561 struct prev_cputime
{
562 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
569 static inline void prev_cputime_init(struct prev_cputime
*prev
)
571 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
572 prev
->utime
= prev
->stime
= 0;
573 raw_spin_lock_init(&prev
->lock
);
578 * struct task_cputime - collected CPU time counts
579 * @utime: time spent in user mode, in &cputime_t units
580 * @stime: time spent in kernel mode, in &cputime_t units
581 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
583 * This structure groups together three kinds of CPU time that are tracked for
584 * threads and thread groups. Most things considering CPU time want to group
585 * these counts together and treat all three of them in parallel.
587 struct task_cputime
{
590 unsigned long long sum_exec_runtime
;
593 /* Alternate field names when used to cache expirations. */
594 #define virt_exp utime
595 #define prof_exp stime
596 #define sched_exp sum_exec_runtime
598 #define INIT_CPUTIME \
599 (struct task_cputime) { \
602 .sum_exec_runtime = 0, \
606 * This is the atomic variant of task_cputime, which can be used for
607 * storing and updating task_cputime statistics without locking.
609 struct task_cputime_atomic
{
612 atomic64_t sum_exec_runtime
;
615 #define INIT_CPUTIME_ATOMIC \
616 (struct task_cputime_atomic) { \
617 .utime = ATOMIC64_INIT(0), \
618 .stime = ATOMIC64_INIT(0), \
619 .sum_exec_runtime = ATOMIC64_INIT(0), \
622 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
625 * Disable preemption until the scheduler is running -- use an unconditional
626 * value so that it also works on !PREEMPT_COUNT kernels.
628 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
630 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
633 * Initial preempt_count value; reflects the preempt_count schedule invariant
634 * which states that during context switches:
636 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
638 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
639 * Note: See finish_task_switch().
641 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
644 * struct thread_group_cputimer - thread group interval timer counts
645 * @cputime_atomic: atomic thread group interval timers.
646 * @running: true when there are timers running and
647 * @cputime_atomic receives updates.
648 * @checking_timer: true when a thread in the group is in the
649 * process of checking for thread group timers.
651 * This structure contains the version of task_cputime, above, that is
652 * used for thread group CPU timer calculations.
654 struct thread_group_cputimer
{
655 struct task_cputime_atomic cputime_atomic
;
660 #include <linux/rwsem.h>
664 * NOTE! "signal_struct" does not have its own
665 * locking, because a shared signal_struct always
666 * implies a shared sighand_struct, so locking
667 * sighand_struct is always a proper superset of
668 * the locking of signal_struct.
670 struct signal_struct
{
674 atomic_t oom_victims
; /* # of TIF_MEDIE threads in this thread group */
675 struct list_head thread_head
;
677 wait_queue_head_t wait_chldexit
; /* for wait4() */
679 /* current thread group signal load-balancing target: */
680 struct task_struct
*curr_target
;
682 /* shared signal handling: */
683 struct sigpending shared_pending
;
685 /* thread group exit support */
688 * - notify group_exit_task when ->count is equal to notify_count
689 * - everyone except group_exit_task is stopped during signal delivery
690 * of fatal signals, group_exit_task processes the signal.
693 struct task_struct
*group_exit_task
;
695 /* thread group stop support, overloads group_exit_code too */
696 int group_stop_count
;
697 unsigned int flags
; /* see SIGNAL_* flags below */
700 * PR_SET_CHILD_SUBREAPER marks a process, like a service
701 * manager, to re-parent orphan (double-forking) child processes
702 * to this process instead of 'init'. The service manager is
703 * able to receive SIGCHLD signals and is able to investigate
704 * the process until it calls wait(). All children of this
705 * process will inherit a flag if they should look for a
706 * child_subreaper process at exit.
708 unsigned int is_child_subreaper
:1;
709 unsigned int has_child_subreaper
:1;
711 /* POSIX.1b Interval Timers */
713 struct list_head posix_timers
;
715 /* ITIMER_REAL timer for the process */
716 struct hrtimer real_timer
;
717 struct pid
*leader_pid
;
718 ktime_t it_real_incr
;
721 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
722 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
723 * values are defined to 0 and 1 respectively
725 struct cpu_itimer it
[2];
728 * Thread group totals for process CPU timers.
729 * See thread_group_cputimer(), et al, for details.
731 struct thread_group_cputimer cputimer
;
733 /* Earliest-expiration cache. */
734 struct task_cputime cputime_expires
;
736 #ifdef CONFIG_NO_HZ_FULL
737 atomic_t tick_dep_mask
;
740 struct list_head cpu_timers
[3];
742 struct pid
*tty_old_pgrp
;
744 /* boolean value for session group leader */
747 struct tty_struct
*tty
; /* NULL if no tty */
749 #ifdef CONFIG_SCHED_AUTOGROUP
750 struct autogroup
*autogroup
;
753 * Cumulative resource counters for dead threads in the group,
754 * and for reaped dead child processes forked by this group.
755 * Live threads maintain their own counters and add to these
756 * in __exit_signal, except for the group leader.
758 seqlock_t stats_lock
;
759 cputime_t utime
, stime
, cutime
, cstime
;
762 struct prev_cputime prev_cputime
;
763 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
764 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
765 unsigned long inblock
, oublock
, cinblock
, coublock
;
766 unsigned long maxrss
, cmaxrss
;
767 struct task_io_accounting ioac
;
770 * Cumulative ns of schedule CPU time fo dead threads in the
771 * group, not including a zombie group leader, (This only differs
772 * from jiffies_to_ns(utime + stime) if sched_clock uses something
773 * other than jiffies.)
775 unsigned long long sum_sched_runtime
;
778 * We don't bother to synchronize most readers of this at all,
779 * because there is no reader checking a limit that actually needs
780 * to get both rlim_cur and rlim_max atomically, and either one
781 * alone is a single word that can safely be read normally.
782 * getrlimit/setrlimit use task_lock(current->group_leader) to
783 * protect this instead of the siglock, because they really
784 * have no need to disable irqs.
786 struct rlimit rlim
[RLIM_NLIMITS
];
788 #ifdef CONFIG_BSD_PROCESS_ACCT
789 struct pacct_struct pacct
; /* per-process accounting information */
791 #ifdef CONFIG_TASKSTATS
792 struct taskstats
*stats
;
796 struct tty_audit_buf
*tty_audit_buf
;
800 * Thread is the potential origin of an oom condition; kill first on
803 bool oom_flag_origin
;
804 short oom_score_adj
; /* OOM kill score adjustment */
805 short oom_score_adj_min
; /* OOM kill score adjustment min value.
806 * Only settable by CAP_SYS_RESOURCE. */
808 struct mutex cred_guard_mutex
; /* guard against foreign influences on
809 * credential calculations
810 * (notably. ptrace) */
814 * Bits in flags field of signal_struct.
816 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
817 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
818 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
819 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
821 * Pending notifications to parent.
823 #define SIGNAL_CLD_STOPPED 0x00000010
824 #define SIGNAL_CLD_CONTINUED 0x00000020
825 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
827 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
829 /* If true, all threads except ->group_exit_task have pending SIGKILL */
830 static inline int signal_group_exit(const struct signal_struct
*sig
)
832 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
833 (sig
->group_exit_task
!= NULL
);
837 * Some day this will be a full-fledged user tracking system..
840 atomic_t __count
; /* reference count */
841 atomic_t processes
; /* How many processes does this user have? */
842 atomic_t sigpending
; /* How many pending signals does this user have? */
843 #ifdef CONFIG_INOTIFY_USER
844 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
845 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
847 #ifdef CONFIG_FANOTIFY
848 atomic_t fanotify_listeners
;
851 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
853 #ifdef CONFIG_POSIX_MQUEUE
854 /* protected by mq_lock */
855 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
857 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
858 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
859 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
862 struct key
*uid_keyring
; /* UID specific keyring */
863 struct key
*session_keyring
; /* UID's default session keyring */
866 /* Hash table maintenance information */
867 struct hlist_node uidhash_node
;
870 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
871 atomic_long_t locked_vm
;
875 extern int uids_sysfs_init(void);
877 extern struct user_struct
*find_user(kuid_t
);
879 extern struct user_struct root_user
;
880 #define INIT_USER (&root_user)
883 struct backing_dev_info
;
884 struct reclaim_state
;
886 #ifdef CONFIG_SCHED_INFO
888 /* cumulative counters */
889 unsigned long pcount
; /* # of times run on this cpu */
890 unsigned long long run_delay
; /* time spent waiting on a runqueue */
893 unsigned long long last_arrival
,/* when we last ran on a cpu */
894 last_queued
; /* when we were last queued to run */
896 #endif /* CONFIG_SCHED_INFO */
898 #ifdef CONFIG_TASK_DELAY_ACCT
899 struct task_delay_info
{
901 unsigned int flags
; /* Private per-task flags */
903 /* For each stat XXX, add following, aligned appropriately
905 * struct timespec XXX_start, XXX_end;
909 * Atomicity of updates to XXX_delay, XXX_count protected by
910 * single lock above (split into XXX_lock if contention is an issue).
914 * XXX_count is incremented on every XXX operation, the delay
915 * associated with the operation is added to XXX_delay.
916 * XXX_delay contains the accumulated delay time in nanoseconds.
918 u64 blkio_start
; /* Shared by blkio, swapin */
919 u64 blkio_delay
; /* wait for sync block io completion */
920 u64 swapin_delay
; /* wait for swapin block io completion */
921 u32 blkio_count
; /* total count of the number of sync block */
922 /* io operations performed */
923 u32 swapin_count
; /* total count of the number of swapin block */
924 /* io operations performed */
927 u64 freepages_delay
; /* wait for memory reclaim */
928 u32 freepages_count
; /* total count of memory reclaim */
930 #endif /* CONFIG_TASK_DELAY_ACCT */
932 static inline int sched_info_on(void)
934 #ifdef CONFIG_SCHEDSTATS
936 #elif defined(CONFIG_TASK_DELAY_ACCT)
937 extern int delayacct_on
;
944 #ifdef CONFIG_SCHEDSTATS
945 void force_schedstat_enabled(void);
956 * Integer metrics need fixed point arithmetic, e.g., sched/fair
957 * has a few: load, load_avg, util_avg, freq, and capacity.
959 * We define a basic fixed point arithmetic range, and then formalize
960 * all these metrics based on that basic range.
962 # define SCHED_FIXEDPOINT_SHIFT 10
963 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
966 * Increase resolution of cpu_capacity calculations
968 #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT
969 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
972 * Wake-queues are lists of tasks with a pending wakeup, whose
973 * callers have already marked the task as woken internally,
974 * and can thus carry on. A common use case is being able to
975 * do the wakeups once the corresponding user lock as been
978 * We hold reference to each task in the list across the wakeup,
979 * thus guaranteeing that the memory is still valid by the time
980 * the actual wakeups are performed in wake_up_q().
982 * One per task suffices, because there's never a need for a task to be
983 * in two wake queues simultaneously; it is forbidden to abandon a task
984 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
985 * already in a wake queue, the wakeup will happen soon and the second
986 * waker can just skip it.
988 * The WAKE_Q macro declares and initializes the list head.
989 * wake_up_q() does NOT reinitialize the list; it's expected to be
990 * called near the end of a function, where the fact that the queue is
991 * not used again will be easy to see by inspection.
993 * Note that this can cause spurious wakeups. schedule() callers
994 * must ensure the call is done inside a loop, confirming that the
995 * wakeup condition has in fact occurred.
998 struct wake_q_node
*next
;
1001 struct wake_q_head
{
1002 struct wake_q_node
*first
;
1003 struct wake_q_node
**lastp
;
1006 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1008 #define WAKE_Q(name) \
1009 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1011 extern void wake_q_add(struct wake_q_head
*head
,
1012 struct task_struct
*task
);
1013 extern void wake_up_q(struct wake_q_head
*head
);
1016 * sched-domains (multiprocessor balancing) declarations:
1019 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
1020 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
1021 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
1022 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
1023 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
1024 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
1025 #define SD_ASYM_CPUCAPACITY 0x0040 /* Groups have different max cpu capacities */
1026 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu capacity */
1027 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
1028 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
1029 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1030 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1031 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1032 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1033 #define SD_NUMA 0x4000 /* cross-node balancing */
1035 #ifdef CONFIG_SCHED_SMT
1036 static inline int cpu_smt_flags(void)
1038 return SD_SHARE_CPUCAPACITY
| SD_SHARE_PKG_RESOURCES
;
1042 #ifdef CONFIG_SCHED_MC
1043 static inline int cpu_core_flags(void)
1045 return SD_SHARE_PKG_RESOURCES
;
1050 static inline int cpu_numa_flags(void)
1056 struct sched_domain_attr
{
1057 int relax_domain_level
;
1060 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1061 .relax_domain_level = -1, \
1064 extern int sched_domain_level_max
;
1068 struct sched_domain
{
1069 /* These fields must be setup */
1070 struct sched_domain
*parent
; /* top domain must be null terminated */
1071 struct sched_domain
*child
; /* bottom domain must be null terminated */
1072 struct sched_group
*groups
; /* the balancing groups of the domain */
1073 unsigned long min_interval
; /* Minimum balance interval ms */
1074 unsigned long max_interval
; /* Maximum balance interval ms */
1075 unsigned int busy_factor
; /* less balancing by factor if busy */
1076 unsigned int imbalance_pct
; /* No balance until over watermark */
1077 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
1078 unsigned int busy_idx
;
1079 unsigned int idle_idx
;
1080 unsigned int newidle_idx
;
1081 unsigned int wake_idx
;
1082 unsigned int forkexec_idx
;
1083 unsigned int smt_gain
;
1085 int nohz_idle
; /* NOHZ IDLE status */
1086 int flags
; /* See SD_* */
1089 /* Runtime fields. */
1090 unsigned long last_balance
; /* init to jiffies. units in jiffies */
1091 unsigned int balance_interval
; /* initialise to 1. units in ms. */
1092 unsigned int nr_balance_failed
; /* initialise to 0 */
1094 /* idle_balance() stats */
1095 u64 max_newidle_lb_cost
;
1096 unsigned long next_decay_max_lb_cost
;
1098 #ifdef CONFIG_SCHEDSTATS
1099 /* load_balance() stats */
1100 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
1101 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
1102 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
1103 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
1104 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
1105 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
1106 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
1107 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
1109 /* Active load balancing */
1110 unsigned int alb_count
;
1111 unsigned int alb_failed
;
1112 unsigned int alb_pushed
;
1114 /* SD_BALANCE_EXEC stats */
1115 unsigned int sbe_count
;
1116 unsigned int sbe_balanced
;
1117 unsigned int sbe_pushed
;
1119 /* SD_BALANCE_FORK stats */
1120 unsigned int sbf_count
;
1121 unsigned int sbf_balanced
;
1122 unsigned int sbf_pushed
;
1124 /* try_to_wake_up() stats */
1125 unsigned int ttwu_wake_remote
;
1126 unsigned int ttwu_move_affine
;
1127 unsigned int ttwu_move_balance
;
1129 #ifdef CONFIG_SCHED_DEBUG
1133 void *private; /* used during construction */
1134 struct rcu_head rcu
; /* used during destruction */
1137 unsigned int span_weight
;
1139 * Span of all CPUs in this domain.
1141 * NOTE: this field is variable length. (Allocated dynamically
1142 * by attaching extra space to the end of the structure,
1143 * depending on how many CPUs the kernel has booted up with)
1145 unsigned long span
[0];
1148 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1150 return to_cpumask(sd
->span
);
1153 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1154 struct sched_domain_attr
*dattr_new
);
1156 /* Allocate an array of sched domains, for partition_sched_domains(). */
1157 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1158 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1160 bool cpus_share_cache(int this_cpu
, int that_cpu
);
1162 typedef const struct cpumask
*(*sched_domain_mask_f
)(int cpu
);
1163 typedef int (*sched_domain_flags_f
)(void);
1165 #define SDTL_OVERLAP 0x01
1168 struct sched_domain
**__percpu sd
;
1169 struct sched_group
**__percpu sg
;
1170 struct sched_group_capacity
**__percpu sgc
;
1173 struct sched_domain_topology_level
{
1174 sched_domain_mask_f mask
;
1175 sched_domain_flags_f sd_flags
;
1178 struct sd_data data
;
1179 #ifdef CONFIG_SCHED_DEBUG
1184 extern void set_sched_topology(struct sched_domain_topology_level
*tl
);
1185 extern void wake_up_if_idle(int cpu
);
1187 #ifdef CONFIG_SCHED_DEBUG
1188 # define SD_INIT_NAME(type) .name = #type
1190 # define SD_INIT_NAME(type)
1193 #else /* CONFIG_SMP */
1195 struct sched_domain_attr
;
1198 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1199 struct sched_domain_attr
*dattr_new
)
1203 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1208 #endif /* !CONFIG_SMP */
1211 struct io_context
; /* See blkdev.h */
1214 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1215 extern void prefetch_stack(struct task_struct
*t
);
1217 static inline void prefetch_stack(struct task_struct
*t
) { }
1220 struct audit_context
; /* See audit.c */
1222 struct pipe_inode_info
;
1223 struct uts_namespace
;
1225 struct load_weight
{
1226 unsigned long weight
;
1231 * The load_avg/util_avg accumulates an infinite geometric series
1232 * (see __update_load_avg() in kernel/sched/fair.c).
1234 * [load_avg definition]
1236 * load_avg = runnable% * scale_load_down(load)
1238 * where runnable% is the time ratio that a sched_entity is runnable.
1239 * For cfs_rq, it is the aggregated load_avg of all runnable and
1240 * blocked sched_entities.
1242 * load_avg may also take frequency scaling into account:
1244 * load_avg = runnable% * scale_load_down(load) * freq%
1246 * where freq% is the CPU frequency normalized to the highest frequency.
1248 * [util_avg definition]
1250 * util_avg = running% * SCHED_CAPACITY_SCALE
1252 * where running% is the time ratio that a sched_entity is running on
1253 * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1254 * and blocked sched_entities.
1256 * util_avg may also factor frequency scaling and CPU capacity scaling:
1258 * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1260 * where freq% is the same as above, and capacity% is the CPU capacity
1261 * normalized to the greatest capacity (due to uarch differences, etc).
1263 * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1264 * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1265 * we therefore scale them to as large a range as necessary. This is for
1266 * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1270 * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1271 * with the highest load (=88761), always runnable on a single cfs_rq,
1272 * and should not overflow as the number already hits PID_MAX_LIMIT.
1274 * For all other cases (including 32-bit kernels), struct load_weight's
1275 * weight will overflow first before we do, because:
1277 * Max(load_avg) <= Max(load.weight)
1279 * Then it is the load_weight's responsibility to consider overflow
1283 u64 last_update_time
, load_sum
;
1284 u32 util_sum
, period_contrib
;
1285 unsigned long load_avg
, util_avg
;
1288 #ifdef CONFIG_SCHEDSTATS
1289 struct sched_statistics
{
1299 s64 sum_sleep_runtime
;
1306 u64 nr_migrations_cold
;
1307 u64 nr_failed_migrations_affine
;
1308 u64 nr_failed_migrations_running
;
1309 u64 nr_failed_migrations_hot
;
1310 u64 nr_forced_migrations
;
1313 u64 nr_wakeups_sync
;
1314 u64 nr_wakeups_migrate
;
1315 u64 nr_wakeups_local
;
1316 u64 nr_wakeups_remote
;
1317 u64 nr_wakeups_affine
;
1318 u64 nr_wakeups_affine_attempts
;
1319 u64 nr_wakeups_passive
;
1320 u64 nr_wakeups_idle
;
1324 struct sched_entity
{
1325 struct load_weight load
; /* for load-balancing */
1326 struct rb_node run_node
;
1327 struct list_head group_node
;
1331 u64 sum_exec_runtime
;
1333 u64 prev_sum_exec_runtime
;
1337 #ifdef CONFIG_SCHEDSTATS
1338 struct sched_statistics statistics
;
1341 #ifdef CONFIG_FAIR_GROUP_SCHED
1343 struct sched_entity
*parent
;
1344 /* rq on which this entity is (to be) queued: */
1345 struct cfs_rq
*cfs_rq
;
1346 /* rq "owned" by this entity/group: */
1347 struct cfs_rq
*my_q
;
1352 * Per entity load average tracking.
1354 * Put into separate cache line so it does not
1355 * collide with read-mostly values above.
1357 struct sched_avg avg ____cacheline_aligned_in_smp
;
1361 struct sched_rt_entity
{
1362 struct list_head run_list
;
1363 unsigned long timeout
;
1364 unsigned long watchdog_stamp
;
1365 unsigned int time_slice
;
1366 unsigned short on_rq
;
1367 unsigned short on_list
;
1369 struct sched_rt_entity
*back
;
1370 #ifdef CONFIG_RT_GROUP_SCHED
1371 struct sched_rt_entity
*parent
;
1372 /* rq on which this entity is (to be) queued: */
1373 struct rt_rq
*rt_rq
;
1374 /* rq "owned" by this entity/group: */
1379 struct sched_dl_entity
{
1380 struct rb_node rb_node
;
1383 * Original scheduling parameters. Copied here from sched_attr
1384 * during sched_setattr(), they will remain the same until
1385 * the next sched_setattr().
1387 u64 dl_runtime
; /* maximum runtime for each instance */
1388 u64 dl_deadline
; /* relative deadline of each instance */
1389 u64 dl_period
; /* separation of two instances (period) */
1390 u64 dl_bw
; /* dl_runtime / dl_deadline */
1393 * Actual scheduling parameters. Initialized with the values above,
1394 * they are continously updated during task execution. Note that
1395 * the remaining runtime could be < 0 in case we are in overrun.
1397 s64 runtime
; /* remaining runtime for this instance */
1398 u64 deadline
; /* absolute deadline for this instance */
1399 unsigned int flags
; /* specifying the scheduler behaviour */
1404 * @dl_throttled tells if we exhausted the runtime. If so, the
1405 * task has to wait for a replenishment to be performed at the
1406 * next firing of dl_timer.
1408 * @dl_boosted tells if we are boosted due to DI. If so we are
1409 * outside bandwidth enforcement mechanism (but only until we
1410 * exit the critical section);
1412 * @dl_yielded tells if task gave up the cpu before consuming
1413 * all its available runtime during the last job.
1415 int dl_throttled
, dl_boosted
, dl_yielded
;
1418 * Bandwidth enforcement timer. Each -deadline task has its
1419 * own bandwidth to be enforced, thus we need one timer per task.
1421 struct hrtimer dl_timer
;
1429 u8 pad
; /* Otherwise the compiler can store garbage here. */
1431 u32 s
; /* Set of bits. */
1435 enum perf_event_task_context
{
1436 perf_invalid_context
= -1,
1437 perf_hw_context
= 0,
1439 perf_nr_task_contexts
,
1442 /* Track pages that require TLB flushes */
1443 struct tlbflush_unmap_batch
{
1445 * Each bit set is a CPU that potentially has a TLB entry for one of
1446 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1448 struct cpumask cpumask
;
1450 /* True if any bit in cpumask is set */
1451 bool flush_required
;
1454 * If true then the PTE was dirty when unmapped. The entry must be
1455 * flushed before IO is initiated or a stale TLB entry potentially
1456 * allows an update without redirtying the page.
1461 struct task_struct
{
1462 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1465 unsigned int flags
; /* per process flags, defined below */
1466 unsigned int ptrace
;
1469 struct llist_node wake_entry
;
1471 unsigned int wakee_flips
;
1472 unsigned long wakee_flip_decay_ts
;
1473 struct task_struct
*last_wakee
;
1479 int prio
, static_prio
, normal_prio
;
1480 unsigned int rt_priority
;
1481 const struct sched_class
*sched_class
;
1482 struct sched_entity se
;
1483 struct sched_rt_entity rt
;
1484 #ifdef CONFIG_CGROUP_SCHED
1485 struct task_group
*sched_task_group
;
1487 struct sched_dl_entity dl
;
1489 #ifdef CONFIG_PREEMPT_NOTIFIERS
1490 /* list of struct preempt_notifier: */
1491 struct hlist_head preempt_notifiers
;
1494 #ifdef CONFIG_BLK_DEV_IO_TRACE
1495 unsigned int btrace_seq
;
1498 unsigned int policy
;
1499 int nr_cpus_allowed
;
1500 cpumask_t cpus_allowed
;
1502 #ifdef CONFIG_PREEMPT_RCU
1503 int rcu_read_lock_nesting
;
1504 union rcu_special rcu_read_unlock_special
;
1505 struct list_head rcu_node_entry
;
1506 struct rcu_node
*rcu_blocked_node
;
1507 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1508 #ifdef CONFIG_TASKS_RCU
1509 unsigned long rcu_tasks_nvcsw
;
1510 bool rcu_tasks_holdout
;
1511 struct list_head rcu_tasks_holdout_list
;
1512 int rcu_tasks_idle_cpu
;
1513 #endif /* #ifdef CONFIG_TASKS_RCU */
1515 #ifdef CONFIG_SCHED_INFO
1516 struct sched_info sched_info
;
1519 struct list_head tasks
;
1521 struct plist_node pushable_tasks
;
1522 struct rb_node pushable_dl_tasks
;
1525 struct mm_struct
*mm
, *active_mm
;
1526 /* per-thread vma caching */
1527 u32 vmacache_seqnum
;
1528 struct vm_area_struct
*vmacache
[VMACACHE_SIZE
];
1529 #if defined(SPLIT_RSS_COUNTING)
1530 struct task_rss_stat rss_stat
;
1534 int exit_code
, exit_signal
;
1535 int pdeath_signal
; /* The signal sent when the parent dies */
1536 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1538 /* Used for emulating ABI behavior of previous Linux versions */
1539 unsigned int personality
;
1541 /* scheduler bits, serialized by scheduler locks */
1542 unsigned sched_reset_on_fork
:1;
1543 unsigned sched_contributes_to_load
:1;
1544 unsigned sched_migrated
:1;
1545 unsigned sched_remote_wakeup
:1;
1546 unsigned :0; /* force alignment to the next boundary */
1548 /* unserialized, strictly 'current' */
1549 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1550 unsigned in_iowait
:1;
1551 #if !defined(TIF_RESTORE_SIGMASK)
1552 unsigned restore_sigmask
:1;
1555 unsigned memcg_may_oom
:1;
1557 unsigned memcg_kmem_skip_account
:1;
1560 #ifdef CONFIG_COMPAT_BRK
1561 unsigned brk_randomized
:1;
1564 unsigned long atomic_flags
; /* Flags needing atomic access. */
1566 struct restart_block restart_block
;
1571 #ifdef CONFIG_CC_STACKPROTECTOR
1572 /* Canary value for the -fstack-protector gcc feature */
1573 unsigned long stack_canary
;
1576 * pointers to (original) parent process, youngest child, younger sibling,
1577 * older sibling, respectively. (p->father can be replaced with
1578 * p->real_parent->pid)
1580 struct task_struct __rcu
*real_parent
; /* real parent process */
1581 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1583 * children/sibling forms the list of my natural children
1585 struct list_head children
; /* list of my children */
1586 struct list_head sibling
; /* linkage in my parent's children list */
1587 struct task_struct
*group_leader
; /* threadgroup leader */
1590 * ptraced is the list of tasks this task is using ptrace on.
1591 * This includes both natural children and PTRACE_ATTACH targets.
1592 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1594 struct list_head ptraced
;
1595 struct list_head ptrace_entry
;
1597 /* PID/PID hash table linkage. */
1598 struct pid_link pids
[PIDTYPE_MAX
];
1599 struct list_head thread_group
;
1600 struct list_head thread_node
;
1602 struct completion
*vfork_done
; /* for vfork() */
1603 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1604 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1606 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1608 struct prev_cputime prev_cputime
;
1609 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1610 seqcount_t vtime_seqcount
;
1611 unsigned long long vtime_snap
;
1613 /* Task is sleeping or running in a CPU with VTIME inactive */
1615 /* Task runs in userspace in a CPU with VTIME active */
1617 /* Task runs in kernelspace in a CPU with VTIME active */
1619 } vtime_snap_whence
;
1622 #ifdef CONFIG_NO_HZ_FULL
1623 atomic_t tick_dep_mask
;
1625 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1626 u64 start_time
; /* monotonic time in nsec */
1627 u64 real_start_time
; /* boot based time in nsec */
1628 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1629 unsigned long min_flt
, maj_flt
;
1631 struct task_cputime cputime_expires
;
1632 struct list_head cpu_timers
[3];
1634 /* process credentials */
1635 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1636 * credentials (COW) */
1637 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1638 * credentials (COW) */
1639 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1640 - access with [gs]et_task_comm (which lock
1641 it with task_lock())
1642 - initialized normally by setup_new_exec */
1643 /* file system info */
1644 struct nameidata
*nameidata
;
1645 #ifdef CONFIG_SYSVIPC
1647 struct sysv_sem sysvsem
;
1648 struct sysv_shm sysvshm
;
1650 #ifdef CONFIG_DETECT_HUNG_TASK
1651 /* hung task detection */
1652 unsigned long last_switch_count
;
1654 /* filesystem information */
1655 struct fs_struct
*fs
;
1656 /* open file information */
1657 struct files_struct
*files
;
1659 struct nsproxy
*nsproxy
;
1660 /* signal handlers */
1661 struct signal_struct
*signal
;
1662 struct sighand_struct
*sighand
;
1664 sigset_t blocked
, real_blocked
;
1665 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1666 struct sigpending pending
;
1668 unsigned long sas_ss_sp
;
1670 unsigned sas_ss_flags
;
1672 struct callback_head
*task_works
;
1674 struct audit_context
*audit_context
;
1675 #ifdef CONFIG_AUDITSYSCALL
1677 unsigned int sessionid
;
1679 struct seccomp seccomp
;
1681 /* Thread group tracking */
1684 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1686 spinlock_t alloc_lock
;
1688 /* Protection of the PI data structures: */
1689 raw_spinlock_t pi_lock
;
1691 struct wake_q_node wake_q
;
1693 #ifdef CONFIG_RT_MUTEXES
1694 /* PI waiters blocked on a rt_mutex held by this task */
1695 struct rb_root pi_waiters
;
1696 struct rb_node
*pi_waiters_leftmost
;
1697 /* Deadlock detection and priority inheritance handling */
1698 struct rt_mutex_waiter
*pi_blocked_on
;
1701 #ifdef CONFIG_DEBUG_MUTEXES
1702 /* mutex deadlock detection */
1703 struct mutex_waiter
*blocked_on
;
1705 #ifdef CONFIG_TRACE_IRQFLAGS
1706 unsigned int irq_events
;
1707 unsigned long hardirq_enable_ip
;
1708 unsigned long hardirq_disable_ip
;
1709 unsigned int hardirq_enable_event
;
1710 unsigned int hardirq_disable_event
;
1711 int hardirqs_enabled
;
1712 int hardirq_context
;
1713 unsigned long softirq_disable_ip
;
1714 unsigned long softirq_enable_ip
;
1715 unsigned int softirq_disable_event
;
1716 unsigned int softirq_enable_event
;
1717 int softirqs_enabled
;
1718 int softirq_context
;
1720 #ifdef CONFIG_LOCKDEP
1721 # define MAX_LOCK_DEPTH 48UL
1724 unsigned int lockdep_recursion
;
1725 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1726 gfp_t lockdep_reclaim_gfp
;
1729 unsigned int in_ubsan
;
1732 /* journalling filesystem info */
1735 /* stacked block device info */
1736 struct bio_list
*bio_list
;
1739 /* stack plugging */
1740 struct blk_plug
*plug
;
1744 struct reclaim_state
*reclaim_state
;
1746 struct backing_dev_info
*backing_dev_info
;
1748 struct io_context
*io_context
;
1750 unsigned long ptrace_message
;
1751 siginfo_t
*last_siginfo
; /* For ptrace use. */
1752 struct task_io_accounting ioac
;
1753 #if defined(CONFIG_TASK_XACCT)
1754 u64 acct_rss_mem1
; /* accumulated rss usage */
1755 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1756 cputime_t acct_timexpd
; /* stime + utime since last update */
1758 #ifdef CONFIG_CPUSETS
1759 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1760 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1761 int cpuset_mem_spread_rotor
;
1762 int cpuset_slab_spread_rotor
;
1764 #ifdef CONFIG_CGROUPS
1765 /* Control Group info protected by css_set_lock */
1766 struct css_set __rcu
*cgroups
;
1767 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1768 struct list_head cg_list
;
1771 struct robust_list_head __user
*robust_list
;
1772 #ifdef CONFIG_COMPAT
1773 struct compat_robust_list_head __user
*compat_robust_list
;
1775 struct list_head pi_state_list
;
1776 struct futex_pi_state
*pi_state_cache
;
1778 #ifdef CONFIG_PERF_EVENTS
1779 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1780 struct mutex perf_event_mutex
;
1781 struct list_head perf_event_list
;
1783 #ifdef CONFIG_DEBUG_PREEMPT
1784 unsigned long preempt_disable_ip
;
1787 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1789 short pref_node_fork
;
1791 #ifdef CONFIG_NUMA_BALANCING
1793 unsigned int numa_scan_period
;
1794 unsigned int numa_scan_period_max
;
1795 int numa_preferred_nid
;
1796 unsigned long numa_migrate_retry
;
1797 u64 node_stamp
; /* migration stamp */
1798 u64 last_task_numa_placement
;
1799 u64 last_sum_exec_runtime
;
1800 struct callback_head numa_work
;
1802 struct list_head numa_entry
;
1803 struct numa_group
*numa_group
;
1806 * numa_faults is an array split into four regions:
1807 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1808 * in this precise order.
1810 * faults_memory: Exponential decaying average of faults on a per-node
1811 * basis. Scheduling placement decisions are made based on these
1812 * counts. The values remain static for the duration of a PTE scan.
1813 * faults_cpu: Track the nodes the process was running on when a NUMA
1814 * hinting fault was incurred.
1815 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1816 * during the current scan window. When the scan completes, the counts
1817 * in faults_memory and faults_cpu decay and these values are copied.
1819 unsigned long *numa_faults
;
1820 unsigned long total_numa_faults
;
1823 * numa_faults_locality tracks if faults recorded during the last
1824 * scan window were remote/local or failed to migrate. The task scan
1825 * period is adapted based on the locality of the faults with different
1826 * weights depending on whether they were shared or private faults
1828 unsigned long numa_faults_locality
[3];
1830 unsigned long numa_pages_migrated
;
1831 #endif /* CONFIG_NUMA_BALANCING */
1833 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1834 struct tlbflush_unmap_batch tlb_ubc
;
1837 struct rcu_head rcu
;
1840 * cache last used pipe for splice
1842 struct pipe_inode_info
*splice_pipe
;
1844 struct page_frag task_frag
;
1846 #ifdef CONFIG_TASK_DELAY_ACCT
1847 struct task_delay_info
*delays
;
1849 #ifdef CONFIG_FAULT_INJECTION
1853 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1854 * balance_dirty_pages() for some dirty throttling pause
1857 int nr_dirtied_pause
;
1858 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1860 #ifdef CONFIG_LATENCYTOP
1861 int latency_record_count
;
1862 struct latency_record latency_record
[LT_SAVECOUNT
];
1865 * time slack values; these are used to round up poll() and
1866 * select() etc timeout values. These are in nanoseconds.
1869 u64 default_timer_slack_ns
;
1872 unsigned int kasan_depth
;
1874 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1875 /* Index of current stored address in ret_stack */
1877 /* Stack of return addresses for return function tracing */
1878 struct ftrace_ret_stack
*ret_stack
;
1879 /* time stamp for last schedule */
1880 unsigned long long ftrace_timestamp
;
1882 * Number of functions that haven't been traced
1883 * because of depth overrun.
1885 atomic_t trace_overrun
;
1886 /* Pause for the tracing */
1887 atomic_t tracing_graph_pause
;
1889 #ifdef CONFIG_TRACING
1890 /* state flags for use by tracers */
1891 unsigned long trace
;
1892 /* bitmask and counter of trace recursion */
1893 unsigned long trace_recursion
;
1894 #endif /* CONFIG_TRACING */
1896 /* Coverage collection mode enabled for this task (0 if disabled). */
1897 enum kcov_mode kcov_mode
;
1898 /* Size of the kcov_area. */
1900 /* Buffer for coverage collection. */
1902 /* kcov desciptor wired with this task or NULL. */
1906 struct mem_cgroup
*memcg_in_oom
;
1907 gfp_t memcg_oom_gfp_mask
;
1908 int memcg_oom_order
;
1910 /* number of pages to reclaim on returning to userland */
1911 unsigned int memcg_nr_pages_over_high
;
1913 #ifdef CONFIG_UPROBES
1914 struct uprobe_task
*utask
;
1916 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1917 unsigned int sequential_io
;
1918 unsigned int sequential_io_avg
;
1920 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1921 unsigned long task_state_change
;
1923 int pagefault_disabled
;
1925 struct task_struct
*oom_reaper_list
;
1927 #ifdef CONFIG_VMAP_STACK
1928 struct vm_struct
*stack_vm_area
;
1930 /* CPU-specific state of this task */
1931 struct thread_struct thread
;
1933 * WARNING: on x86, 'thread_struct' contains a variable-sized
1934 * structure. It *MUST* be at the end of 'task_struct'.
1936 * Do not put anything below here!
1940 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1941 extern int arch_task_struct_size __read_mostly
;
1943 # define arch_task_struct_size (sizeof(struct task_struct))
1946 #ifdef CONFIG_VMAP_STACK
1947 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1949 return t
->stack_vm_area
;
1952 static inline struct vm_struct
*task_stack_vm_area(const struct task_struct
*t
)
1958 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1959 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1961 static inline int tsk_nr_cpus_allowed(struct task_struct
*p
)
1963 return p
->nr_cpus_allowed
;
1966 #define TNF_MIGRATED 0x01
1967 #define TNF_NO_GROUP 0x02
1968 #define TNF_SHARED 0x04
1969 #define TNF_FAULT_LOCAL 0x08
1970 #define TNF_MIGRATE_FAIL 0x10
1972 static inline bool in_vfork(struct task_struct
*tsk
)
1977 * need RCU to access ->real_parent if CLONE_VM was used along with
1980 * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
1983 * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
1984 * ->real_parent is not necessarily the task doing vfork(), so in
1985 * theory we can't rely on task_lock() if we want to dereference it.
1987 * And in this case we can't trust the real_parent->mm == tsk->mm
1988 * check, it can be false negative. But we do not care, if init or
1989 * another oom-unkillable task does this it should blame itself.
1992 ret
= tsk
->vfork_done
&& tsk
->real_parent
->mm
== tsk
->mm
;
1998 #ifdef CONFIG_NUMA_BALANCING
1999 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
2000 extern pid_t
task_numa_group_id(struct task_struct
*p
);
2001 extern void set_numabalancing_state(bool enabled
);
2002 extern void task_numa_free(struct task_struct
*p
);
2003 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
2004 int src_nid
, int dst_cpu
);
2006 static inline void task_numa_fault(int last_node
, int node
, int pages
,
2010 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
2014 static inline void set_numabalancing_state(bool enabled
)
2017 static inline void task_numa_free(struct task_struct
*p
)
2020 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
2021 struct page
*page
, int src_nid
, int dst_cpu
)
2027 static inline struct pid
*task_pid(struct task_struct
*task
)
2029 return task
->pids
[PIDTYPE_PID
].pid
;
2032 static inline struct pid
*task_tgid(struct task_struct
*task
)
2034 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
2038 * Without tasklist or rcu lock it is not safe to dereference
2039 * the result of task_pgrp/task_session even if task == current,
2040 * we can race with another thread doing sys_setsid/sys_setpgid.
2042 static inline struct pid
*task_pgrp(struct task_struct
*task
)
2044 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
2047 static inline struct pid
*task_session(struct task_struct
*task
)
2049 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
2052 struct pid_namespace
;
2055 * the helpers to get the task's different pids as they are seen
2056 * from various namespaces
2058 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
2059 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
2061 * task_xid_nr_ns() : id seen from the ns specified;
2063 * set_task_vxid() : assigns a virtual id to a task;
2065 * see also pid_nr() etc in include/linux/pid.h
2067 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
2068 struct pid_namespace
*ns
);
2070 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
2075 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
2076 struct pid_namespace
*ns
)
2078 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
2081 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
2083 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
2087 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
2092 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
2094 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
2096 return pid_vnr(task_tgid(tsk
));
2100 static inline int pid_alive(const struct task_struct
*p
);
2101 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
2107 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
2113 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
2115 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
2118 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
2119 struct pid_namespace
*ns
)
2121 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
2124 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
2126 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
2130 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
2131 struct pid_namespace
*ns
)
2133 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
2136 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
2138 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
2141 /* obsolete, do not use */
2142 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
2144 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
2148 * pid_alive - check that a task structure is not stale
2149 * @p: Task structure to be checked.
2151 * Test if a process is not yet dead (at most zombie state)
2152 * If pid_alive fails, then pointers within the task structure
2153 * can be stale and must not be dereferenced.
2155 * Return: 1 if the process is alive. 0 otherwise.
2157 static inline int pid_alive(const struct task_struct
*p
)
2159 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
2163 * is_global_init - check if a task structure is init. Since init
2164 * is free to have sub-threads we need to check tgid.
2165 * @tsk: Task structure to be checked.
2167 * Check if a task structure is the first user space task the kernel created.
2169 * Return: 1 if the task structure is init. 0 otherwise.
2171 static inline int is_global_init(struct task_struct
*tsk
)
2173 return task_tgid_nr(tsk
) == 1;
2176 extern struct pid
*cad_pid
;
2178 extern void free_task(struct task_struct
*tsk
);
2179 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2181 extern void __put_task_struct(struct task_struct
*t
);
2183 static inline void put_task_struct(struct task_struct
*t
)
2185 if (atomic_dec_and_test(&t
->usage
))
2186 __put_task_struct(t
);
2189 struct task_struct
*task_rcu_dereference(struct task_struct
**ptask
);
2190 struct task_struct
*try_get_task_struct(struct task_struct
**ptask
);
2192 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2193 extern void task_cputime(struct task_struct
*t
,
2194 cputime_t
*utime
, cputime_t
*stime
);
2195 extern void task_cputime_scaled(struct task_struct
*t
,
2196 cputime_t
*utimescaled
, cputime_t
*stimescaled
);
2197 extern cputime_t
task_gtime(struct task_struct
*t
);
2199 static inline void task_cputime(struct task_struct
*t
,
2200 cputime_t
*utime
, cputime_t
*stime
)
2208 static inline void task_cputime_scaled(struct task_struct
*t
,
2209 cputime_t
*utimescaled
,
2210 cputime_t
*stimescaled
)
2213 *utimescaled
= t
->utimescaled
;
2215 *stimescaled
= t
->stimescaled
;
2218 static inline cputime_t
task_gtime(struct task_struct
*t
)
2223 extern void task_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2224 extern void thread_group_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2229 #define PF_EXITING 0x00000004 /* getting shut down */
2230 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2231 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2232 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2233 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2234 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2235 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2236 #define PF_DUMPCORE 0x00000200 /* dumped core */
2237 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2238 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2239 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2240 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2241 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2242 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2243 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2244 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2245 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2246 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2247 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2248 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2249 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2250 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2251 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2252 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2253 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2254 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2255 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2258 * Only the _current_ task can read/write to tsk->flags, but other
2259 * tasks can access tsk->flags in readonly mode for example
2260 * with tsk_used_math (like during threaded core dumping).
2261 * There is however an exception to this rule during ptrace
2262 * or during fork: the ptracer task is allowed to write to the
2263 * child->flags of its traced child (same goes for fork, the parent
2264 * can write to the child->flags), because we're guaranteed the
2265 * child is not running and in turn not changing child->flags
2266 * at the same time the parent does it.
2268 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2269 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2270 #define clear_used_math() clear_stopped_child_used_math(current)
2271 #define set_used_math() set_stopped_child_used_math(current)
2272 #define conditional_stopped_child_used_math(condition, child) \
2273 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2274 #define conditional_used_math(condition) \
2275 conditional_stopped_child_used_math(condition, current)
2276 #define copy_to_stopped_child_used_math(child) \
2277 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2278 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2279 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2280 #define used_math() tsk_used_math(current)
2282 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2283 * __GFP_FS is also cleared as it implies __GFP_IO.
2285 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
2287 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
2288 flags
&= ~(__GFP_IO
| __GFP_FS
);
2292 static inline unsigned int memalloc_noio_save(void)
2294 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
2295 current
->flags
|= PF_MEMALLOC_NOIO
;
2299 static inline void memalloc_noio_restore(unsigned int flags
)
2301 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
2304 /* Per-process atomic flags. */
2305 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2306 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2307 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2308 #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */
2311 #define TASK_PFA_TEST(name, func) \
2312 static inline bool task_##func(struct task_struct *p) \
2313 { return test_bit(PFA_##name, &p->atomic_flags); }
2314 #define TASK_PFA_SET(name, func) \
2315 static inline void task_set_##func(struct task_struct *p) \
2316 { set_bit(PFA_##name, &p->atomic_flags); }
2317 #define TASK_PFA_CLEAR(name, func) \
2318 static inline void task_clear_##func(struct task_struct *p) \
2319 { clear_bit(PFA_##name, &p->atomic_flags); }
2321 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
2322 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
2324 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
2325 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
2326 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2328 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2329 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2330 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2332 TASK_PFA_TEST(LMK_WAITING
, lmk_waiting
)
2333 TASK_PFA_SET(LMK_WAITING
, lmk_waiting
)
2336 * task->jobctl flags
2338 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2340 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2341 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2342 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2343 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2344 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2345 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2346 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2348 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2349 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2350 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2351 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2352 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2353 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2354 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2356 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2357 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2359 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2360 unsigned long mask
);
2361 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2362 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2363 unsigned long mask
);
2365 static inline void rcu_copy_process(struct task_struct
*p
)
2367 #ifdef CONFIG_PREEMPT_RCU
2368 p
->rcu_read_lock_nesting
= 0;
2369 p
->rcu_read_unlock_special
.s
= 0;
2370 p
->rcu_blocked_node
= NULL
;
2371 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2372 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2373 #ifdef CONFIG_TASKS_RCU
2374 p
->rcu_tasks_holdout
= false;
2375 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2376 p
->rcu_tasks_idle_cpu
= -1;
2377 #endif /* #ifdef CONFIG_TASKS_RCU */
2380 static inline void tsk_restore_flags(struct task_struct
*task
,
2381 unsigned long orig_flags
, unsigned long flags
)
2383 task
->flags
&= ~flags
;
2384 task
->flags
|= orig_flags
& flags
;
2387 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2388 const struct cpumask
*trial
);
2389 extern int task_can_attach(struct task_struct
*p
,
2390 const struct cpumask
*cs_cpus_allowed
);
2392 extern void do_set_cpus_allowed(struct task_struct
*p
,
2393 const struct cpumask
*new_mask
);
2395 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2396 const struct cpumask
*new_mask
);
2398 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2399 const struct cpumask
*new_mask
)
2402 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2403 const struct cpumask
*new_mask
)
2405 if (!cpumask_test_cpu(0, new_mask
))
2411 #ifdef CONFIG_NO_HZ_COMMON
2412 void calc_load_enter_idle(void);
2413 void calc_load_exit_idle(void);
2415 static inline void calc_load_enter_idle(void) { }
2416 static inline void calc_load_exit_idle(void) { }
2417 #endif /* CONFIG_NO_HZ_COMMON */
2420 * Do not use outside of architecture code which knows its limitations.
2422 * sched_clock() has no promise of monotonicity or bounded drift between
2423 * CPUs, use (which you should not) requires disabling IRQs.
2425 * Please use one of the three interfaces below.
2427 extern unsigned long long notrace
sched_clock(void);
2429 * See the comment in kernel/sched/clock.c
2431 extern u64
running_clock(void);
2432 extern u64
sched_clock_cpu(int cpu
);
2435 extern void sched_clock_init(void);
2437 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2438 static inline void sched_clock_tick(void)
2442 static inline void sched_clock_idle_sleep_event(void)
2446 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
2450 static inline u64
cpu_clock(int cpu
)
2452 return sched_clock();
2455 static inline u64
local_clock(void)
2457 return sched_clock();
2461 * Architectures can set this to 1 if they have specified
2462 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2463 * but then during bootup it turns out that sched_clock()
2464 * is reliable after all:
2466 extern int sched_clock_stable(void);
2467 extern void set_sched_clock_stable(void);
2468 extern void clear_sched_clock_stable(void);
2470 extern void sched_clock_tick(void);
2471 extern void sched_clock_idle_sleep_event(void);
2472 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2475 * As outlined in clock.c, provides a fast, high resolution, nanosecond
2476 * time source that is monotonic per cpu argument and has bounded drift
2479 * ######################### BIG FAT WARNING ##########################
2480 * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2481 * # go backwards !! #
2482 * ####################################################################
2484 static inline u64
cpu_clock(int cpu
)
2486 return sched_clock_cpu(cpu
);
2489 static inline u64
local_clock(void)
2491 return sched_clock_cpu(raw_smp_processor_id());
2495 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2497 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2498 * The reason for this explicit opt-in is not to have perf penalty with
2499 * slow sched_clocks.
2501 extern void enable_sched_clock_irqtime(void);
2502 extern void disable_sched_clock_irqtime(void);
2504 static inline void enable_sched_clock_irqtime(void) {}
2505 static inline void disable_sched_clock_irqtime(void) {}
2508 extern unsigned long long
2509 task_sched_runtime(struct task_struct
*task
);
2511 /* sched_exec is called by processes performing an exec */
2513 extern void sched_exec(void);
2515 #define sched_exec() {}
2518 extern void sched_clock_idle_sleep_event(void);
2519 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2521 #ifdef CONFIG_HOTPLUG_CPU
2522 extern void idle_task_exit(void);
2524 static inline void idle_task_exit(void) {}
2527 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2528 extern void wake_up_nohz_cpu(int cpu
);
2530 static inline void wake_up_nohz_cpu(int cpu
) { }
2533 #ifdef CONFIG_NO_HZ_FULL
2534 extern u64
scheduler_tick_max_deferment(void);
2537 #ifdef CONFIG_SCHED_AUTOGROUP
2538 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2539 extern void sched_autogroup_detach(struct task_struct
*p
);
2540 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2541 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2542 #ifdef CONFIG_PROC_FS
2543 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2544 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2547 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2548 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2549 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2550 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2553 extern int yield_to(struct task_struct
*p
, bool preempt
);
2554 extern void set_user_nice(struct task_struct
*p
, long nice
);
2555 extern int task_prio(const struct task_struct
*p
);
2557 * task_nice - return the nice value of a given task.
2558 * @p: the task in question.
2560 * Return: The nice value [ -20 ... 0 ... 19 ].
2562 static inline int task_nice(const struct task_struct
*p
)
2564 return PRIO_TO_NICE((p
)->static_prio
);
2566 extern int can_nice(const struct task_struct
*p
, const int nice
);
2567 extern int task_curr(const struct task_struct
*p
);
2568 extern int idle_cpu(int cpu
);
2569 extern int sched_setscheduler(struct task_struct
*, int,
2570 const struct sched_param
*);
2571 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2572 const struct sched_param
*);
2573 extern int sched_setattr(struct task_struct
*,
2574 const struct sched_attr
*);
2575 extern struct task_struct
*idle_task(int cpu
);
2577 * is_idle_task - is the specified task an idle task?
2578 * @p: the task in question.
2580 * Return: 1 if @p is an idle task. 0 otherwise.
2582 static inline bool is_idle_task(const struct task_struct
*p
)
2586 extern struct task_struct
*curr_task(int cpu
);
2587 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2591 union thread_union
{
2592 struct thread_info thread_info
;
2593 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2596 #ifndef __HAVE_ARCH_KSTACK_END
2597 static inline int kstack_end(void *addr
)
2599 /* Reliable end of stack detection:
2600 * Some APM bios versions misalign the stack
2602 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2606 extern union thread_union init_thread_union
;
2607 extern struct task_struct init_task
;
2609 extern struct mm_struct init_mm
;
2611 extern struct pid_namespace init_pid_ns
;
2614 * find a task by one of its numerical ids
2616 * find_task_by_pid_ns():
2617 * finds a task by its pid in the specified namespace
2618 * find_task_by_vpid():
2619 * finds a task by its virtual pid
2621 * see also find_vpid() etc in include/linux/pid.h
2624 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2625 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2626 struct pid_namespace
*ns
);
2628 /* per-UID process charging. */
2629 extern struct user_struct
* alloc_uid(kuid_t
);
2630 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2632 atomic_inc(&u
->__count
);
2635 extern void free_uid(struct user_struct
*);
2637 #include <asm/current.h>
2639 extern void xtime_update(unsigned long ticks
);
2641 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2642 extern int wake_up_process(struct task_struct
*tsk
);
2643 extern void wake_up_new_task(struct task_struct
*tsk
);
2645 extern void kick_process(struct task_struct
*tsk
);
2647 static inline void kick_process(struct task_struct
*tsk
) { }
2649 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2650 extern void sched_dead(struct task_struct
*p
);
2652 extern void proc_caches_init(void);
2653 extern void flush_signals(struct task_struct
*);
2654 extern void ignore_signals(struct task_struct
*);
2655 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2656 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2658 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2660 struct task_struct
*tsk
= current
;
2664 spin_lock_irq(&tsk
->sighand
->siglock
);
2665 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2666 spin_unlock_irq(&tsk
->sighand
->siglock
);
2671 static inline void kernel_signal_stop(void)
2673 spin_lock_irq(¤t
->sighand
->siglock
);
2674 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2675 __set_current_state(TASK_STOPPED
);
2676 spin_unlock_irq(¤t
->sighand
->siglock
);
2681 extern void release_task(struct task_struct
* p
);
2682 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2683 extern int force_sigsegv(int, struct task_struct
*);
2684 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2685 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2686 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2687 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2688 const struct cred
*, u32
);
2689 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2690 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2691 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2692 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2693 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2694 extern void force_sig(int, struct task_struct
*);
2695 extern int send_sig(int, struct task_struct
*, int);
2696 extern int zap_other_threads(struct task_struct
*p
);
2697 extern struct sigqueue
*sigqueue_alloc(void);
2698 extern void sigqueue_free(struct sigqueue
*);
2699 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2700 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2702 #ifdef TIF_RESTORE_SIGMASK
2704 * Legacy restore_sigmask accessors. These are inefficient on
2705 * SMP architectures because they require atomic operations.
2709 * set_restore_sigmask() - make sure saved_sigmask processing gets done
2711 * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2712 * will run before returning to user mode, to process the flag. For
2713 * all callers, TIF_SIGPENDING is already set or it's no harm to set
2714 * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the
2715 * arch code will notice on return to user mode, in case those bits
2716 * are scarce. We set TIF_SIGPENDING here to ensure that the arch
2717 * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2719 static inline void set_restore_sigmask(void)
2721 set_thread_flag(TIF_RESTORE_SIGMASK
);
2722 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2724 static inline void clear_restore_sigmask(void)
2726 clear_thread_flag(TIF_RESTORE_SIGMASK
);
2728 static inline bool test_restore_sigmask(void)
2730 return test_thread_flag(TIF_RESTORE_SIGMASK
);
2732 static inline bool test_and_clear_restore_sigmask(void)
2734 return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK
);
2737 #else /* TIF_RESTORE_SIGMASK */
2739 /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2740 static inline void set_restore_sigmask(void)
2742 current
->restore_sigmask
= true;
2743 WARN_ON(!test_thread_flag(TIF_SIGPENDING
));
2745 static inline void clear_restore_sigmask(void)
2747 current
->restore_sigmask
= false;
2749 static inline bool test_restore_sigmask(void)
2751 return current
->restore_sigmask
;
2753 static inline bool test_and_clear_restore_sigmask(void)
2755 if (!current
->restore_sigmask
)
2757 current
->restore_sigmask
= false;
2762 static inline void restore_saved_sigmask(void)
2764 if (test_and_clear_restore_sigmask())
2765 __set_current_blocked(¤t
->saved_sigmask
);
2768 static inline sigset_t
*sigmask_to_save(void)
2770 sigset_t
*res
= ¤t
->blocked
;
2771 if (unlikely(test_restore_sigmask()))
2772 res
= ¤t
->saved_sigmask
;
2776 static inline int kill_cad_pid(int sig
, int priv
)
2778 return kill_pid(cad_pid
, sig
, priv
);
2781 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2782 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2783 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2784 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2787 * True if we are on the alternate signal stack.
2789 static inline int on_sig_stack(unsigned long sp
)
2792 * If the signal stack is SS_AUTODISARM then, by construction, we
2793 * can't be on the signal stack unless user code deliberately set
2794 * SS_AUTODISARM when we were already on it.
2796 * This improves reliability: if user state gets corrupted such that
2797 * the stack pointer points very close to the end of the signal stack,
2798 * then this check will enable the signal to be handled anyway.
2800 if (current
->sas_ss_flags
& SS_AUTODISARM
)
2803 #ifdef CONFIG_STACK_GROWSUP
2804 return sp
>= current
->sas_ss_sp
&&
2805 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2807 return sp
> current
->sas_ss_sp
&&
2808 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2812 static inline int sas_ss_flags(unsigned long sp
)
2814 if (!current
->sas_ss_size
)
2817 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2820 static inline void sas_ss_reset(struct task_struct
*p
)
2824 p
->sas_ss_flags
= SS_DISABLE
;
2827 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2829 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2830 #ifdef CONFIG_STACK_GROWSUP
2831 return current
->sas_ss_sp
;
2833 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2839 * Routines for handling mm_structs
2841 extern struct mm_struct
* mm_alloc(void);
2843 /* mmdrop drops the mm and the page tables */
2844 extern void __mmdrop(struct mm_struct
*);
2845 static inline void mmdrop(struct mm_struct
*mm
)
2847 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2851 static inline bool mmget_not_zero(struct mm_struct
*mm
)
2853 return atomic_inc_not_zero(&mm
->mm_users
);
2856 /* mmput gets rid of the mappings and all user-space */
2857 extern void mmput(struct mm_struct
*);
2859 /* same as above but performs the slow path from the async context. Can
2860 * be called from the atomic context as well
2862 extern void mmput_async(struct mm_struct
*);
2865 /* Grab a reference to a task's mm, if it is not already going away */
2866 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2868 * Grab a reference to a task's mm, if it is not already going away
2869 * and ptrace_may_access with the mode parameter passed to it
2872 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2873 /* Remove the current tasks stale references to the old mm_struct */
2874 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2876 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2877 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2878 struct task_struct
*, unsigned long);
2880 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2881 struct task_struct
*);
2883 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2884 * via pt_regs, so ignore the tls argument passed via C. */
2885 static inline int copy_thread_tls(
2886 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2887 struct task_struct
*p
, unsigned long tls
)
2889 return copy_thread(clone_flags
, sp
, arg
, p
);
2892 extern void flush_thread(void);
2894 #ifdef CONFIG_HAVE_EXIT_THREAD
2895 extern void exit_thread(struct task_struct
*tsk
);
2897 static inline void exit_thread(struct task_struct
*tsk
)
2902 extern void exit_files(struct task_struct
*);
2903 extern void __cleanup_sighand(struct sighand_struct
*);
2905 extern void exit_itimers(struct signal_struct
*);
2906 extern void flush_itimer_signals(void);
2908 extern void do_group_exit(int);
2910 extern int do_execve(struct filename
*,
2911 const char __user
* const __user
*,
2912 const char __user
* const __user
*);
2913 extern int do_execveat(int, struct filename
*,
2914 const char __user
* const __user
*,
2915 const char __user
* const __user
*,
2917 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
2918 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2919 struct task_struct
*fork_idle(int);
2920 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2922 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
2923 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
2925 __set_task_comm(tsk
, from
, false);
2927 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2930 void scheduler_ipi(void);
2931 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2933 static inline void scheduler_ipi(void) { }
2934 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2941 #define tasklist_empty() \
2942 list_empty(&init_task.tasks)
2944 #define next_task(p) \
2945 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2947 #define for_each_process(p) \
2948 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2950 extern bool current_is_single_threaded(void);
2953 * Careful: do_each_thread/while_each_thread is a double loop so
2954 * 'break' will not work as expected - use goto instead.
2956 #define do_each_thread(g, t) \
2957 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2959 #define while_each_thread(g, t) \
2960 while ((t = next_thread(t)) != g)
2962 #define __for_each_thread(signal, t) \
2963 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2965 #define for_each_thread(p, t) \
2966 __for_each_thread((p)->signal, t)
2968 /* Careful: this is a double loop, 'break' won't work as expected. */
2969 #define for_each_process_thread(p, t) \
2970 for_each_process(p) for_each_thread(p, t)
2972 static inline int get_nr_threads(struct task_struct
*tsk
)
2974 return tsk
->signal
->nr_threads
;
2977 static inline bool thread_group_leader(struct task_struct
*p
)
2979 return p
->exit_signal
>= 0;
2982 /* Do to the insanities of de_thread it is possible for a process
2983 * to have the pid of the thread group leader without actually being
2984 * the thread group leader. For iteration through the pids in proc
2985 * all we care about is that we have a task with the appropriate
2986 * pid, we don't actually care if we have the right task.
2988 static inline bool has_group_leader_pid(struct task_struct
*p
)
2990 return task_pid(p
) == p
->signal
->leader_pid
;
2994 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2996 return p1
->signal
== p2
->signal
;
2999 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
3001 return list_entry_rcu(p
->thread_group
.next
,
3002 struct task_struct
, thread_group
);
3005 static inline int thread_group_empty(struct task_struct
*p
)
3007 return list_empty(&p
->thread_group
);
3010 #define delay_group_leader(p) \
3011 (thread_group_leader(p) && !thread_group_empty(p))
3014 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
3015 * subscriptions and synchronises with wait4(). Also used in procfs. Also
3016 * pins the final release of task.io_context. Also protects ->cpuset and
3017 * ->cgroup.subsys[]. And ->vfork_done.
3019 * Nests both inside and outside of read_lock(&tasklist_lock).
3020 * It must not be nested with write_lock_irq(&tasklist_lock),
3021 * neither inside nor outside.
3023 static inline void task_lock(struct task_struct
*p
)
3025 spin_lock(&p
->alloc_lock
);
3028 static inline void task_unlock(struct task_struct
*p
)
3030 spin_unlock(&p
->alloc_lock
);
3033 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
3034 unsigned long *flags
);
3036 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
3037 unsigned long *flags
)
3039 struct sighand_struct
*ret
;
3041 ret
= __lock_task_sighand(tsk
, flags
);
3042 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
3046 static inline void unlock_task_sighand(struct task_struct
*tsk
,
3047 unsigned long *flags
)
3049 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
3053 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
3054 * @tsk: task causing the changes
3056 * All operations which modify a threadgroup - a new thread joining the
3057 * group, death of a member thread (the assertion of PF_EXITING) and
3058 * exec(2) dethreading the process and replacing the leader - are wrapped
3059 * by threadgroup_change_{begin|end}(). This is to provide a place which
3060 * subsystems needing threadgroup stability can hook into for
3063 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
3066 cgroup_threadgroup_change_begin(tsk
);
3070 * threadgroup_change_end - mark the end of changes to a threadgroup
3071 * @tsk: task causing the changes
3073 * See threadgroup_change_begin().
3075 static inline void threadgroup_change_end(struct task_struct
*tsk
)
3077 cgroup_threadgroup_change_end(tsk
);
3080 #ifndef __HAVE_THREAD_FUNCTIONS
3082 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
3083 #define task_stack_page(task) ((task)->stack)
3085 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
3087 *task_thread_info(p
) = *task_thread_info(org
);
3088 task_thread_info(p
)->task
= p
;
3092 * Return the address of the last usable long on the stack.
3094 * When the stack grows down, this is just above the thread
3095 * info struct. Going any lower will corrupt the threadinfo.
3097 * When the stack grows up, this is the highest address.
3098 * Beyond that position, we corrupt data on the next page.
3100 static inline unsigned long *end_of_stack(struct task_struct
*p
)
3102 #ifdef CONFIG_STACK_GROWSUP
3103 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
3105 return (unsigned long *)(task_thread_info(p
) + 1);
3110 #define task_stack_end_corrupted(task) \
3111 (*(end_of_stack(task)) != STACK_END_MAGIC)
3113 static inline int object_is_on_stack(void *obj
)
3115 void *stack
= task_stack_page(current
);
3117 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
3120 extern void thread_stack_cache_init(void);
3122 #ifdef CONFIG_DEBUG_STACK_USAGE
3123 static inline unsigned long stack_not_used(struct task_struct
*p
)
3125 unsigned long *n
= end_of_stack(p
);
3127 do { /* Skip over canary */
3128 # ifdef CONFIG_STACK_GROWSUP
3135 # ifdef CONFIG_STACK_GROWSUP
3136 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
3138 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
3142 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
3144 /* set thread flags in other task's structures
3145 * - see asm/thread_info.h for TIF_xxxx flags available
3147 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3149 set_ti_thread_flag(task_thread_info(tsk
), flag
);
3152 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3154 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3157 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3159 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
3162 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3164 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
3167 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
3169 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
3172 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
3174 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3177 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
3179 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
3182 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
3184 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
3187 static inline int restart_syscall(void)
3189 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
3190 return -ERESTARTNOINTR
;
3193 static inline int signal_pending(struct task_struct
*p
)
3195 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
3198 static inline int __fatal_signal_pending(struct task_struct
*p
)
3200 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
3203 static inline int fatal_signal_pending(struct task_struct
*p
)
3205 return signal_pending(p
) && __fatal_signal_pending(p
);
3208 static inline int signal_pending_state(long state
, struct task_struct
*p
)
3210 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
3212 if (!signal_pending(p
))
3215 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
3219 * cond_resched() and cond_resched_lock(): latency reduction via
3220 * explicit rescheduling in places that are safe. The return
3221 * value indicates whether a reschedule was done in fact.
3222 * cond_resched_lock() will drop the spinlock before scheduling,
3223 * cond_resched_softirq() will enable bhs before scheduling.
3225 extern int _cond_resched(void);
3227 #define cond_resched() ({ \
3228 ___might_sleep(__FILE__, __LINE__, 0); \
3232 extern int __cond_resched_lock(spinlock_t
*lock
);
3234 #define cond_resched_lock(lock) ({ \
3235 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3236 __cond_resched_lock(lock); \
3239 extern int __cond_resched_softirq(void);
3241 #define cond_resched_softirq() ({ \
3242 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
3243 __cond_resched_softirq(); \
3246 static inline void cond_resched_rcu(void)
3248 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3255 static inline unsigned long get_preempt_disable_ip(struct task_struct
*p
)
3257 #ifdef CONFIG_DEBUG_PREEMPT
3258 return p
->preempt_disable_ip
;
3265 * Does a critical section need to be broken due to another
3266 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3267 * but a general need for low latency)
3269 static inline int spin_needbreak(spinlock_t
*lock
)
3271 #ifdef CONFIG_PREEMPT
3272 return spin_is_contended(lock
);
3279 * Idle thread specific functions to determine the need_resched
3282 #ifdef TIF_POLLING_NRFLAG
3283 static inline int tsk_is_polling(struct task_struct
*p
)
3285 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
3288 static inline void __current_set_polling(void)
3290 set_thread_flag(TIF_POLLING_NRFLAG
);
3293 static inline bool __must_check
current_set_polling_and_test(void)
3295 __current_set_polling();
3298 * Polling state must be visible before we test NEED_RESCHED,
3299 * paired by resched_curr()
3301 smp_mb__after_atomic();
3303 return unlikely(tif_need_resched());
3306 static inline void __current_clr_polling(void)
3308 clear_thread_flag(TIF_POLLING_NRFLAG
);
3311 static inline bool __must_check
current_clr_polling_and_test(void)
3313 __current_clr_polling();
3316 * Polling state must be visible before we test NEED_RESCHED,
3317 * paired by resched_curr()
3319 smp_mb__after_atomic();
3321 return unlikely(tif_need_resched());
3325 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
3326 static inline void __current_set_polling(void) { }
3327 static inline void __current_clr_polling(void) { }
3329 static inline bool __must_check
current_set_polling_and_test(void)
3331 return unlikely(tif_need_resched());
3333 static inline bool __must_check
current_clr_polling_and_test(void)
3335 return unlikely(tif_need_resched());
3339 static inline void current_clr_polling(void)
3341 __current_clr_polling();
3344 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3345 * Once the bit is cleared, we'll get IPIs with every new
3346 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3349 smp_mb(); /* paired with resched_curr() */
3351 preempt_fold_need_resched();
3354 static __always_inline
bool need_resched(void)
3356 return unlikely(tif_need_resched());
3360 * Thread group CPU time accounting.
3362 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
3363 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
3366 * Reevaluate whether the task has signals pending delivery.
3367 * Wake the task if so.
3368 * This is required every time the blocked sigset_t changes.
3369 * callers must hold sighand->siglock.
3371 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
3372 extern void recalc_sigpending(void);
3374 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
3376 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
3378 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
3380 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
3382 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
3386 * Wrappers for p->thread_info->cpu access. No-op on UP.
3390 static inline unsigned int task_cpu(const struct task_struct
*p
)
3392 return task_thread_info(p
)->cpu
;
3395 static inline int task_node(const struct task_struct
*p
)
3397 return cpu_to_node(task_cpu(p
));
3400 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
3404 static inline unsigned int task_cpu(const struct task_struct
*p
)
3409 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
3413 #endif /* CONFIG_SMP */
3415 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3416 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3418 #ifdef CONFIG_CGROUP_SCHED
3419 extern struct task_group root_task_group
;
3420 #endif /* CONFIG_CGROUP_SCHED */
3422 extern int task_can_switch_user(struct user_struct
*up
,
3423 struct task_struct
*tsk
);
3425 #ifdef CONFIG_TASK_XACCT
3426 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3428 tsk
->ioac
.rchar
+= amt
;
3431 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3433 tsk
->ioac
.wchar
+= amt
;
3436 static inline void inc_syscr(struct task_struct
*tsk
)
3441 static inline void inc_syscw(struct task_struct
*tsk
)
3446 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3450 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3454 static inline void inc_syscr(struct task_struct
*tsk
)
3458 static inline void inc_syscw(struct task_struct
*tsk
)
3463 #ifndef TASK_SIZE_OF
3464 #define TASK_SIZE_OF(tsk) TASK_SIZE
3468 extern void mm_update_next_owner(struct mm_struct
*mm
);
3470 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3473 #endif /* CONFIG_MEMCG */
3475 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3478 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3481 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3484 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3487 static inline unsigned long rlimit(unsigned int limit
)
3489 return task_rlimit(current
, limit
);
3492 static inline unsigned long rlimit_max(unsigned int limit
)
3494 return task_rlimit_max(current
, limit
);
3497 #define SCHED_CPUFREQ_RT (1U << 0)
3498 #define SCHED_CPUFREQ_DL (1U << 1)
3500 #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL)
3502 #ifdef CONFIG_CPU_FREQ
3503 struct update_util_data
{
3504 void (*func
)(struct update_util_data
*data
, u64 time
, unsigned int flags
);
3507 void cpufreq_add_update_util_hook(int cpu
, struct update_util_data
*data
,
3508 void (*func
)(struct update_util_data
*data
, u64 time
,
3509 unsigned int flags
));
3510 void cpufreq_remove_update_util_hook(int cpu
);
3511 #endif /* CONFIG_CPU_FREQ */