4 #include <uapi/linux/sched.h>
6 #include <linux/sched/prio.h>
13 #include <asm/param.h> /* for HZ */
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/kcov.h>
55 #include <linux/task_io_accounting.h>
56 #include <linux/latencytop.h>
57 #include <linux/cred.h>
58 #include <linux/llist.h>
59 #include <linux/uidgid.h>
60 #include <linux/gfp.h>
61 #include <linux/magic.h>
62 #include <linux/cgroup-defs.h>
64 #include <asm/processor.h>
66 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
69 * Extended scheduling parameters data structure.
71 * This is needed because the original struct sched_param can not be
72 * altered without introducing ABI issues with legacy applications
73 * (e.g., in sched_getparam()).
75 * However, the possibility of specifying more than just a priority for
76 * the tasks may be useful for a wide variety of application fields, e.g.,
77 * multimedia, streaming, automation and control, and many others.
79 * This variant (sched_attr) is meant at describing a so-called
80 * sporadic time-constrained task. In such model a task is specified by:
81 * - the activation period or minimum instance inter-arrival time;
82 * - the maximum (or average, depending on the actual scheduling
83 * discipline) computation time of all instances, a.k.a. runtime;
84 * - the deadline (relative to the actual activation time) of each
86 * Very briefly, a periodic (sporadic) task asks for the execution of
87 * some specific computation --which is typically called an instance--
88 * (at most) every period. Moreover, each instance typically lasts no more
89 * than the runtime and must be completed by time instant t equal to
90 * the instance activation time + the deadline.
92 * This is reflected by the actual fields of the sched_attr structure:
94 * @size size of the structure, for fwd/bwd compat.
96 * @sched_policy task's scheduling policy
97 * @sched_flags for customizing the scheduler behaviour
98 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
99 * @sched_priority task's static priority (SCHED_FIFO/RR)
100 * @sched_deadline representative of the task's deadline
101 * @sched_runtime representative of the task's runtime
102 * @sched_period representative of the task's period
104 * Given this task model, there are a multiplicity of scheduling algorithms
105 * and policies, that can be used to ensure all the tasks will make their
106 * timing constraints.
108 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
109 * only user of this new interface. More information about the algorithm
110 * available in the scheduling class file or in Documentation/.
118 /* SCHED_NORMAL, SCHED_BATCH */
121 /* SCHED_FIFO, SCHED_RR */
130 struct futex_pi_state
;
131 struct robust_list_head
;
134 struct perf_event_context
;
139 #define VMACACHE_BITS 2
140 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
141 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
144 * These are the constant used to fake the fixed-point load-average
145 * counting. Some notes:
146 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
147 * a load-average precision of 10 bits integer + 11 bits fractional
148 * - if you want to count load-averages more often, you need more
149 * precision, or rounding will get you. With 2-second counting freq,
150 * the EXP_n values would be 1981, 2034 and 2043 if still using only
153 extern unsigned long avenrun
[]; /* Load averages */
154 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
156 #define FSHIFT 11 /* nr of bits of precision */
157 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
158 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
159 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
160 #define EXP_5 2014 /* 1/exp(5sec/5min) */
161 #define EXP_15 2037 /* 1/exp(5sec/15min) */
163 #define CALC_LOAD(load,exp,n) \
165 load += n*(FIXED_1-exp); \
168 extern unsigned long total_forks
;
169 extern int nr_threads
;
170 DECLARE_PER_CPU(unsigned long, process_counts
);
171 extern int nr_processes(void);
172 extern unsigned long nr_running(void);
173 extern bool single_task_running(void);
174 extern unsigned long nr_iowait(void);
175 extern unsigned long nr_iowait_cpu(int cpu
);
176 extern void get_iowait_load(unsigned long *nr_waiters
, unsigned long *load
);
178 extern void calc_global_load(unsigned long ticks
);
180 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
181 extern void update_cpu_load_nohz(int active
);
183 static inline void update_cpu_load_nohz(int active
) { }
186 extern void dump_cpu_task(int cpu
);
191 #ifdef CONFIG_SCHED_DEBUG
192 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
193 extern void proc_sched_set_task(struct task_struct
*p
);
197 * Task state bitmask. NOTE! These bits are also
198 * encoded in fs/proc/array.c: get_task_state().
200 * We have two separate sets of flags: task->state
201 * is about runnability, while task->exit_state are
202 * about the task exiting. Confusing, but this way
203 * modifying one set can't modify the other one by
206 #define TASK_RUNNING 0
207 #define TASK_INTERRUPTIBLE 1
208 #define TASK_UNINTERRUPTIBLE 2
209 #define __TASK_STOPPED 4
210 #define __TASK_TRACED 8
211 /* in tsk->exit_state */
213 #define EXIT_ZOMBIE 32
214 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
215 /* in tsk->state again */
217 #define TASK_WAKEKILL 128
218 #define TASK_WAKING 256
219 #define TASK_PARKED 512
220 #define TASK_NOLOAD 1024
221 #define TASK_STATE_MAX 2048
223 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPN"
225 extern char ___assert_task_state
[1 - 2*!!(
226 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
228 /* Convenience macros for the sake of set_task_state */
229 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
230 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
231 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
233 #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
235 /* Convenience macros for the sake of wake_up */
236 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
237 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
239 /* get_task_state() */
240 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
241 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
242 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
244 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
245 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
246 #define task_is_stopped_or_traced(task) \
247 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
248 #define task_contributes_to_load(task) \
249 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
250 (task->flags & PF_FROZEN) == 0 && \
251 (task->state & TASK_NOLOAD) == 0)
253 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
255 #define __set_task_state(tsk, state_value) \
257 (tsk)->task_state_change = _THIS_IP_; \
258 (tsk)->state = (state_value); \
260 #define set_task_state(tsk, state_value) \
262 (tsk)->task_state_change = _THIS_IP_; \
263 smp_store_mb((tsk)->state, (state_value)); \
267 * set_current_state() includes a barrier so that the write of current->state
268 * is correctly serialised wrt the caller's subsequent test of whether to
271 * set_current_state(TASK_UNINTERRUPTIBLE);
272 * if (do_i_need_to_sleep())
275 * If the caller does not need such serialisation then use __set_current_state()
277 #define __set_current_state(state_value) \
279 current->task_state_change = _THIS_IP_; \
280 current->state = (state_value); \
282 #define set_current_state(state_value) \
284 current->task_state_change = _THIS_IP_; \
285 smp_store_mb(current->state, (state_value)); \
290 #define __set_task_state(tsk, state_value) \
291 do { (tsk)->state = (state_value); } while (0)
292 #define set_task_state(tsk, state_value) \
293 smp_store_mb((tsk)->state, (state_value))
296 * set_current_state() includes a barrier so that the write of current->state
297 * is correctly serialised wrt the caller's subsequent test of whether to
300 * set_current_state(TASK_UNINTERRUPTIBLE);
301 * if (do_i_need_to_sleep())
304 * If the caller does not need such serialisation then use __set_current_state()
306 #define __set_current_state(state_value) \
307 do { current->state = (state_value); } while (0)
308 #define set_current_state(state_value) \
309 smp_store_mb(current->state, (state_value))
313 /* Task command name length */
314 #define TASK_COMM_LEN 16
316 #include <linux/spinlock.h>
319 * This serializes "schedule()" and also protects
320 * the run-queue from deletions/modifications (but
321 * _adding_ to the beginning of the run-queue has
324 extern rwlock_t tasklist_lock
;
325 extern spinlock_t mmlist_lock
;
329 #ifdef CONFIG_PROVE_RCU
330 extern int lockdep_tasklist_lock_is_held(void);
331 #endif /* #ifdef CONFIG_PROVE_RCU */
333 extern void sched_init(void);
334 extern void sched_init_smp(void);
335 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
336 extern void init_idle(struct task_struct
*idle
, int cpu
);
337 extern void init_idle_bootup_task(struct task_struct
*idle
);
339 extern cpumask_var_t cpu_isolated_map
;
341 extern int runqueue_is_locked(int cpu
);
343 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
344 extern void nohz_balance_enter_idle(int cpu
);
345 extern void set_cpu_sd_state_idle(void);
346 extern int get_nohz_timer_target(void);
348 static inline void nohz_balance_enter_idle(int cpu
) { }
349 static inline void set_cpu_sd_state_idle(void) { }
353 * Only dump TASK_* tasks. (0 for all tasks)
355 extern void show_state_filter(unsigned long state_filter
);
357 static inline void show_state(void)
359 show_state_filter(0);
362 extern void show_regs(struct pt_regs
*);
365 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
366 * task), SP is the stack pointer of the first frame that should be shown in the back
367 * trace (or NULL if the entire call-chain of the task should be shown).
369 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
371 extern void cpu_init (void);
372 extern void trap_init(void);
373 extern void update_process_times(int user
);
374 extern void scheduler_tick(void);
376 extern void sched_show_task(struct task_struct
*p
);
378 #ifdef CONFIG_LOCKUP_DETECTOR
379 extern void touch_softlockup_watchdog_sched(void);
380 extern void touch_softlockup_watchdog(void);
381 extern void touch_softlockup_watchdog_sync(void);
382 extern void touch_all_softlockup_watchdogs(void);
383 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
385 size_t *lenp
, loff_t
*ppos
);
386 extern unsigned int softlockup_panic
;
387 extern unsigned int hardlockup_panic
;
388 void lockup_detector_init(void);
390 static inline void touch_softlockup_watchdog_sched(void)
393 static inline void touch_softlockup_watchdog(void)
396 static inline void touch_softlockup_watchdog_sync(void)
399 static inline void touch_all_softlockup_watchdogs(void)
402 static inline void lockup_detector_init(void)
407 #ifdef CONFIG_DETECT_HUNG_TASK
408 void reset_hung_task_detector(void);
410 static inline void reset_hung_task_detector(void)
415 /* Attach to any functions which should be ignored in wchan output. */
416 #define __sched __attribute__((__section__(".sched.text")))
418 /* Linker adds these: start and end of __sched functions */
419 extern char __sched_text_start
[], __sched_text_end
[];
421 /* Is this address in the __sched functions? */
422 extern int in_sched_functions(unsigned long addr
);
424 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
425 extern signed long schedule_timeout(signed long timeout
);
426 extern signed long schedule_timeout_interruptible(signed long timeout
);
427 extern signed long schedule_timeout_killable(signed long timeout
);
428 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
429 asmlinkage
void schedule(void);
430 extern void schedule_preempt_disabled(void);
432 extern long io_schedule_timeout(long timeout
);
434 static inline void io_schedule(void)
436 io_schedule_timeout(MAX_SCHEDULE_TIMEOUT
);
440 struct user_namespace
;
443 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
445 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
446 unsigned long, unsigned long);
448 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
449 unsigned long len
, unsigned long pgoff
,
450 unsigned long flags
);
452 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
455 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
456 #define SUID_DUMP_USER 1 /* Dump as user of process */
457 #define SUID_DUMP_ROOT 2 /* Dump as root */
461 /* for SUID_DUMP_* above */
462 #define MMF_DUMPABLE_BITS 2
463 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
465 extern void set_dumpable(struct mm_struct
*mm
, int value
);
467 * This returns the actual value of the suid_dumpable flag. For things
468 * that are using this for checking for privilege transitions, it must
469 * test against SUID_DUMP_USER rather than treating it as a boolean
472 static inline int __get_dumpable(unsigned long mm_flags
)
474 return mm_flags
& MMF_DUMPABLE_MASK
;
477 static inline int get_dumpable(struct mm_struct
*mm
)
479 return __get_dumpable(mm
->flags
);
482 /* coredump filter bits */
483 #define MMF_DUMP_ANON_PRIVATE 2
484 #define MMF_DUMP_ANON_SHARED 3
485 #define MMF_DUMP_MAPPED_PRIVATE 4
486 #define MMF_DUMP_MAPPED_SHARED 5
487 #define MMF_DUMP_ELF_HEADERS 6
488 #define MMF_DUMP_HUGETLB_PRIVATE 7
489 #define MMF_DUMP_HUGETLB_SHARED 8
490 #define MMF_DUMP_DAX_PRIVATE 9
491 #define MMF_DUMP_DAX_SHARED 10
493 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
494 #define MMF_DUMP_FILTER_BITS 9
495 #define MMF_DUMP_FILTER_MASK \
496 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
497 #define MMF_DUMP_FILTER_DEFAULT \
498 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
499 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
501 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
502 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
504 # define MMF_DUMP_MASK_DEFAULT_ELF 0
506 /* leave room for more dump flags */
507 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
508 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
509 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
511 #define MMF_HAS_UPROBES 19 /* has uprobes */
512 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
514 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
516 struct sighand_struct
{
518 struct k_sigaction action
[_NSIG
];
520 wait_queue_head_t signalfd_wqh
;
523 struct pacct_struct
{
526 unsigned long ac_mem
;
527 cputime_t ac_utime
, ac_stime
;
528 unsigned long ac_minflt
, ac_majflt
;
539 * struct prev_cputime - snaphsot of system and user cputime
540 * @utime: time spent in user mode
541 * @stime: time spent in system mode
542 * @lock: protects the above two fields
544 * Stores previous user/system time values such that we can guarantee
547 struct prev_cputime
{
548 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
555 static inline void prev_cputime_init(struct prev_cputime
*prev
)
557 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
558 prev
->utime
= prev
->stime
= 0;
559 raw_spin_lock_init(&prev
->lock
);
564 * struct task_cputime - collected CPU time counts
565 * @utime: time spent in user mode, in &cputime_t units
566 * @stime: time spent in kernel mode, in &cputime_t units
567 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
569 * This structure groups together three kinds of CPU time that are tracked for
570 * threads and thread groups. Most things considering CPU time want to group
571 * these counts together and treat all three of them in parallel.
573 struct task_cputime
{
576 unsigned long long sum_exec_runtime
;
579 /* Alternate field names when used to cache expirations. */
580 #define virt_exp utime
581 #define prof_exp stime
582 #define sched_exp sum_exec_runtime
584 #define INIT_CPUTIME \
585 (struct task_cputime) { \
588 .sum_exec_runtime = 0, \
592 * This is the atomic variant of task_cputime, which can be used for
593 * storing and updating task_cputime statistics without locking.
595 struct task_cputime_atomic
{
598 atomic64_t sum_exec_runtime
;
601 #define INIT_CPUTIME_ATOMIC \
602 (struct task_cputime_atomic) { \
603 .utime = ATOMIC64_INIT(0), \
604 .stime = ATOMIC64_INIT(0), \
605 .sum_exec_runtime = ATOMIC64_INIT(0), \
608 #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
611 * Disable preemption until the scheduler is running -- use an unconditional
612 * value so that it also works on !PREEMPT_COUNT kernels.
614 * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
616 #define INIT_PREEMPT_COUNT PREEMPT_OFFSET
619 * Initial preempt_count value; reflects the preempt_count schedule invariant
620 * which states that during context switches:
622 * preempt_count() == 2*PREEMPT_DISABLE_OFFSET
624 * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
625 * Note: See finish_task_switch().
627 #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
630 * struct thread_group_cputimer - thread group interval timer counts
631 * @cputime_atomic: atomic thread group interval timers.
632 * @running: true when there are timers running and
633 * @cputime_atomic receives updates.
634 * @checking_timer: true when a thread in the group is in the
635 * process of checking for thread group timers.
637 * This structure contains the version of task_cputime, above, that is
638 * used for thread group CPU timer calculations.
640 struct thread_group_cputimer
{
641 struct task_cputime_atomic cputime_atomic
;
646 #include <linux/rwsem.h>
650 * NOTE! "signal_struct" does not have its own
651 * locking, because a shared signal_struct always
652 * implies a shared sighand_struct, so locking
653 * sighand_struct is always a proper superset of
654 * the locking of signal_struct.
656 struct signal_struct
{
660 struct list_head thread_head
;
662 wait_queue_head_t wait_chldexit
; /* for wait4() */
664 /* current thread group signal load-balancing target: */
665 struct task_struct
*curr_target
;
667 /* shared signal handling: */
668 struct sigpending shared_pending
;
670 /* thread group exit support */
673 * - notify group_exit_task when ->count is equal to notify_count
674 * - everyone except group_exit_task is stopped during signal delivery
675 * of fatal signals, group_exit_task processes the signal.
678 struct task_struct
*group_exit_task
;
680 /* thread group stop support, overloads group_exit_code too */
681 int group_stop_count
;
682 unsigned int flags
; /* see SIGNAL_* flags below */
685 * PR_SET_CHILD_SUBREAPER marks a process, like a service
686 * manager, to re-parent orphan (double-forking) child processes
687 * to this process instead of 'init'. The service manager is
688 * able to receive SIGCHLD signals and is able to investigate
689 * the process until it calls wait(). All children of this
690 * process will inherit a flag if they should look for a
691 * child_subreaper process at exit.
693 unsigned int is_child_subreaper
:1;
694 unsigned int has_child_subreaper
:1;
696 /* POSIX.1b Interval Timers */
698 struct list_head posix_timers
;
700 /* ITIMER_REAL timer for the process */
701 struct hrtimer real_timer
;
702 struct pid
*leader_pid
;
703 ktime_t it_real_incr
;
706 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
707 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
708 * values are defined to 0 and 1 respectively
710 struct cpu_itimer it
[2];
713 * Thread group totals for process CPU timers.
714 * See thread_group_cputimer(), et al, for details.
716 struct thread_group_cputimer cputimer
;
718 /* Earliest-expiration cache. */
719 struct task_cputime cputime_expires
;
721 #ifdef CONFIG_NO_HZ_FULL
722 unsigned long tick_dep_mask
;
725 struct list_head cpu_timers
[3];
727 struct pid
*tty_old_pgrp
;
729 /* boolean value for session group leader */
732 struct tty_struct
*tty
; /* NULL if no tty */
734 #ifdef CONFIG_SCHED_AUTOGROUP
735 struct autogroup
*autogroup
;
738 * Cumulative resource counters for dead threads in the group,
739 * and for reaped dead child processes forked by this group.
740 * Live threads maintain their own counters and add to these
741 * in __exit_signal, except for the group leader.
743 seqlock_t stats_lock
;
744 cputime_t utime
, stime
, cutime
, cstime
;
747 struct prev_cputime prev_cputime
;
748 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
749 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
750 unsigned long inblock
, oublock
, cinblock
, coublock
;
751 unsigned long maxrss
, cmaxrss
;
752 struct task_io_accounting ioac
;
755 * Cumulative ns of schedule CPU time fo dead threads in the
756 * group, not including a zombie group leader, (This only differs
757 * from jiffies_to_ns(utime + stime) if sched_clock uses something
758 * other than jiffies.)
760 unsigned long long sum_sched_runtime
;
763 * We don't bother to synchronize most readers of this at all,
764 * because there is no reader checking a limit that actually needs
765 * to get both rlim_cur and rlim_max atomically, and either one
766 * alone is a single word that can safely be read normally.
767 * getrlimit/setrlimit use task_lock(current->group_leader) to
768 * protect this instead of the siglock, because they really
769 * have no need to disable irqs.
771 struct rlimit rlim
[RLIM_NLIMITS
];
773 #ifdef CONFIG_BSD_PROCESS_ACCT
774 struct pacct_struct pacct
; /* per-process accounting information */
776 #ifdef CONFIG_TASKSTATS
777 struct taskstats
*stats
;
781 struct tty_audit_buf
*tty_audit_buf
;
784 oom_flags_t oom_flags
;
785 short oom_score_adj
; /* OOM kill score adjustment */
786 short oom_score_adj_min
; /* OOM kill score adjustment min value.
787 * Only settable by CAP_SYS_RESOURCE. */
789 struct mutex cred_guard_mutex
; /* guard against foreign influences on
790 * credential calculations
791 * (notably. ptrace) */
795 * Bits in flags field of signal_struct.
797 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
798 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
799 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
800 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
802 * Pending notifications to parent.
804 #define SIGNAL_CLD_STOPPED 0x00000010
805 #define SIGNAL_CLD_CONTINUED 0x00000020
806 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
808 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
810 /* If true, all threads except ->group_exit_task have pending SIGKILL */
811 static inline int signal_group_exit(const struct signal_struct
*sig
)
813 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
814 (sig
->group_exit_task
!= NULL
);
818 * Some day this will be a full-fledged user tracking system..
821 atomic_t __count
; /* reference count */
822 atomic_t processes
; /* How many processes does this user have? */
823 atomic_t sigpending
; /* How many pending signals does this user have? */
824 #ifdef CONFIG_INOTIFY_USER
825 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
826 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
828 #ifdef CONFIG_FANOTIFY
829 atomic_t fanotify_listeners
;
832 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
834 #ifdef CONFIG_POSIX_MQUEUE
835 /* protected by mq_lock */
836 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
838 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
839 unsigned long unix_inflight
; /* How many files in flight in unix sockets */
840 atomic_long_t pipe_bufs
; /* how many pages are allocated in pipe buffers */
843 struct key
*uid_keyring
; /* UID specific keyring */
844 struct key
*session_keyring
; /* UID's default session keyring */
847 /* Hash table maintenance information */
848 struct hlist_node uidhash_node
;
851 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
852 atomic_long_t locked_vm
;
856 extern int uids_sysfs_init(void);
858 extern struct user_struct
*find_user(kuid_t
);
860 extern struct user_struct root_user
;
861 #define INIT_USER (&root_user)
864 struct backing_dev_info
;
865 struct reclaim_state
;
867 #ifdef CONFIG_SCHED_INFO
869 /* cumulative counters */
870 unsigned long pcount
; /* # of times run on this cpu */
871 unsigned long long run_delay
; /* time spent waiting on a runqueue */
874 unsigned long long last_arrival
,/* when we last ran on a cpu */
875 last_queued
; /* when we were last queued to run */
877 #endif /* CONFIG_SCHED_INFO */
879 #ifdef CONFIG_TASK_DELAY_ACCT
880 struct task_delay_info
{
882 unsigned int flags
; /* Private per-task flags */
884 /* For each stat XXX, add following, aligned appropriately
886 * struct timespec XXX_start, XXX_end;
890 * Atomicity of updates to XXX_delay, XXX_count protected by
891 * single lock above (split into XXX_lock if contention is an issue).
895 * XXX_count is incremented on every XXX operation, the delay
896 * associated with the operation is added to XXX_delay.
897 * XXX_delay contains the accumulated delay time in nanoseconds.
899 u64 blkio_start
; /* Shared by blkio, swapin */
900 u64 blkio_delay
; /* wait for sync block io completion */
901 u64 swapin_delay
; /* wait for swapin block io completion */
902 u32 blkio_count
; /* total count of the number of sync block */
903 /* io operations performed */
904 u32 swapin_count
; /* total count of the number of swapin block */
905 /* io operations performed */
908 u64 freepages_delay
; /* wait for memory reclaim */
909 u32 freepages_count
; /* total count of memory reclaim */
911 #endif /* CONFIG_TASK_DELAY_ACCT */
913 static inline int sched_info_on(void)
915 #ifdef CONFIG_SCHEDSTATS
917 #elif defined(CONFIG_TASK_DELAY_ACCT)
918 extern int delayacct_on
;
925 #ifdef CONFIG_SCHEDSTATS
926 void force_schedstat_enabled(void);
937 * Increase resolution of cpu_capacity calculations
939 #define SCHED_CAPACITY_SHIFT 10
940 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
943 * Wake-queues are lists of tasks with a pending wakeup, whose
944 * callers have already marked the task as woken internally,
945 * and can thus carry on. A common use case is being able to
946 * do the wakeups once the corresponding user lock as been
949 * We hold reference to each task in the list across the wakeup,
950 * thus guaranteeing that the memory is still valid by the time
951 * the actual wakeups are performed in wake_up_q().
953 * One per task suffices, because there's never a need for a task to be
954 * in two wake queues simultaneously; it is forbidden to abandon a task
955 * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
956 * already in a wake queue, the wakeup will happen soon and the second
957 * waker can just skip it.
959 * The WAKE_Q macro declares and initializes the list head.
960 * wake_up_q() does NOT reinitialize the list; it's expected to be
961 * called near the end of a function, where the fact that the queue is
962 * not used again will be easy to see by inspection.
964 * Note that this can cause spurious wakeups. schedule() callers
965 * must ensure the call is done inside a loop, confirming that the
966 * wakeup condition has in fact occurred.
969 struct wake_q_node
*next
;
973 struct wake_q_node
*first
;
974 struct wake_q_node
**lastp
;
977 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
979 #define WAKE_Q(name) \
980 struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
982 extern void wake_q_add(struct wake_q_head
*head
,
983 struct task_struct
*task
);
984 extern void wake_up_q(struct wake_q_head
*head
);
987 * sched-domains (multiprocessor balancing) declarations:
990 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
991 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
992 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
993 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
994 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
995 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
996 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
997 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
998 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
999 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
1000 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
1001 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
1002 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
1003 #define SD_NUMA 0x4000 /* cross-node balancing */
1005 #ifdef CONFIG_SCHED_SMT
1006 static inline int cpu_smt_flags(void)
1008 return SD_SHARE_CPUCAPACITY
| SD_SHARE_PKG_RESOURCES
;
1012 #ifdef CONFIG_SCHED_MC
1013 static inline int cpu_core_flags(void)
1015 return SD_SHARE_PKG_RESOURCES
;
1020 static inline int cpu_numa_flags(void)
1026 struct sched_domain_attr
{
1027 int relax_domain_level
;
1030 #define SD_ATTR_INIT (struct sched_domain_attr) { \
1031 .relax_domain_level = -1, \
1034 extern int sched_domain_level_max
;
1038 struct sched_domain
{
1039 /* These fields must be setup */
1040 struct sched_domain
*parent
; /* top domain must be null terminated */
1041 struct sched_domain
*child
; /* bottom domain must be null terminated */
1042 struct sched_group
*groups
; /* the balancing groups of the domain */
1043 unsigned long min_interval
; /* Minimum balance interval ms */
1044 unsigned long max_interval
; /* Maximum balance interval ms */
1045 unsigned int busy_factor
; /* less balancing by factor if busy */
1046 unsigned int imbalance_pct
; /* No balance until over watermark */
1047 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
1048 unsigned int busy_idx
;
1049 unsigned int idle_idx
;
1050 unsigned int newidle_idx
;
1051 unsigned int wake_idx
;
1052 unsigned int forkexec_idx
;
1053 unsigned int smt_gain
;
1055 int nohz_idle
; /* NOHZ IDLE status */
1056 int flags
; /* See SD_* */
1059 /* Runtime fields. */
1060 unsigned long last_balance
; /* init to jiffies. units in jiffies */
1061 unsigned int balance_interval
; /* initialise to 1. units in ms. */
1062 unsigned int nr_balance_failed
; /* initialise to 0 */
1064 /* idle_balance() stats */
1065 u64 max_newidle_lb_cost
;
1066 unsigned long next_decay_max_lb_cost
;
1068 #ifdef CONFIG_SCHEDSTATS
1069 /* load_balance() stats */
1070 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
1071 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
1072 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
1073 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
1074 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
1075 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
1076 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
1077 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
1079 /* Active load balancing */
1080 unsigned int alb_count
;
1081 unsigned int alb_failed
;
1082 unsigned int alb_pushed
;
1084 /* SD_BALANCE_EXEC stats */
1085 unsigned int sbe_count
;
1086 unsigned int sbe_balanced
;
1087 unsigned int sbe_pushed
;
1089 /* SD_BALANCE_FORK stats */
1090 unsigned int sbf_count
;
1091 unsigned int sbf_balanced
;
1092 unsigned int sbf_pushed
;
1094 /* try_to_wake_up() stats */
1095 unsigned int ttwu_wake_remote
;
1096 unsigned int ttwu_move_affine
;
1097 unsigned int ttwu_move_balance
;
1099 #ifdef CONFIG_SCHED_DEBUG
1103 void *private; /* used during construction */
1104 struct rcu_head rcu
; /* used during destruction */
1107 unsigned int span_weight
;
1109 * Span of all CPUs in this domain.
1111 * NOTE: this field is variable length. (Allocated dynamically
1112 * by attaching extra space to the end of the structure,
1113 * depending on how many CPUs the kernel has booted up with)
1115 unsigned long span
[0];
1118 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1120 return to_cpumask(sd
->span
);
1123 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1124 struct sched_domain_attr
*dattr_new
);
1126 /* Allocate an array of sched domains, for partition_sched_domains(). */
1127 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1128 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1130 bool cpus_share_cache(int this_cpu
, int that_cpu
);
1132 typedef const struct cpumask
*(*sched_domain_mask_f
)(int cpu
);
1133 typedef int (*sched_domain_flags_f
)(void);
1135 #define SDTL_OVERLAP 0x01
1138 struct sched_domain
**__percpu sd
;
1139 struct sched_group
**__percpu sg
;
1140 struct sched_group_capacity
**__percpu sgc
;
1143 struct sched_domain_topology_level
{
1144 sched_domain_mask_f mask
;
1145 sched_domain_flags_f sd_flags
;
1148 struct sd_data data
;
1149 #ifdef CONFIG_SCHED_DEBUG
1154 extern void set_sched_topology(struct sched_domain_topology_level
*tl
);
1155 extern void wake_up_if_idle(int cpu
);
1157 #ifdef CONFIG_SCHED_DEBUG
1158 # define SD_INIT_NAME(type) .name = #type
1160 # define SD_INIT_NAME(type)
1163 #else /* CONFIG_SMP */
1165 struct sched_domain_attr
;
1168 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1169 struct sched_domain_attr
*dattr_new
)
1173 static inline bool cpus_share_cache(int this_cpu
, int that_cpu
)
1178 #endif /* !CONFIG_SMP */
1181 struct io_context
; /* See blkdev.h */
1184 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1185 extern void prefetch_stack(struct task_struct
*t
);
1187 static inline void prefetch_stack(struct task_struct
*t
) { }
1190 struct audit_context
; /* See audit.c */
1192 struct pipe_inode_info
;
1193 struct uts_namespace
;
1195 struct load_weight
{
1196 unsigned long weight
;
1201 * The load_avg/util_avg accumulates an infinite geometric series.
1202 * 1) load_avg factors frequency scaling into the amount of time that a
1203 * sched_entity is runnable on a rq into its weight. For cfs_rq, it is the
1204 * aggregated such weights of all runnable and blocked sched_entities.
1205 * 2) util_avg factors frequency and cpu scaling into the amount of time
1206 * that a sched_entity is running on a CPU, in the range [0..SCHED_LOAD_SCALE].
1207 * For cfs_rq, it is the aggregated such times of all runnable and
1208 * blocked sched_entities.
1209 * The 64 bit load_sum can:
1210 * 1) for cfs_rq, afford 4353082796 (=2^64/47742/88761) entities with
1211 * the highest weight (=88761) always runnable, we should not overflow
1212 * 2) for entity, support any load.weight always runnable
1215 u64 last_update_time
, load_sum
;
1216 u32 util_sum
, period_contrib
;
1217 unsigned long load_avg
, util_avg
;
1220 #ifdef CONFIG_SCHEDSTATS
1221 struct sched_statistics
{
1231 s64 sum_sleep_runtime
;
1238 u64 nr_migrations_cold
;
1239 u64 nr_failed_migrations_affine
;
1240 u64 nr_failed_migrations_running
;
1241 u64 nr_failed_migrations_hot
;
1242 u64 nr_forced_migrations
;
1245 u64 nr_wakeups_sync
;
1246 u64 nr_wakeups_migrate
;
1247 u64 nr_wakeups_local
;
1248 u64 nr_wakeups_remote
;
1249 u64 nr_wakeups_affine
;
1250 u64 nr_wakeups_affine_attempts
;
1251 u64 nr_wakeups_passive
;
1252 u64 nr_wakeups_idle
;
1256 struct sched_entity
{
1257 struct load_weight load
; /* for load-balancing */
1258 struct rb_node run_node
;
1259 struct list_head group_node
;
1263 u64 sum_exec_runtime
;
1265 u64 prev_sum_exec_runtime
;
1269 #ifdef CONFIG_SCHEDSTATS
1270 struct sched_statistics statistics
;
1273 #ifdef CONFIG_FAIR_GROUP_SCHED
1275 struct sched_entity
*parent
;
1276 /* rq on which this entity is (to be) queued: */
1277 struct cfs_rq
*cfs_rq
;
1278 /* rq "owned" by this entity/group: */
1279 struct cfs_rq
*my_q
;
1284 * Per entity load average tracking.
1286 * Put into separate cache line so it does not
1287 * collide with read-mostly values above.
1289 struct sched_avg avg ____cacheline_aligned_in_smp
;
1293 struct sched_rt_entity
{
1294 struct list_head run_list
;
1295 unsigned long timeout
;
1296 unsigned long watchdog_stamp
;
1297 unsigned int time_slice
;
1298 unsigned short on_rq
;
1299 unsigned short on_list
;
1301 struct sched_rt_entity
*back
;
1302 #ifdef CONFIG_RT_GROUP_SCHED
1303 struct sched_rt_entity
*parent
;
1304 /* rq on which this entity is (to be) queued: */
1305 struct rt_rq
*rt_rq
;
1306 /* rq "owned" by this entity/group: */
1311 struct sched_dl_entity
{
1312 struct rb_node rb_node
;
1315 * Original scheduling parameters. Copied here from sched_attr
1316 * during sched_setattr(), they will remain the same until
1317 * the next sched_setattr().
1319 u64 dl_runtime
; /* maximum runtime for each instance */
1320 u64 dl_deadline
; /* relative deadline of each instance */
1321 u64 dl_period
; /* separation of two instances (period) */
1322 u64 dl_bw
; /* dl_runtime / dl_deadline */
1325 * Actual scheduling parameters. Initialized with the values above,
1326 * they are continously updated during task execution. Note that
1327 * the remaining runtime could be < 0 in case we are in overrun.
1329 s64 runtime
; /* remaining runtime for this instance */
1330 u64 deadline
; /* absolute deadline for this instance */
1331 unsigned int flags
; /* specifying the scheduler behaviour */
1336 * @dl_throttled tells if we exhausted the runtime. If so, the
1337 * task has to wait for a replenishment to be performed at the
1338 * next firing of dl_timer.
1340 * @dl_boosted tells if we are boosted due to DI. If so we are
1341 * outside bandwidth enforcement mechanism (but only until we
1342 * exit the critical section);
1344 * @dl_yielded tells if task gave up the cpu before consuming
1345 * all its available runtime during the last job.
1347 int dl_throttled
, dl_boosted
, dl_yielded
;
1350 * Bandwidth enforcement timer. Each -deadline task has its
1351 * own bandwidth to be enforced, thus we need one timer per task.
1353 struct hrtimer dl_timer
;
1361 u8 pad
; /* Otherwise the compiler can store garbage here. */
1363 u32 s
; /* Set of bits. */
1367 enum perf_event_task_context
{
1368 perf_invalid_context
= -1,
1369 perf_hw_context
= 0,
1371 perf_nr_task_contexts
,
1374 /* Track pages that require TLB flushes */
1375 struct tlbflush_unmap_batch
{
1377 * Each bit set is a CPU that potentially has a TLB entry for one of
1378 * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1380 struct cpumask cpumask
;
1382 /* True if any bit in cpumask is set */
1383 bool flush_required
;
1386 * If true then the PTE was dirty when unmapped. The entry must be
1387 * flushed before IO is initiated or a stale TLB entry potentially
1388 * allows an update without redirtying the page.
1393 struct task_struct
{
1394 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1397 unsigned int flags
; /* per process flags, defined below */
1398 unsigned int ptrace
;
1401 struct llist_node wake_entry
;
1403 unsigned int wakee_flips
;
1404 unsigned long wakee_flip_decay_ts
;
1405 struct task_struct
*last_wakee
;
1411 int prio
, static_prio
, normal_prio
;
1412 unsigned int rt_priority
;
1413 const struct sched_class
*sched_class
;
1414 struct sched_entity se
;
1415 struct sched_rt_entity rt
;
1416 #ifdef CONFIG_CGROUP_SCHED
1417 struct task_group
*sched_task_group
;
1419 struct sched_dl_entity dl
;
1421 #ifdef CONFIG_PREEMPT_NOTIFIERS
1422 /* list of struct preempt_notifier: */
1423 struct hlist_head preempt_notifiers
;
1426 #ifdef CONFIG_BLK_DEV_IO_TRACE
1427 unsigned int btrace_seq
;
1430 unsigned int policy
;
1431 int nr_cpus_allowed
;
1432 cpumask_t cpus_allowed
;
1434 #ifdef CONFIG_PREEMPT_RCU
1435 int rcu_read_lock_nesting
;
1436 union rcu_special rcu_read_unlock_special
;
1437 struct list_head rcu_node_entry
;
1438 struct rcu_node
*rcu_blocked_node
;
1439 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1440 #ifdef CONFIG_TASKS_RCU
1441 unsigned long rcu_tasks_nvcsw
;
1442 bool rcu_tasks_holdout
;
1443 struct list_head rcu_tasks_holdout_list
;
1444 int rcu_tasks_idle_cpu
;
1445 #endif /* #ifdef CONFIG_TASKS_RCU */
1447 #ifdef CONFIG_SCHED_INFO
1448 struct sched_info sched_info
;
1451 struct list_head tasks
;
1453 struct plist_node pushable_tasks
;
1454 struct rb_node pushable_dl_tasks
;
1457 struct mm_struct
*mm
, *active_mm
;
1458 /* per-thread vma caching */
1459 u32 vmacache_seqnum
;
1460 struct vm_area_struct
*vmacache
[VMACACHE_SIZE
];
1461 #if defined(SPLIT_RSS_COUNTING)
1462 struct task_rss_stat rss_stat
;
1466 int exit_code
, exit_signal
;
1467 int pdeath_signal
; /* The signal sent when the parent dies */
1468 unsigned long jobctl
; /* JOBCTL_*, siglock protected */
1470 /* Used for emulating ABI behavior of previous Linux versions */
1471 unsigned int personality
;
1473 /* scheduler bits, serialized by scheduler locks */
1474 unsigned sched_reset_on_fork
:1;
1475 unsigned sched_contributes_to_load
:1;
1476 unsigned sched_migrated
:1;
1477 unsigned :0; /* force alignment to the next boundary */
1479 /* unserialized, strictly 'current' */
1480 unsigned in_execve
:1; /* bit to tell LSMs we're in execve */
1481 unsigned in_iowait
:1;
1483 unsigned memcg_may_oom
:1;
1485 unsigned memcg_kmem_skip_account
:1;
1488 #ifdef CONFIG_COMPAT_BRK
1489 unsigned brk_randomized
:1;
1492 unsigned long atomic_flags
; /* Flags needing atomic access. */
1494 struct restart_block restart_block
;
1499 #ifdef CONFIG_CC_STACKPROTECTOR
1500 /* Canary value for the -fstack-protector gcc feature */
1501 unsigned long stack_canary
;
1504 * pointers to (original) parent process, youngest child, younger sibling,
1505 * older sibling, respectively. (p->father can be replaced with
1506 * p->real_parent->pid)
1508 struct task_struct __rcu
*real_parent
; /* real parent process */
1509 struct task_struct __rcu
*parent
; /* recipient of SIGCHLD, wait4() reports */
1511 * children/sibling forms the list of my natural children
1513 struct list_head children
; /* list of my children */
1514 struct list_head sibling
; /* linkage in my parent's children list */
1515 struct task_struct
*group_leader
; /* threadgroup leader */
1518 * ptraced is the list of tasks this task is using ptrace on.
1519 * This includes both natural children and PTRACE_ATTACH targets.
1520 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1522 struct list_head ptraced
;
1523 struct list_head ptrace_entry
;
1525 /* PID/PID hash table linkage. */
1526 struct pid_link pids
[PIDTYPE_MAX
];
1527 struct list_head thread_group
;
1528 struct list_head thread_node
;
1530 struct completion
*vfork_done
; /* for vfork() */
1531 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1532 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1534 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1536 struct prev_cputime prev_cputime
;
1537 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1538 seqcount_t vtime_seqcount
;
1539 unsigned long long vtime_snap
;
1541 /* Task is sleeping or running in a CPU with VTIME inactive */
1543 /* Task runs in userspace in a CPU with VTIME active */
1545 /* Task runs in kernelspace in a CPU with VTIME active */
1547 } vtime_snap_whence
;
1550 #ifdef CONFIG_NO_HZ_FULL
1551 unsigned long tick_dep_mask
;
1553 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1554 u64 start_time
; /* monotonic time in nsec */
1555 u64 real_start_time
; /* boot based time in nsec */
1556 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1557 unsigned long min_flt
, maj_flt
;
1559 struct task_cputime cputime_expires
;
1560 struct list_head cpu_timers
[3];
1562 /* process credentials */
1563 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1564 * credentials (COW) */
1565 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1566 * credentials (COW) */
1567 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1568 - access with [gs]et_task_comm (which lock
1569 it with task_lock())
1570 - initialized normally by setup_new_exec */
1571 /* file system info */
1572 struct nameidata
*nameidata
;
1573 #ifdef CONFIG_SYSVIPC
1575 struct sysv_sem sysvsem
;
1576 struct sysv_shm sysvshm
;
1578 #ifdef CONFIG_DETECT_HUNG_TASK
1579 /* hung task detection */
1580 unsigned long last_switch_count
;
1582 /* filesystem information */
1583 struct fs_struct
*fs
;
1584 /* open file information */
1585 struct files_struct
*files
;
1587 struct nsproxy
*nsproxy
;
1588 /* signal handlers */
1589 struct signal_struct
*signal
;
1590 struct sighand_struct
*sighand
;
1592 sigset_t blocked
, real_blocked
;
1593 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1594 struct sigpending pending
;
1596 unsigned long sas_ss_sp
;
1599 struct callback_head
*task_works
;
1601 struct audit_context
*audit_context
;
1602 #ifdef CONFIG_AUDITSYSCALL
1604 unsigned int sessionid
;
1606 struct seccomp seccomp
;
1608 /* Thread group tracking */
1611 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1613 spinlock_t alloc_lock
;
1615 /* Protection of the PI data structures: */
1616 raw_spinlock_t pi_lock
;
1618 struct wake_q_node wake_q
;
1620 #ifdef CONFIG_RT_MUTEXES
1621 /* PI waiters blocked on a rt_mutex held by this task */
1622 struct rb_root pi_waiters
;
1623 struct rb_node
*pi_waiters_leftmost
;
1624 /* Deadlock detection and priority inheritance handling */
1625 struct rt_mutex_waiter
*pi_blocked_on
;
1628 #ifdef CONFIG_DEBUG_MUTEXES
1629 /* mutex deadlock detection */
1630 struct mutex_waiter
*blocked_on
;
1632 #ifdef CONFIG_TRACE_IRQFLAGS
1633 unsigned int irq_events
;
1634 unsigned long hardirq_enable_ip
;
1635 unsigned long hardirq_disable_ip
;
1636 unsigned int hardirq_enable_event
;
1637 unsigned int hardirq_disable_event
;
1638 int hardirqs_enabled
;
1639 int hardirq_context
;
1640 unsigned long softirq_disable_ip
;
1641 unsigned long softirq_enable_ip
;
1642 unsigned int softirq_disable_event
;
1643 unsigned int softirq_enable_event
;
1644 int softirqs_enabled
;
1645 int softirq_context
;
1647 #ifdef CONFIG_LOCKDEP
1648 # define MAX_LOCK_DEPTH 48UL
1651 unsigned int lockdep_recursion
;
1652 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1653 gfp_t lockdep_reclaim_gfp
;
1656 unsigned int in_ubsan
;
1659 /* journalling filesystem info */
1662 /* stacked block device info */
1663 struct bio_list
*bio_list
;
1666 /* stack plugging */
1667 struct blk_plug
*plug
;
1671 struct reclaim_state
*reclaim_state
;
1673 struct backing_dev_info
*backing_dev_info
;
1675 struct io_context
*io_context
;
1677 unsigned long ptrace_message
;
1678 siginfo_t
*last_siginfo
; /* For ptrace use. */
1679 struct task_io_accounting ioac
;
1680 #if defined(CONFIG_TASK_XACCT)
1681 u64 acct_rss_mem1
; /* accumulated rss usage */
1682 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1683 cputime_t acct_timexpd
; /* stime + utime since last update */
1685 #ifdef CONFIG_CPUSETS
1686 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1687 seqcount_t mems_allowed_seq
; /* Seqence no to catch updates */
1688 int cpuset_mem_spread_rotor
;
1689 int cpuset_slab_spread_rotor
;
1691 #ifdef CONFIG_CGROUPS
1692 /* Control Group info protected by css_set_lock */
1693 struct css_set __rcu
*cgroups
;
1694 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1695 struct list_head cg_list
;
1698 struct robust_list_head __user
*robust_list
;
1699 #ifdef CONFIG_COMPAT
1700 struct compat_robust_list_head __user
*compat_robust_list
;
1702 struct list_head pi_state_list
;
1703 struct futex_pi_state
*pi_state_cache
;
1705 #ifdef CONFIG_PERF_EVENTS
1706 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1707 struct mutex perf_event_mutex
;
1708 struct list_head perf_event_list
;
1710 #ifdef CONFIG_DEBUG_PREEMPT
1711 unsigned long preempt_disable_ip
;
1714 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1716 short pref_node_fork
;
1718 #ifdef CONFIG_NUMA_BALANCING
1720 unsigned int numa_scan_period
;
1721 unsigned int numa_scan_period_max
;
1722 int numa_preferred_nid
;
1723 unsigned long numa_migrate_retry
;
1724 u64 node_stamp
; /* migration stamp */
1725 u64 last_task_numa_placement
;
1726 u64 last_sum_exec_runtime
;
1727 struct callback_head numa_work
;
1729 struct list_head numa_entry
;
1730 struct numa_group
*numa_group
;
1733 * numa_faults is an array split into four regions:
1734 * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1735 * in this precise order.
1737 * faults_memory: Exponential decaying average of faults on a per-node
1738 * basis. Scheduling placement decisions are made based on these
1739 * counts. The values remain static for the duration of a PTE scan.
1740 * faults_cpu: Track the nodes the process was running on when a NUMA
1741 * hinting fault was incurred.
1742 * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1743 * during the current scan window. When the scan completes, the counts
1744 * in faults_memory and faults_cpu decay and these values are copied.
1746 unsigned long *numa_faults
;
1747 unsigned long total_numa_faults
;
1750 * numa_faults_locality tracks if faults recorded during the last
1751 * scan window were remote/local or failed to migrate. The task scan
1752 * period is adapted based on the locality of the faults with different
1753 * weights depending on whether they were shared or private faults
1755 unsigned long numa_faults_locality
[3];
1757 unsigned long numa_pages_migrated
;
1758 #endif /* CONFIG_NUMA_BALANCING */
1760 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1761 struct tlbflush_unmap_batch tlb_ubc
;
1764 struct rcu_head rcu
;
1767 * cache last used pipe for splice
1769 struct pipe_inode_info
*splice_pipe
;
1771 struct page_frag task_frag
;
1773 #ifdef CONFIG_TASK_DELAY_ACCT
1774 struct task_delay_info
*delays
;
1776 #ifdef CONFIG_FAULT_INJECTION
1780 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1781 * balance_dirty_pages() for some dirty throttling pause
1784 int nr_dirtied_pause
;
1785 unsigned long dirty_paused_when
; /* start of a write-and-pause period */
1787 #ifdef CONFIG_LATENCYTOP
1788 int latency_record_count
;
1789 struct latency_record latency_record
[LT_SAVECOUNT
];
1792 * time slack values; these are used to round up poll() and
1793 * select() etc timeout values. These are in nanoseconds.
1796 u64 default_timer_slack_ns
;
1799 unsigned int kasan_depth
;
1801 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1802 /* Index of current stored address in ret_stack */
1804 /* Stack of return addresses for return function tracing */
1805 struct ftrace_ret_stack
*ret_stack
;
1806 /* time stamp for last schedule */
1807 unsigned long long ftrace_timestamp
;
1809 * Number of functions that haven't been traced
1810 * because of depth overrun.
1812 atomic_t trace_overrun
;
1813 /* Pause for the tracing */
1814 atomic_t tracing_graph_pause
;
1816 #ifdef CONFIG_TRACING
1817 /* state flags for use by tracers */
1818 unsigned long trace
;
1819 /* bitmask and counter of trace recursion */
1820 unsigned long trace_recursion
;
1821 #endif /* CONFIG_TRACING */
1823 /* Coverage collection mode enabled for this task (0 if disabled). */
1824 enum kcov_mode kcov_mode
;
1825 /* Size of the kcov_area. */
1827 /* Buffer for coverage collection. */
1829 /* kcov desciptor wired with this task or NULL. */
1833 struct mem_cgroup
*memcg_in_oom
;
1834 gfp_t memcg_oom_gfp_mask
;
1835 int memcg_oom_order
;
1837 /* number of pages to reclaim on returning to userland */
1838 unsigned int memcg_nr_pages_over_high
;
1840 #ifdef CONFIG_UPROBES
1841 struct uprobe_task
*utask
;
1843 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1844 unsigned int sequential_io
;
1845 unsigned int sequential_io_avg
;
1847 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1848 unsigned long task_state_change
;
1850 int pagefault_disabled
;
1851 /* CPU-specific state of this task */
1852 struct thread_struct thread
;
1854 * WARNING: on x86, 'thread_struct' contains a variable-sized
1855 * structure. It *MUST* be at the end of 'task_struct'.
1857 * Do not put anything below here!
1861 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1862 extern int arch_task_struct_size __read_mostly
;
1864 # define arch_task_struct_size (sizeof(struct task_struct))
1867 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1868 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1870 #define TNF_MIGRATED 0x01
1871 #define TNF_NO_GROUP 0x02
1872 #define TNF_SHARED 0x04
1873 #define TNF_FAULT_LOCAL 0x08
1874 #define TNF_MIGRATE_FAIL 0x10
1876 #ifdef CONFIG_NUMA_BALANCING
1877 extern void task_numa_fault(int last_node
, int node
, int pages
, int flags
);
1878 extern pid_t
task_numa_group_id(struct task_struct
*p
);
1879 extern void set_numabalancing_state(bool enabled
);
1880 extern void task_numa_free(struct task_struct
*p
);
1881 extern bool should_numa_migrate_memory(struct task_struct
*p
, struct page
*page
,
1882 int src_nid
, int dst_cpu
);
1884 static inline void task_numa_fault(int last_node
, int node
, int pages
,
1888 static inline pid_t
task_numa_group_id(struct task_struct
*p
)
1892 static inline void set_numabalancing_state(bool enabled
)
1895 static inline void task_numa_free(struct task_struct
*p
)
1898 static inline bool should_numa_migrate_memory(struct task_struct
*p
,
1899 struct page
*page
, int src_nid
, int dst_cpu
)
1905 static inline struct pid
*task_pid(struct task_struct
*task
)
1907 return task
->pids
[PIDTYPE_PID
].pid
;
1910 static inline struct pid
*task_tgid(struct task_struct
*task
)
1912 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1916 * Without tasklist or rcu lock it is not safe to dereference
1917 * the result of task_pgrp/task_session even if task == current,
1918 * we can race with another thread doing sys_setsid/sys_setpgid.
1920 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1922 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1925 static inline struct pid
*task_session(struct task_struct
*task
)
1927 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1930 struct pid_namespace
;
1933 * the helpers to get the task's different pids as they are seen
1934 * from various namespaces
1936 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1937 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1939 * task_xid_nr_ns() : id seen from the ns specified;
1941 * set_task_vxid() : assigns a virtual id to a task;
1943 * see also pid_nr() etc in include/linux/pid.h
1945 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1946 struct pid_namespace
*ns
);
1948 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1953 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1954 struct pid_namespace
*ns
)
1956 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1959 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1961 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1965 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1970 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1972 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1974 return pid_vnr(task_tgid(tsk
));
1978 static inline int pid_alive(const struct task_struct
*p
);
1979 static inline pid_t
task_ppid_nr_ns(const struct task_struct
*tsk
, struct pid_namespace
*ns
)
1985 pid
= task_tgid_nr_ns(rcu_dereference(tsk
->real_parent
), ns
);
1991 static inline pid_t
task_ppid_nr(const struct task_struct
*tsk
)
1993 return task_ppid_nr_ns(tsk
, &init_pid_ns
);
1996 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1997 struct pid_namespace
*ns
)
1999 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
2002 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
2004 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
2008 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
2009 struct pid_namespace
*ns
)
2011 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
2014 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
2016 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
2019 /* obsolete, do not use */
2020 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
2022 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
2026 * pid_alive - check that a task structure is not stale
2027 * @p: Task structure to be checked.
2029 * Test if a process is not yet dead (at most zombie state)
2030 * If pid_alive fails, then pointers within the task structure
2031 * can be stale and must not be dereferenced.
2033 * Return: 1 if the process is alive. 0 otherwise.
2035 static inline int pid_alive(const struct task_struct
*p
)
2037 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
2041 * is_global_init - check if a task structure is init. Since init
2042 * is free to have sub-threads we need to check tgid.
2043 * @tsk: Task structure to be checked.
2045 * Check if a task structure is the first user space task the kernel created.
2047 * Return: 1 if the task structure is init. 0 otherwise.
2049 static inline int is_global_init(struct task_struct
*tsk
)
2051 return task_tgid_nr(tsk
) == 1;
2054 extern struct pid
*cad_pid
;
2056 extern void free_task(struct task_struct
*tsk
);
2057 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2059 extern void __put_task_struct(struct task_struct
*t
);
2061 static inline void put_task_struct(struct task_struct
*t
)
2063 if (atomic_dec_and_test(&t
->usage
))
2064 __put_task_struct(t
);
2067 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2068 extern void task_cputime(struct task_struct
*t
,
2069 cputime_t
*utime
, cputime_t
*stime
);
2070 extern void task_cputime_scaled(struct task_struct
*t
,
2071 cputime_t
*utimescaled
, cputime_t
*stimescaled
);
2072 extern cputime_t
task_gtime(struct task_struct
*t
);
2074 static inline void task_cputime(struct task_struct
*t
,
2075 cputime_t
*utime
, cputime_t
*stime
)
2083 static inline void task_cputime_scaled(struct task_struct
*t
,
2084 cputime_t
*utimescaled
,
2085 cputime_t
*stimescaled
)
2088 *utimescaled
= t
->utimescaled
;
2090 *stimescaled
= t
->stimescaled
;
2093 static inline cputime_t
task_gtime(struct task_struct
*t
)
2098 extern void task_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2099 extern void thread_group_cputime_adjusted(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
2104 #define PF_EXITING 0x00000004 /* getting shut down */
2105 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
2106 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
2107 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
2108 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
2109 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
2110 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
2111 #define PF_DUMPCORE 0x00000200 /* dumped core */
2112 #define PF_SIGNALED 0x00000400 /* killed by a signal */
2113 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
2114 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
2115 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
2116 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
2117 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
2118 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
2119 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
2120 #define PF_KSWAPD 0x00040000 /* I am kswapd */
2121 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
2122 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
2123 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
2124 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
2125 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
2126 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
2127 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
2128 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
2129 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
2130 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
2133 * Only the _current_ task can read/write to tsk->flags, but other
2134 * tasks can access tsk->flags in readonly mode for example
2135 * with tsk_used_math (like during threaded core dumping).
2136 * There is however an exception to this rule during ptrace
2137 * or during fork: the ptracer task is allowed to write to the
2138 * child->flags of its traced child (same goes for fork, the parent
2139 * can write to the child->flags), because we're guaranteed the
2140 * child is not running and in turn not changing child->flags
2141 * at the same time the parent does it.
2143 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2144 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2145 #define clear_used_math() clear_stopped_child_used_math(current)
2146 #define set_used_math() set_stopped_child_used_math(current)
2147 #define conditional_stopped_child_used_math(condition, child) \
2148 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2149 #define conditional_used_math(condition) \
2150 conditional_stopped_child_used_math(condition, current)
2151 #define copy_to_stopped_child_used_math(child) \
2152 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2153 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2154 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2155 #define used_math() tsk_used_math(current)
2157 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2158 * __GFP_FS is also cleared as it implies __GFP_IO.
2160 static inline gfp_t
memalloc_noio_flags(gfp_t flags
)
2162 if (unlikely(current
->flags
& PF_MEMALLOC_NOIO
))
2163 flags
&= ~(__GFP_IO
| __GFP_FS
);
2167 static inline unsigned int memalloc_noio_save(void)
2169 unsigned int flags
= current
->flags
& PF_MEMALLOC_NOIO
;
2170 current
->flags
|= PF_MEMALLOC_NOIO
;
2174 static inline void memalloc_noio_restore(unsigned int flags
)
2176 current
->flags
= (current
->flags
& ~PF_MEMALLOC_NOIO
) | flags
;
2179 /* Per-process atomic flags. */
2180 #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */
2181 #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */
2182 #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */
2185 #define TASK_PFA_TEST(name, func) \
2186 static inline bool task_##func(struct task_struct *p) \
2187 { return test_bit(PFA_##name, &p->atomic_flags); }
2188 #define TASK_PFA_SET(name, func) \
2189 static inline void task_set_##func(struct task_struct *p) \
2190 { set_bit(PFA_##name, &p->atomic_flags); }
2191 #define TASK_PFA_CLEAR(name, func) \
2192 static inline void task_clear_##func(struct task_struct *p) \
2193 { clear_bit(PFA_##name, &p->atomic_flags); }
2195 TASK_PFA_TEST(NO_NEW_PRIVS
, no_new_privs
)
2196 TASK_PFA_SET(NO_NEW_PRIVS
, no_new_privs
)
2198 TASK_PFA_TEST(SPREAD_PAGE
, spread_page
)
2199 TASK_PFA_SET(SPREAD_PAGE
, spread_page
)
2200 TASK_PFA_CLEAR(SPREAD_PAGE
, spread_page
)
2202 TASK_PFA_TEST(SPREAD_SLAB
, spread_slab
)
2203 TASK_PFA_SET(SPREAD_SLAB
, spread_slab
)
2204 TASK_PFA_CLEAR(SPREAD_SLAB
, spread_slab
)
2207 * task->jobctl flags
2209 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
2211 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
2212 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
2213 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
2214 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
2215 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
2216 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
2217 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
2219 #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2220 #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT)
2221 #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT)
2222 #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT)
2223 #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2224 #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT)
2225 #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT)
2227 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2228 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2230 extern bool task_set_jobctl_pending(struct task_struct
*task
,
2231 unsigned long mask
);
2232 extern void task_clear_jobctl_trapping(struct task_struct
*task
);
2233 extern void task_clear_jobctl_pending(struct task_struct
*task
,
2234 unsigned long mask
);
2236 static inline void rcu_copy_process(struct task_struct
*p
)
2238 #ifdef CONFIG_PREEMPT_RCU
2239 p
->rcu_read_lock_nesting
= 0;
2240 p
->rcu_read_unlock_special
.s
= 0;
2241 p
->rcu_blocked_node
= NULL
;
2242 INIT_LIST_HEAD(&p
->rcu_node_entry
);
2243 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2244 #ifdef CONFIG_TASKS_RCU
2245 p
->rcu_tasks_holdout
= false;
2246 INIT_LIST_HEAD(&p
->rcu_tasks_holdout_list
);
2247 p
->rcu_tasks_idle_cpu
= -1;
2248 #endif /* #ifdef CONFIG_TASKS_RCU */
2251 static inline void tsk_restore_flags(struct task_struct
*task
,
2252 unsigned long orig_flags
, unsigned long flags
)
2254 task
->flags
&= ~flags
;
2255 task
->flags
|= orig_flags
& flags
;
2258 extern int cpuset_cpumask_can_shrink(const struct cpumask
*cur
,
2259 const struct cpumask
*trial
);
2260 extern int task_can_attach(struct task_struct
*p
,
2261 const struct cpumask
*cs_cpus_allowed
);
2263 extern void do_set_cpus_allowed(struct task_struct
*p
,
2264 const struct cpumask
*new_mask
);
2266 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
2267 const struct cpumask
*new_mask
);
2269 static inline void do_set_cpus_allowed(struct task_struct
*p
,
2270 const struct cpumask
*new_mask
)
2273 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
2274 const struct cpumask
*new_mask
)
2276 if (!cpumask_test_cpu(0, new_mask
))
2282 #ifdef CONFIG_NO_HZ_COMMON
2283 void calc_load_enter_idle(void);
2284 void calc_load_exit_idle(void);
2286 static inline void calc_load_enter_idle(void) { }
2287 static inline void calc_load_exit_idle(void) { }
2288 #endif /* CONFIG_NO_HZ_COMMON */
2291 * Do not use outside of architecture code which knows its limitations.
2293 * sched_clock() has no promise of monotonicity or bounded drift between
2294 * CPUs, use (which you should not) requires disabling IRQs.
2296 * Please use one of the three interfaces below.
2298 extern unsigned long long notrace
sched_clock(void);
2300 * See the comment in kernel/sched/clock.c
2302 extern u64
cpu_clock(int cpu
);
2303 extern u64
local_clock(void);
2304 extern u64
running_clock(void);
2305 extern u64
sched_clock_cpu(int cpu
);
2308 extern void sched_clock_init(void);
2310 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2311 static inline void sched_clock_tick(void)
2315 static inline void sched_clock_idle_sleep_event(void)
2319 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
2324 * Architectures can set this to 1 if they have specified
2325 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2326 * but then during bootup it turns out that sched_clock()
2327 * is reliable after all:
2329 extern int sched_clock_stable(void);
2330 extern void set_sched_clock_stable(void);
2331 extern void clear_sched_clock_stable(void);
2333 extern void sched_clock_tick(void);
2334 extern void sched_clock_idle_sleep_event(void);
2335 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2338 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2340 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2341 * The reason for this explicit opt-in is not to have perf penalty with
2342 * slow sched_clocks.
2344 extern void enable_sched_clock_irqtime(void);
2345 extern void disable_sched_clock_irqtime(void);
2347 static inline void enable_sched_clock_irqtime(void) {}
2348 static inline void disable_sched_clock_irqtime(void) {}
2351 extern unsigned long long
2352 task_sched_runtime(struct task_struct
*task
);
2354 /* sched_exec is called by processes performing an exec */
2356 extern void sched_exec(void);
2358 #define sched_exec() {}
2361 extern void sched_clock_idle_sleep_event(void);
2362 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
2364 #ifdef CONFIG_HOTPLUG_CPU
2365 extern void idle_task_exit(void);
2367 static inline void idle_task_exit(void) {}
2370 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2371 extern void wake_up_nohz_cpu(int cpu
);
2373 static inline void wake_up_nohz_cpu(int cpu
) { }
2376 #ifdef CONFIG_NO_HZ_FULL
2377 extern u64
scheduler_tick_max_deferment(void);
2380 #ifdef CONFIG_SCHED_AUTOGROUP
2381 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2382 extern void sched_autogroup_detach(struct task_struct
*p
);
2383 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2384 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2385 #ifdef CONFIG_PROC_FS
2386 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2387 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int nice
);
2390 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2391 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2392 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2393 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2396 extern int yield_to(struct task_struct
*p
, bool preempt
);
2397 extern void set_user_nice(struct task_struct
*p
, long nice
);
2398 extern int task_prio(const struct task_struct
*p
);
2400 * task_nice - return the nice value of a given task.
2401 * @p: the task in question.
2403 * Return: The nice value [ -20 ... 0 ... 19 ].
2405 static inline int task_nice(const struct task_struct
*p
)
2407 return PRIO_TO_NICE((p
)->static_prio
);
2409 extern int can_nice(const struct task_struct
*p
, const int nice
);
2410 extern int task_curr(const struct task_struct
*p
);
2411 extern int idle_cpu(int cpu
);
2412 extern int sched_setscheduler(struct task_struct
*, int,
2413 const struct sched_param
*);
2414 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2415 const struct sched_param
*);
2416 extern int sched_setattr(struct task_struct
*,
2417 const struct sched_attr
*);
2418 extern struct task_struct
*idle_task(int cpu
);
2420 * is_idle_task - is the specified task an idle task?
2421 * @p: the task in question.
2423 * Return: 1 if @p is an idle task. 0 otherwise.
2425 static inline bool is_idle_task(const struct task_struct
*p
)
2429 extern struct task_struct
*curr_task(int cpu
);
2430 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2434 union thread_union
{
2435 struct thread_info thread_info
;
2436 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2439 #ifndef __HAVE_ARCH_KSTACK_END
2440 static inline int kstack_end(void *addr
)
2442 /* Reliable end of stack detection:
2443 * Some APM bios versions misalign the stack
2445 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2449 extern union thread_union init_thread_union
;
2450 extern struct task_struct init_task
;
2452 extern struct mm_struct init_mm
;
2454 extern struct pid_namespace init_pid_ns
;
2457 * find a task by one of its numerical ids
2459 * find_task_by_pid_ns():
2460 * finds a task by its pid in the specified namespace
2461 * find_task_by_vpid():
2462 * finds a task by its virtual pid
2464 * see also find_vpid() etc in include/linux/pid.h
2467 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2468 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2469 struct pid_namespace
*ns
);
2471 /* per-UID process charging. */
2472 extern struct user_struct
* alloc_uid(kuid_t
);
2473 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2475 atomic_inc(&u
->__count
);
2478 extern void free_uid(struct user_struct
*);
2480 #include <asm/current.h>
2482 extern void xtime_update(unsigned long ticks
);
2484 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2485 extern int wake_up_process(struct task_struct
*tsk
);
2486 extern void wake_up_new_task(struct task_struct
*tsk
);
2488 extern void kick_process(struct task_struct
*tsk
);
2490 static inline void kick_process(struct task_struct
*tsk
) { }
2492 extern int sched_fork(unsigned long clone_flags
, struct task_struct
*p
);
2493 extern void sched_dead(struct task_struct
*p
);
2495 extern void proc_caches_init(void);
2496 extern void flush_signals(struct task_struct
*);
2497 extern void ignore_signals(struct task_struct
*);
2498 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2499 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2501 static inline int kernel_dequeue_signal(siginfo_t
*info
)
2503 struct task_struct
*tsk
= current
;
2507 spin_lock_irq(&tsk
->sighand
->siglock
);
2508 ret
= dequeue_signal(tsk
, &tsk
->blocked
, info
?: &__info
);
2509 spin_unlock_irq(&tsk
->sighand
->siglock
);
2514 static inline void kernel_signal_stop(void)
2516 spin_lock_irq(¤t
->sighand
->siglock
);
2517 if (current
->jobctl
& JOBCTL_STOP_DEQUEUED
)
2518 __set_current_state(TASK_STOPPED
);
2519 spin_unlock_irq(¤t
->sighand
->siglock
);
2524 extern void release_task(struct task_struct
* p
);
2525 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2526 extern int force_sigsegv(int, struct task_struct
*);
2527 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2528 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2529 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2530 extern int kill_pid_info_as_cred(int, struct siginfo
*, struct pid
*,
2531 const struct cred
*, u32
);
2532 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2533 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2534 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2535 extern __must_check
bool do_notify_parent(struct task_struct
*, int);
2536 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2537 extern void force_sig(int, struct task_struct
*);
2538 extern int send_sig(int, struct task_struct
*, int);
2539 extern int zap_other_threads(struct task_struct
*p
);
2540 extern struct sigqueue
*sigqueue_alloc(void);
2541 extern void sigqueue_free(struct sigqueue
*);
2542 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2543 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2545 static inline void restore_saved_sigmask(void)
2547 if (test_and_clear_restore_sigmask())
2548 __set_current_blocked(¤t
->saved_sigmask
);
2551 static inline sigset_t
*sigmask_to_save(void)
2553 sigset_t
*res
= ¤t
->blocked
;
2554 if (unlikely(test_restore_sigmask()))
2555 res
= ¤t
->saved_sigmask
;
2559 static inline int kill_cad_pid(int sig
, int priv
)
2561 return kill_pid(cad_pid
, sig
, priv
);
2564 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2565 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2566 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2567 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2570 * True if we are on the alternate signal stack.
2572 static inline int on_sig_stack(unsigned long sp
)
2574 #ifdef CONFIG_STACK_GROWSUP
2575 return sp
>= current
->sas_ss_sp
&&
2576 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2578 return sp
> current
->sas_ss_sp
&&
2579 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2583 static inline int sas_ss_flags(unsigned long sp
)
2585 if (!current
->sas_ss_size
)
2588 return on_sig_stack(sp
) ? SS_ONSTACK
: 0;
2591 static inline unsigned long sigsp(unsigned long sp
, struct ksignal
*ksig
)
2593 if (unlikely((ksig
->ka
.sa
.sa_flags
& SA_ONSTACK
)) && ! sas_ss_flags(sp
))
2594 #ifdef CONFIG_STACK_GROWSUP
2595 return current
->sas_ss_sp
;
2597 return current
->sas_ss_sp
+ current
->sas_ss_size
;
2603 * Routines for handling mm_structs
2605 extern struct mm_struct
* mm_alloc(void);
2607 /* mmdrop drops the mm and the page tables */
2608 extern void __mmdrop(struct mm_struct
*);
2609 static inline void mmdrop(struct mm_struct
* mm
)
2611 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2615 /* mmput gets rid of the mappings and all user-space */
2616 extern void mmput(struct mm_struct
*);
2617 /* Grab a reference to a task's mm, if it is not already going away */
2618 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2620 * Grab a reference to a task's mm, if it is not already going away
2621 * and ptrace_may_access with the mode parameter passed to it
2624 extern struct mm_struct
*mm_access(struct task_struct
*task
, unsigned int mode
);
2625 /* Remove the current tasks stale references to the old mm_struct */
2626 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2628 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2629 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2630 struct task_struct
*, unsigned long);
2632 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2633 struct task_struct
*);
2635 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2636 * via pt_regs, so ignore the tls argument passed via C. */
2637 static inline int copy_thread_tls(
2638 unsigned long clone_flags
, unsigned long sp
, unsigned long arg
,
2639 struct task_struct
*p
, unsigned long tls
)
2641 return copy_thread(clone_flags
, sp
, arg
, p
);
2644 extern void flush_thread(void);
2645 extern void exit_thread(void);
2647 extern void exit_files(struct task_struct
*);
2648 extern void __cleanup_sighand(struct sighand_struct
*);
2650 extern void exit_itimers(struct signal_struct
*);
2651 extern void flush_itimer_signals(void);
2653 extern void do_group_exit(int);
2655 extern int do_execve(struct filename
*,
2656 const char __user
* const __user
*,
2657 const char __user
* const __user
*);
2658 extern int do_execveat(int, struct filename
*,
2659 const char __user
* const __user
*,
2660 const char __user
* const __user
*,
2662 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*, unsigned long);
2663 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user
*, int __user
*);
2664 struct task_struct
*fork_idle(int);
2665 extern pid_t
kernel_thread(int (*fn
)(void *), void *arg
, unsigned long flags
);
2667 extern void __set_task_comm(struct task_struct
*tsk
, const char *from
, bool exec
);
2668 static inline void set_task_comm(struct task_struct
*tsk
, const char *from
)
2670 __set_task_comm(tsk
, from
, false);
2672 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2675 void scheduler_ipi(void);
2676 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2678 static inline void scheduler_ipi(void) { }
2679 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2686 #define tasklist_empty() \
2687 list_empty(&init_task.tasks)
2689 #define next_task(p) \
2690 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2692 #define for_each_process(p) \
2693 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2695 extern bool current_is_single_threaded(void);
2698 * Careful: do_each_thread/while_each_thread is a double loop so
2699 * 'break' will not work as expected - use goto instead.
2701 #define do_each_thread(g, t) \
2702 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2704 #define while_each_thread(g, t) \
2705 while ((t = next_thread(t)) != g)
2707 #define __for_each_thread(signal, t) \
2708 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2710 #define for_each_thread(p, t) \
2711 __for_each_thread((p)->signal, t)
2713 /* Careful: this is a double loop, 'break' won't work as expected. */
2714 #define for_each_process_thread(p, t) \
2715 for_each_process(p) for_each_thread(p, t)
2717 static inline int get_nr_threads(struct task_struct
*tsk
)
2719 return tsk
->signal
->nr_threads
;
2722 static inline bool thread_group_leader(struct task_struct
*p
)
2724 return p
->exit_signal
>= 0;
2727 /* Do to the insanities of de_thread it is possible for a process
2728 * to have the pid of the thread group leader without actually being
2729 * the thread group leader. For iteration through the pids in proc
2730 * all we care about is that we have a task with the appropriate
2731 * pid, we don't actually care if we have the right task.
2733 static inline bool has_group_leader_pid(struct task_struct
*p
)
2735 return task_pid(p
) == p
->signal
->leader_pid
;
2739 bool same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2741 return p1
->signal
== p2
->signal
;
2744 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2746 return list_entry_rcu(p
->thread_group
.next
,
2747 struct task_struct
, thread_group
);
2750 static inline int thread_group_empty(struct task_struct
*p
)
2752 return list_empty(&p
->thread_group
);
2755 #define delay_group_leader(p) \
2756 (thread_group_leader(p) && !thread_group_empty(p))
2759 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2760 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2761 * pins the final release of task.io_context. Also protects ->cpuset and
2762 * ->cgroup.subsys[]. And ->vfork_done.
2764 * Nests both inside and outside of read_lock(&tasklist_lock).
2765 * It must not be nested with write_lock_irq(&tasklist_lock),
2766 * neither inside nor outside.
2768 static inline void task_lock(struct task_struct
*p
)
2770 spin_lock(&p
->alloc_lock
);
2773 static inline void task_unlock(struct task_struct
*p
)
2775 spin_unlock(&p
->alloc_lock
);
2778 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2779 unsigned long *flags
);
2781 static inline struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2782 unsigned long *flags
)
2784 struct sighand_struct
*ret
;
2786 ret
= __lock_task_sighand(tsk
, flags
);
2787 (void)__cond_lock(&tsk
->sighand
->siglock
, ret
);
2791 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2792 unsigned long *flags
)
2794 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2798 * threadgroup_change_begin - mark the beginning of changes to a threadgroup
2799 * @tsk: task causing the changes
2801 * All operations which modify a threadgroup - a new thread joining the
2802 * group, death of a member thread (the assertion of PF_EXITING) and
2803 * exec(2) dethreading the process and replacing the leader - are wrapped
2804 * by threadgroup_change_{begin|end}(). This is to provide a place which
2805 * subsystems needing threadgroup stability can hook into for
2808 static inline void threadgroup_change_begin(struct task_struct
*tsk
)
2811 cgroup_threadgroup_change_begin(tsk
);
2815 * threadgroup_change_end - mark the end of changes to a threadgroup
2816 * @tsk: task causing the changes
2818 * See threadgroup_change_begin().
2820 static inline void threadgroup_change_end(struct task_struct
*tsk
)
2822 cgroup_threadgroup_change_end(tsk
);
2825 #ifndef __HAVE_THREAD_FUNCTIONS
2827 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2828 #define task_stack_page(task) ((task)->stack)
2830 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2832 *task_thread_info(p
) = *task_thread_info(org
);
2833 task_thread_info(p
)->task
= p
;
2837 * Return the address of the last usable long on the stack.
2839 * When the stack grows down, this is just above the thread
2840 * info struct. Going any lower will corrupt the threadinfo.
2842 * When the stack grows up, this is the highest address.
2843 * Beyond that position, we corrupt data on the next page.
2845 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2847 #ifdef CONFIG_STACK_GROWSUP
2848 return (unsigned long *)((unsigned long)task_thread_info(p
) + THREAD_SIZE
) - 1;
2850 return (unsigned long *)(task_thread_info(p
) + 1);
2855 #define task_stack_end_corrupted(task) \
2856 (*(end_of_stack(task)) != STACK_END_MAGIC)
2858 static inline int object_is_on_stack(void *obj
)
2860 void *stack
= task_stack_page(current
);
2862 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2865 extern void thread_info_cache_init(void);
2867 #ifdef CONFIG_DEBUG_STACK_USAGE
2868 static inline unsigned long stack_not_used(struct task_struct
*p
)
2870 unsigned long *n
= end_of_stack(p
);
2872 do { /* Skip over canary */
2873 # ifdef CONFIG_STACK_GROWSUP
2880 # ifdef CONFIG_STACK_GROWSUP
2881 return (unsigned long)end_of_stack(p
) - (unsigned long)n
;
2883 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2887 extern void set_task_stack_end_magic(struct task_struct
*tsk
);
2889 /* set thread flags in other task's structures
2890 * - see asm/thread_info.h for TIF_xxxx flags available
2892 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2894 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2897 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2899 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2902 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2904 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2907 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2909 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2912 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2914 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2917 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2919 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2922 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2924 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2927 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2929 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2932 static inline int restart_syscall(void)
2934 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2935 return -ERESTARTNOINTR
;
2938 static inline int signal_pending(struct task_struct
*p
)
2940 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2943 static inline int __fatal_signal_pending(struct task_struct
*p
)
2945 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2948 static inline int fatal_signal_pending(struct task_struct
*p
)
2950 return signal_pending(p
) && __fatal_signal_pending(p
);
2953 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2955 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2957 if (!signal_pending(p
))
2960 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2964 * cond_resched() and cond_resched_lock(): latency reduction via
2965 * explicit rescheduling in places that are safe. The return
2966 * value indicates whether a reschedule was done in fact.
2967 * cond_resched_lock() will drop the spinlock before scheduling,
2968 * cond_resched_softirq() will enable bhs before scheduling.
2970 extern int _cond_resched(void);
2972 #define cond_resched() ({ \
2973 ___might_sleep(__FILE__, __LINE__, 0); \
2977 extern int __cond_resched_lock(spinlock_t
*lock
);
2979 #define cond_resched_lock(lock) ({ \
2980 ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
2981 __cond_resched_lock(lock); \
2984 extern int __cond_resched_softirq(void);
2986 #define cond_resched_softirq() ({ \
2987 ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2988 __cond_resched_softirq(); \
2991 static inline void cond_resched_rcu(void)
2993 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3001 * Does a critical section need to be broken due to another
3002 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3003 * but a general need for low latency)
3005 static inline int spin_needbreak(spinlock_t
*lock
)
3007 #ifdef CONFIG_PREEMPT
3008 return spin_is_contended(lock
);
3015 * Idle thread specific functions to determine the need_resched
3018 #ifdef TIF_POLLING_NRFLAG
3019 static inline int tsk_is_polling(struct task_struct
*p
)
3021 return test_tsk_thread_flag(p
, TIF_POLLING_NRFLAG
);
3024 static inline void __current_set_polling(void)
3026 set_thread_flag(TIF_POLLING_NRFLAG
);
3029 static inline bool __must_check
current_set_polling_and_test(void)
3031 __current_set_polling();
3034 * Polling state must be visible before we test NEED_RESCHED,
3035 * paired by resched_curr()
3037 smp_mb__after_atomic();
3039 return unlikely(tif_need_resched());
3042 static inline void __current_clr_polling(void)
3044 clear_thread_flag(TIF_POLLING_NRFLAG
);
3047 static inline bool __must_check
current_clr_polling_and_test(void)
3049 __current_clr_polling();
3052 * Polling state must be visible before we test NEED_RESCHED,
3053 * paired by resched_curr()
3055 smp_mb__after_atomic();
3057 return unlikely(tif_need_resched());
3061 static inline int tsk_is_polling(struct task_struct
*p
) { return 0; }
3062 static inline void __current_set_polling(void) { }
3063 static inline void __current_clr_polling(void) { }
3065 static inline bool __must_check
current_set_polling_and_test(void)
3067 return unlikely(tif_need_resched());
3069 static inline bool __must_check
current_clr_polling_and_test(void)
3071 return unlikely(tif_need_resched());
3075 static inline void current_clr_polling(void)
3077 __current_clr_polling();
3080 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3081 * Once the bit is cleared, we'll get IPIs with every new
3082 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3085 smp_mb(); /* paired with resched_curr() */
3087 preempt_fold_need_resched();
3090 static __always_inline
bool need_resched(void)
3092 return unlikely(tif_need_resched());
3096 * Thread group CPU time accounting.
3098 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
3099 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
3102 * Reevaluate whether the task has signals pending delivery.
3103 * Wake the task if so.
3104 * This is required every time the blocked sigset_t changes.
3105 * callers must hold sighand->siglock.
3107 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
3108 extern void recalc_sigpending(void);
3110 extern void signal_wake_up_state(struct task_struct
*t
, unsigned int state
);
3112 static inline void signal_wake_up(struct task_struct
*t
, bool resume
)
3114 signal_wake_up_state(t
, resume
? TASK_WAKEKILL
: 0);
3116 static inline void ptrace_signal_wake_up(struct task_struct
*t
, bool resume
)
3118 signal_wake_up_state(t
, resume
? __TASK_TRACED
: 0);
3122 * Wrappers for p->thread_info->cpu access. No-op on UP.
3126 static inline unsigned int task_cpu(const struct task_struct
*p
)
3128 return task_thread_info(p
)->cpu
;
3131 static inline int task_node(const struct task_struct
*p
)
3133 return cpu_to_node(task_cpu(p
));
3136 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
3140 static inline unsigned int task_cpu(const struct task_struct
*p
)
3145 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
3149 #endif /* CONFIG_SMP */
3151 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
3152 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
3154 #ifdef CONFIG_CGROUP_SCHED
3155 extern struct task_group root_task_group
;
3156 #endif /* CONFIG_CGROUP_SCHED */
3158 extern int task_can_switch_user(struct user_struct
*up
,
3159 struct task_struct
*tsk
);
3161 #ifdef CONFIG_TASK_XACCT
3162 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3164 tsk
->ioac
.rchar
+= amt
;
3167 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3169 tsk
->ioac
.wchar
+= amt
;
3172 static inline void inc_syscr(struct task_struct
*tsk
)
3177 static inline void inc_syscw(struct task_struct
*tsk
)
3182 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
3186 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
3190 static inline void inc_syscr(struct task_struct
*tsk
)
3194 static inline void inc_syscw(struct task_struct
*tsk
)
3199 #ifndef TASK_SIZE_OF
3200 #define TASK_SIZE_OF(tsk) TASK_SIZE
3204 extern void mm_update_next_owner(struct mm_struct
*mm
);
3206 static inline void mm_update_next_owner(struct mm_struct
*mm
)
3209 #endif /* CONFIG_MEMCG */
3211 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
3214 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
3217 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
3220 return READ_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
3223 static inline unsigned long rlimit(unsigned int limit
)
3225 return task_rlimit(current
, limit
);
3228 static inline unsigned long rlimit_max(unsigned int limit
)
3230 return task_rlimit_max(current
, limit
);
3233 #ifdef CONFIG_CPU_FREQ
3234 struct update_util_data
{
3235 void (*func
)(struct update_util_data
*data
,
3236 u64 time
, unsigned long util
, unsigned long max
);
3239 void cpufreq_set_update_util_data(int cpu
, struct update_util_data
*data
);
3240 #endif /* CONFIG_CPU_FREQ */