7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25 and is now available for re-use. */
26 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
27 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
28 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
29 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
31 #define CLONE_IO 0x80000000 /* Clone io context */
36 #define SCHED_NORMAL 0
40 /* SCHED_ISO: reserved but not implemented yet */
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK 0x40000000
51 #include <asm/param.h> /* for HZ */
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
66 #include <asm/system.h>
68 #include <asm/ptrace.h>
69 #include <asm/cputime.h>
71 #include <linux/smp.h>
72 #include <linux/sem.h>
73 #include <linux/signal.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/latencytop.h>
92 #include <linux/cred.h>
94 #include <asm/processor.h>
97 struct futex_pi_state
;
98 struct robust_list_head
;
101 struct perf_event_context
;
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
120 extern unsigned long avenrun
[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
130 #define CALC_LOAD(load,exp,n) \
132 load += n*(FIXED_1-exp); \
135 extern unsigned long total_forks
;
136 extern int nr_threads
;
137 DECLARE_PER_CPU(unsigned long, process_counts
);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu
);
143 extern unsigned long this_cpu_load(void);
146 extern void calc_global_load(unsigned long ticks
);
148 extern unsigned long get_parent_ip(unsigned long addr
);
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
155 extern void proc_sched_set_task(struct task_struct
*p
);
157 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
160 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
163 static inline void proc_sched_set_task(struct task_struct
*p
)
167 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
190 /* in tsk->state again */
192 #define TASK_WAKEKILL 128
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 extern char ___assert_task_state
[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR
)-1 != ilog2(TASK_STATE_MAX
)+1)];
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
238 * If the caller does not need such serialisation then use __set_current_state()
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
248 #include <linux/spinlock.h>
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
256 extern rwlock_t tasklist_lock
;
257 extern spinlock_t mmlist_lock
;
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
268 extern void init_idle(struct task_struct
*idle
, int cpu
);
269 extern void init_idle_bootup_task(struct task_struct
*idle
);
271 extern int runqueue_is_locked(int cpu
);
273 extern cpumask_var_t nohz_cpu_mask
;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern void select_nohz_load_balancer(int stop_tick
);
276 extern int get_nohz_timer_target(void);
278 static inline void select_nohz_load_balancer(int stop_tick
) { }
282 * Only dump TASK_* tasks. (0 for all tasks)
284 extern void show_state_filter(unsigned long state_filter
);
286 static inline void show_state(void)
288 show_state_filter(0);
291 extern void show_regs(struct pt_regs
*);
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
298 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout
);
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user
);
306 extern void scheduler_tick(void);
308 extern void sched_show_task(struct task_struct
*p
);
310 #ifdef CONFIG_LOCKUP_DETECTOR
311 extern void touch_softlockup_watchdog(void);
312 extern void touch_softlockup_watchdog_sync(void);
313 extern void touch_all_softlockup_watchdogs(void);
314 extern int proc_dowatchdog_thresh(struct ctl_table
*table
, int write
,
316 size_t *lenp
, loff_t
*ppos
);
317 extern unsigned int softlockup_panic
;
318 void lockup_detector_init(void);
320 static inline void touch_softlockup_watchdog(void)
323 static inline void touch_softlockup_watchdog_sync(void)
326 static inline void touch_all_softlockup_watchdogs(void)
329 static inline void lockup_detector_init(void)
334 #ifdef CONFIG_DETECT_HUNG_TASK
335 extern unsigned int sysctl_hung_task_panic
;
336 extern unsigned long sysctl_hung_task_check_count
;
337 extern unsigned long sysctl_hung_task_timeout_secs
;
338 extern unsigned long sysctl_hung_task_warnings
;
339 extern int proc_dohung_task_timeout_secs(struct ctl_table
*table
, int write
,
341 size_t *lenp
, loff_t
*ppos
);
343 /* Avoid need for ifdefs elsewhere in the code */
344 enum { sysctl_hung_task_timeout_secs
= 0 };
347 /* Attach to any functions which should be ignored in wchan output. */
348 #define __sched __attribute__((__section__(".sched.text")))
350 /* Linker adds these: start and end of __sched functions */
351 extern char __sched_text_start
[], __sched_text_end
[];
353 /* Is this address in the __sched functions? */
354 extern int in_sched_functions(unsigned long addr
);
356 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
357 extern signed long schedule_timeout(signed long timeout
);
358 extern signed long schedule_timeout_interruptible(signed long timeout
);
359 extern signed long schedule_timeout_killable(signed long timeout
);
360 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
361 asmlinkage
void schedule(void);
362 extern int mutex_spin_on_owner(struct mutex
*lock
, struct task_struct
*owner
);
365 struct user_namespace
;
368 * Default maximum number of active map areas, this limits the number of vmas
369 * per mm struct. Users can overwrite this number by sysctl but there is a
372 * When a program's coredump is generated as ELF format, a section is created
373 * per a vma. In ELF, the number of sections is represented in unsigned short.
374 * This means the number of sections should be smaller than 65535 at coredump.
375 * Because the kernel adds some informative sections to a image of program at
376 * generating coredump, we need some margin. The number of extra sections is
377 * 1-3 now and depends on arch. We use "5" as safe margin, here.
379 #define MAPCOUNT_ELF_CORE_MARGIN (5)
380 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
382 extern int sysctl_max_map_count
;
384 #include <linux/aio.h>
387 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
389 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
390 unsigned long, unsigned long);
392 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
393 unsigned long len
, unsigned long pgoff
,
394 unsigned long flags
);
395 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
396 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
398 static inline void arch_pick_mmap_layout(struct mm_struct
*mm
) {}
402 extern void set_dumpable(struct mm_struct
*mm
, int value
);
403 extern int get_dumpable(struct mm_struct
*mm
);
407 #define MMF_DUMPABLE 0 /* core dump is permitted */
408 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
410 #define MMF_DUMPABLE_BITS 2
411 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
413 /* coredump filter bits */
414 #define MMF_DUMP_ANON_PRIVATE 2
415 #define MMF_DUMP_ANON_SHARED 3
416 #define MMF_DUMP_MAPPED_PRIVATE 4
417 #define MMF_DUMP_MAPPED_SHARED 5
418 #define MMF_DUMP_ELF_HEADERS 6
419 #define MMF_DUMP_HUGETLB_PRIVATE 7
420 #define MMF_DUMP_HUGETLB_SHARED 8
422 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
423 #define MMF_DUMP_FILTER_BITS 7
424 #define MMF_DUMP_FILTER_MASK \
425 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
426 #define MMF_DUMP_FILTER_DEFAULT \
427 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
428 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
430 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
431 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
433 # define MMF_DUMP_MASK_DEFAULT_ELF 0
435 /* leave room for more dump flags */
436 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
437 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
441 struct sighand_struct
{
443 struct k_sigaction action
[_NSIG
];
445 wait_queue_head_t signalfd_wqh
;
448 struct pacct_struct
{
451 unsigned long ac_mem
;
452 cputime_t ac_utime
, ac_stime
;
453 unsigned long ac_minflt
, ac_majflt
;
464 * struct task_cputime - collected CPU time counts
465 * @utime: time spent in user mode, in &cputime_t units
466 * @stime: time spent in kernel mode, in &cputime_t units
467 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
469 * This structure groups together three kinds of CPU time that are
470 * tracked for threads and thread groups. Most things considering
471 * CPU time want to group these counts together and treat all three
472 * of them in parallel.
474 struct task_cputime
{
477 unsigned long long sum_exec_runtime
;
479 /* Alternate field names when used to cache expirations. */
480 #define prof_exp stime
481 #define virt_exp utime
482 #define sched_exp sum_exec_runtime
484 #define INIT_CPUTIME \
485 (struct task_cputime) { \
486 .utime = cputime_zero, \
487 .stime = cputime_zero, \
488 .sum_exec_runtime = 0, \
492 * Disable preemption until the scheduler is running.
493 * Reset by start_kernel()->sched_init()->init_idle().
495 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
496 * before the scheduler is active -- see should_resched().
498 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
501 * struct thread_group_cputimer - thread group interval timer counts
502 * @cputime: thread group interval timers.
503 * @running: non-zero when there are timers running and
504 * @cputime receives updates.
505 * @lock: lock for fields in this struct.
507 * This structure contains the version of task_cputime, above, that is
508 * used for thread group CPU timer calculations.
510 struct thread_group_cputimer
{
511 struct task_cputime cputime
;
516 #include <linux/rwsem.h>
520 * NOTE! "signal_struct" does not have its own
521 * locking, because a shared signal_struct always
522 * implies a shared sighand_struct, so locking
523 * sighand_struct is always a proper superset of
524 * the locking of signal_struct.
526 struct signal_struct
{
531 wait_queue_head_t wait_chldexit
; /* for wait4() */
533 /* current thread group signal load-balancing target: */
534 struct task_struct
*curr_target
;
536 /* shared signal handling: */
537 struct sigpending shared_pending
;
539 /* thread group exit support */
542 * - notify group_exit_task when ->count is equal to notify_count
543 * - everyone except group_exit_task is stopped during signal delivery
544 * of fatal signals, group_exit_task processes the signal.
547 struct task_struct
*group_exit_task
;
549 /* thread group stop support, overloads group_exit_code too */
550 int group_stop_count
;
551 unsigned int flags
; /* see SIGNAL_* flags below */
553 /* POSIX.1b Interval Timers */
554 struct list_head posix_timers
;
556 /* ITIMER_REAL timer for the process */
557 struct hrtimer real_timer
;
558 struct pid
*leader_pid
;
559 ktime_t it_real_incr
;
562 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
563 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
564 * values are defined to 0 and 1 respectively
566 struct cpu_itimer it
[2];
569 * Thread group totals for process CPU timers.
570 * See thread_group_cputimer(), et al, for details.
572 struct thread_group_cputimer cputimer
;
574 /* Earliest-expiration cache. */
575 struct task_cputime cputime_expires
;
577 struct list_head cpu_timers
[3];
579 struct pid
*tty_old_pgrp
;
581 /* boolean value for session group leader */
584 struct tty_struct
*tty
; /* NULL if no tty */
586 #ifdef CONFIG_SCHED_AUTOGROUP
587 struct autogroup
*autogroup
;
590 * Cumulative resource counters for dead threads in the group,
591 * and for reaped dead child processes forked by this group.
592 * Live threads maintain their own counters and add to these
593 * in __exit_signal, except for the group leader.
595 cputime_t utime
, stime
, cutime
, cstime
;
598 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
599 cputime_t prev_utime
, prev_stime
;
601 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
602 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
603 unsigned long inblock
, oublock
, cinblock
, coublock
;
604 unsigned long maxrss
, cmaxrss
;
605 struct task_io_accounting ioac
;
608 * Cumulative ns of schedule CPU time fo dead threads in the
609 * group, not including a zombie group leader, (This only differs
610 * from jiffies_to_ns(utime + stime) if sched_clock uses something
611 * other than jiffies.)
613 unsigned long long sum_sched_runtime
;
616 * We don't bother to synchronize most readers of this at all,
617 * because there is no reader checking a limit that actually needs
618 * to get both rlim_cur and rlim_max atomically, and either one
619 * alone is a single word that can safely be read normally.
620 * getrlimit/setrlimit use task_lock(current->group_leader) to
621 * protect this instead of the siglock, because they really
622 * have no need to disable irqs.
624 struct rlimit rlim
[RLIM_NLIMITS
];
626 #ifdef CONFIG_BSD_PROCESS_ACCT
627 struct pacct_struct pacct
; /* per-process accounting information */
629 #ifdef CONFIG_TASKSTATS
630 struct taskstats
*stats
;
634 struct tty_audit_buf
*tty_audit_buf
;
636 #ifdef CONFIG_CGROUPS
638 * The threadgroup_fork_lock prevents threads from forking with
639 * CLONE_THREAD while held for writing. Use this for fork-sensitive
640 * threadgroup-wide operations. It's taken for reading in fork.c in
642 * Currently only needed write-side by cgroups.
644 struct rw_semaphore threadgroup_fork_lock
;
647 int oom_adj
; /* OOM kill score adjustment (bit shift) */
648 int oom_score_adj
; /* OOM kill score adjustment */
649 int oom_score_adj_min
; /* OOM kill score adjustment minimum value.
650 * Only settable by CAP_SYS_RESOURCE. */
652 struct mutex cred_guard_mutex
; /* guard against foreign influences on
653 * credential calculations
654 * (notably. ptrace) */
657 /* Context switch must be unlocked if interrupts are to be enabled */
658 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
659 # define __ARCH_WANT_UNLOCKED_CTXSW
663 * Bits in flags field of signal_struct.
665 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
666 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
667 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
669 * Pending notifications to parent.
671 #define SIGNAL_CLD_STOPPED 0x00000010
672 #define SIGNAL_CLD_CONTINUED 0x00000020
673 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
675 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
677 /* If true, all threads except ->group_exit_task have pending SIGKILL */
678 static inline int signal_group_exit(const struct signal_struct
*sig
)
680 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
681 (sig
->group_exit_task
!= NULL
);
685 * Some day this will be a full-fledged user tracking system..
688 atomic_t __count
; /* reference count */
689 atomic_t processes
; /* How many processes does this user have? */
690 atomic_t files
; /* How many open files does this user have? */
691 atomic_t sigpending
; /* How many pending signals does this user have? */
692 #ifdef CONFIG_INOTIFY_USER
693 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
694 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
696 #ifdef CONFIG_FANOTIFY
697 atomic_t fanotify_listeners
;
700 atomic_long_t epoll_watches
; /* The number of file descriptors currently watched */
702 #ifdef CONFIG_POSIX_MQUEUE
703 /* protected by mq_lock */
704 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
706 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
709 struct key
*uid_keyring
; /* UID specific keyring */
710 struct key
*session_keyring
; /* UID's default session keyring */
713 /* Hash table maintenance information */
714 struct hlist_node uidhash_node
;
716 struct user_namespace
*user_ns
;
718 #ifdef CONFIG_PERF_EVENTS
719 atomic_long_t locked_vm
;
723 extern int uids_sysfs_init(void);
725 extern struct user_struct
*find_user(uid_t
);
727 extern struct user_struct root_user
;
728 #define INIT_USER (&root_user)
731 struct backing_dev_info
;
732 struct reclaim_state
;
734 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
736 /* cumulative counters */
737 unsigned long pcount
; /* # of times run on this cpu */
738 unsigned long long run_delay
; /* time spent waiting on a runqueue */
741 unsigned long long last_arrival
,/* when we last ran on a cpu */
742 last_queued
; /* when we were last queued to run */
744 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
746 #ifdef CONFIG_TASK_DELAY_ACCT
747 struct task_delay_info
{
749 unsigned int flags
; /* Private per-task flags */
751 /* For each stat XXX, add following, aligned appropriately
753 * struct timespec XXX_start, XXX_end;
757 * Atomicity of updates to XXX_delay, XXX_count protected by
758 * single lock above (split into XXX_lock if contention is an issue).
762 * XXX_count is incremented on every XXX operation, the delay
763 * associated with the operation is added to XXX_delay.
764 * XXX_delay contains the accumulated delay time in nanoseconds.
766 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
767 u64 blkio_delay
; /* wait for sync block io completion */
768 u64 swapin_delay
; /* wait for swapin block io completion */
769 u32 blkio_count
; /* total count of the number of sync block */
770 /* io operations performed */
771 u32 swapin_count
; /* total count of the number of swapin block */
772 /* io operations performed */
774 struct timespec freepages_start
, freepages_end
;
775 u64 freepages_delay
; /* wait for memory reclaim */
776 u32 freepages_count
; /* total count of memory reclaim */
778 #endif /* CONFIG_TASK_DELAY_ACCT */
780 static inline int sched_info_on(void)
782 #ifdef CONFIG_SCHEDSTATS
784 #elif defined(CONFIG_TASK_DELAY_ACCT)
785 extern int delayacct_on
;
800 * Increase resolution of nice-level calculations for 64-bit architectures.
801 * The extra resolution improves shares distribution and load balancing of
802 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
803 * hierarchies, especially on larger systems. This is not a user-visible change
804 * and does not change the user-interface for setting shares/weights.
806 * We increase resolution only if we have enough bits to allow this increased
807 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
808 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
811 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */
812 # define SCHED_LOAD_RESOLUTION 10
813 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
814 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
816 # define SCHED_LOAD_RESOLUTION 0
817 # define scale_load(w) (w)
818 # define scale_load_down(w) (w)
821 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
822 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
825 * Increase resolution of cpu_power calculations
827 #define SCHED_POWER_SHIFT 10
828 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
831 * sched-domains (multiprocessor balancing) declarations:
834 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
835 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
836 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
837 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
838 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
839 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
840 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
841 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
842 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
843 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
844 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
845 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
846 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
847 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
849 enum powersavings_balance_level
{
850 POWERSAVINGS_BALANCE_NONE
= 0, /* No power saving load balance */
851 POWERSAVINGS_BALANCE_BASIC
, /* Fill one thread/core/package
852 * first for long running threads
854 POWERSAVINGS_BALANCE_WAKEUP
, /* Also bias task wakeups to semi-idle
855 * cpu package for power savings
857 MAX_POWERSAVINGS_BALANCE_LEVELS
860 extern int sched_mc_power_savings
, sched_smt_power_savings
;
862 static inline int sd_balance_for_mc_power(void)
864 if (sched_smt_power_savings
)
865 return SD_POWERSAVINGS_BALANCE
;
867 if (!sched_mc_power_savings
)
868 return SD_PREFER_SIBLING
;
873 static inline int sd_balance_for_package_power(void)
875 if (sched_mc_power_savings
| sched_smt_power_savings
)
876 return SD_POWERSAVINGS_BALANCE
;
878 return SD_PREFER_SIBLING
;
881 extern int __weak
arch_sd_sibiling_asym_packing(void);
884 * Optimise SD flags for power savings:
885 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
886 * Keep default SD flags if sched_{smt,mc}_power_saving=0
889 static inline int sd_power_saving_flags(void)
891 if (sched_mc_power_savings
| sched_smt_power_savings
)
892 return SD_BALANCE_NEWIDLE
;
897 struct sched_group_power
{
900 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
903 unsigned int power
, power_orig
;
907 struct sched_group
*next
; /* Must be a circular list */
910 unsigned int group_weight
;
911 struct sched_group_power
*sgp
;
914 * The CPUs this group covers.
916 * NOTE: this field is variable length. (Allocated dynamically
917 * by attaching extra space to the end of the structure,
918 * depending on how many CPUs the kernel has booted up with)
920 unsigned long cpumask
[0];
923 static inline struct cpumask
*sched_group_cpus(struct sched_group
*sg
)
925 return to_cpumask(sg
->cpumask
);
928 struct sched_domain_attr
{
929 int relax_domain_level
;
932 #define SD_ATTR_INIT (struct sched_domain_attr) { \
933 .relax_domain_level = -1, \
936 extern int sched_domain_level_max
;
938 struct sched_domain
{
939 /* These fields must be setup */
940 struct sched_domain
*parent
; /* top domain must be null terminated */
941 struct sched_domain
*child
; /* bottom domain must be null terminated */
942 struct sched_group
*groups
; /* the balancing groups of the domain */
943 unsigned long min_interval
; /* Minimum balance interval ms */
944 unsigned long max_interval
; /* Maximum balance interval ms */
945 unsigned int busy_factor
; /* less balancing by factor if busy */
946 unsigned int imbalance_pct
; /* No balance until over watermark */
947 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
948 unsigned int busy_idx
;
949 unsigned int idle_idx
;
950 unsigned int newidle_idx
;
951 unsigned int wake_idx
;
952 unsigned int forkexec_idx
;
953 unsigned int smt_gain
;
954 int flags
; /* See SD_* */
957 /* Runtime fields. */
958 unsigned long last_balance
; /* init to jiffies. units in jiffies */
959 unsigned int balance_interval
; /* initialise to 1. units in ms. */
960 unsigned int nr_balance_failed
; /* initialise to 0 */
964 #ifdef CONFIG_SCHEDSTATS
965 /* load_balance() stats */
966 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
967 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
968 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
969 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
970 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
971 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
972 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
973 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
975 /* Active load balancing */
976 unsigned int alb_count
;
977 unsigned int alb_failed
;
978 unsigned int alb_pushed
;
980 /* SD_BALANCE_EXEC stats */
981 unsigned int sbe_count
;
982 unsigned int sbe_balanced
;
983 unsigned int sbe_pushed
;
985 /* SD_BALANCE_FORK stats */
986 unsigned int sbf_count
;
987 unsigned int sbf_balanced
;
988 unsigned int sbf_pushed
;
990 /* try_to_wake_up() stats */
991 unsigned int ttwu_wake_remote
;
992 unsigned int ttwu_move_affine
;
993 unsigned int ttwu_move_balance
;
995 #ifdef CONFIG_SCHED_DEBUG
999 void *private; /* used during construction */
1000 struct rcu_head rcu
; /* used during destruction */
1003 unsigned int span_weight
;
1005 * Span of all CPUs in this domain.
1007 * NOTE: this field is variable length. (Allocated dynamically
1008 * by attaching extra space to the end of the structure,
1009 * depending on how many CPUs the kernel has booted up with)
1011 unsigned long span
[0];
1014 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
1016 return to_cpumask(sd
->span
);
1019 extern void partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1020 struct sched_domain_attr
*dattr_new
);
1022 /* Allocate an array of sched domains, for partition_sched_domains(). */
1023 cpumask_var_t
*alloc_sched_domains(unsigned int ndoms
);
1024 void free_sched_domains(cpumask_var_t doms
[], unsigned int ndoms
);
1026 /* Test a flag in parent sched domain */
1027 static inline int test_sd_parent(struct sched_domain
*sd
, int flag
)
1029 if (sd
->parent
&& (sd
->parent
->flags
& flag
))
1035 unsigned long default_scale_freq_power(struct sched_domain
*sd
, int cpu
);
1036 unsigned long default_scale_smt_power(struct sched_domain
*sd
, int cpu
);
1038 #else /* CONFIG_SMP */
1040 struct sched_domain_attr
;
1043 partition_sched_domains(int ndoms_new
, cpumask_var_t doms_new
[],
1044 struct sched_domain_attr
*dattr_new
)
1047 #endif /* !CONFIG_SMP */
1050 struct io_context
; /* See blkdev.h */
1053 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1054 extern void prefetch_stack(struct task_struct
*t
);
1056 static inline void prefetch_stack(struct task_struct
*t
) { }
1059 struct audit_context
; /* See audit.c */
1061 struct pipe_inode_info
;
1062 struct uts_namespace
;
1065 struct sched_domain
;
1070 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1071 #define WF_FORK 0x02 /* child wakeup after fork */
1072 #define WF_MIGRATED 0x04 /* internal use, task got migrated */
1074 #define ENQUEUE_WAKEUP 1
1075 #define ENQUEUE_HEAD 2
1077 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1079 #define ENQUEUE_WAKING 0
1082 #define DEQUEUE_SLEEP 1
1084 struct sched_class
{
1085 const struct sched_class
*next
;
1087 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1088 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1089 void (*yield_task
) (struct rq
*rq
);
1090 bool (*yield_to_task
) (struct rq
*rq
, struct task_struct
*p
, bool preempt
);
1092 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1094 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
1095 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
1098 int (*select_task_rq
)(struct task_struct
*p
, int sd_flag
, int flags
);
1100 void (*pre_schedule
) (struct rq
*this_rq
, struct task_struct
*task
);
1101 void (*post_schedule
) (struct rq
*this_rq
);
1102 void (*task_waking
) (struct task_struct
*task
);
1103 void (*task_woken
) (struct rq
*this_rq
, struct task_struct
*task
);
1105 void (*set_cpus_allowed
)(struct task_struct
*p
,
1106 const struct cpumask
*newmask
);
1108 void (*rq_online
)(struct rq
*rq
);
1109 void (*rq_offline
)(struct rq
*rq
);
1112 void (*set_curr_task
) (struct rq
*rq
);
1113 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
, int queued
);
1114 void (*task_fork
) (struct task_struct
*p
);
1116 void (*switched_from
) (struct rq
*this_rq
, struct task_struct
*task
);
1117 void (*switched_to
) (struct rq
*this_rq
, struct task_struct
*task
);
1118 void (*prio_changed
) (struct rq
*this_rq
, struct task_struct
*task
,
1121 unsigned int (*get_rr_interval
) (struct rq
*rq
,
1122 struct task_struct
*task
);
1124 #ifdef CONFIG_FAIR_GROUP_SCHED
1125 void (*task_move_group
) (struct task_struct
*p
, int on_rq
);
1129 struct load_weight
{
1130 unsigned long weight
, inv_weight
;
1133 #ifdef CONFIG_SCHEDSTATS
1134 struct sched_statistics
{
1144 s64 sum_sleep_runtime
;
1151 u64 nr_migrations_cold
;
1152 u64 nr_failed_migrations_affine
;
1153 u64 nr_failed_migrations_running
;
1154 u64 nr_failed_migrations_hot
;
1155 u64 nr_forced_migrations
;
1158 u64 nr_wakeups_sync
;
1159 u64 nr_wakeups_migrate
;
1160 u64 nr_wakeups_local
;
1161 u64 nr_wakeups_remote
;
1162 u64 nr_wakeups_affine
;
1163 u64 nr_wakeups_affine_attempts
;
1164 u64 nr_wakeups_passive
;
1165 u64 nr_wakeups_idle
;
1169 struct sched_entity
{
1170 struct load_weight load
; /* for load-balancing */
1171 struct rb_node run_node
;
1172 struct list_head group_node
;
1176 u64 sum_exec_runtime
;
1178 u64 prev_sum_exec_runtime
;
1182 #ifdef CONFIG_SCHEDSTATS
1183 struct sched_statistics statistics
;
1186 #ifdef CONFIG_FAIR_GROUP_SCHED
1187 struct sched_entity
*parent
;
1188 /* rq on which this entity is (to be) queued: */
1189 struct cfs_rq
*cfs_rq
;
1190 /* rq "owned" by this entity/group: */
1191 struct cfs_rq
*my_q
;
1195 struct sched_rt_entity
{
1196 struct list_head run_list
;
1197 unsigned long timeout
;
1198 unsigned int time_slice
;
1199 int nr_cpus_allowed
;
1201 struct sched_rt_entity
*back
;
1202 #ifdef CONFIG_RT_GROUP_SCHED
1203 struct sched_rt_entity
*parent
;
1204 /* rq on which this entity is (to be) queued: */
1205 struct rt_rq
*rt_rq
;
1206 /* rq "owned" by this entity/group: */
1213 enum perf_event_task_context
{
1214 perf_invalid_context
= -1,
1215 perf_hw_context
= 0,
1217 perf_nr_task_contexts
,
1220 struct task_struct
{
1221 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1224 unsigned int flags
; /* per process flags, defined below */
1225 unsigned int ptrace
;
1228 struct task_struct
*wake_entry
;
1233 int prio
, static_prio
, normal_prio
;
1234 unsigned int rt_priority
;
1235 const struct sched_class
*sched_class
;
1236 struct sched_entity se
;
1237 struct sched_rt_entity rt
;
1239 #ifdef CONFIG_PREEMPT_NOTIFIERS
1240 /* list of struct preempt_notifier: */
1241 struct hlist_head preempt_notifiers
;
1245 * fpu_counter contains the number of consecutive context switches
1246 * that the FPU is used. If this is over a threshold, the lazy fpu
1247 * saving becomes unlazy to save the trap. This is an unsigned char
1248 * so that after 256 times the counter wraps and the behavior turns
1249 * lazy again; this to deal with bursty apps that only use FPU for
1252 unsigned char fpu_counter
;
1253 #ifdef CONFIG_BLK_DEV_IO_TRACE
1254 unsigned int btrace_seq
;
1257 unsigned int policy
;
1258 cpumask_t cpus_allowed
;
1260 #ifdef CONFIG_PREEMPT_RCU
1261 int rcu_read_lock_nesting
;
1262 char rcu_read_unlock_special
;
1263 #if defined(CONFIG_RCU_BOOST) && defined(CONFIG_TREE_PREEMPT_RCU)
1265 #endif /* #if defined(CONFIG_RCU_BOOST) && defined(CONFIG_TREE_PREEMPT_RCU) */
1266 struct list_head rcu_node_entry
;
1267 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1268 #ifdef CONFIG_TREE_PREEMPT_RCU
1269 struct rcu_node
*rcu_blocked_node
;
1270 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1271 #ifdef CONFIG_RCU_BOOST
1272 struct rt_mutex
*rcu_boost_mutex
;
1273 #endif /* #ifdef CONFIG_RCU_BOOST */
1275 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1276 struct sched_info sched_info
;
1279 struct list_head tasks
;
1281 struct plist_node pushable_tasks
;
1284 struct mm_struct
*mm
, *active_mm
;
1285 #ifdef CONFIG_COMPAT_BRK
1286 unsigned brk_randomized
:1;
1288 #if defined(SPLIT_RSS_COUNTING)
1289 struct task_rss_stat rss_stat
;
1293 int exit_code
, exit_signal
;
1294 int pdeath_signal
; /* The signal sent when the parent dies */
1295 unsigned int group_stop
; /* GROUP_STOP_*, siglock protected */
1297 unsigned int personality
;
1298 unsigned did_exec
:1;
1299 unsigned in_execve
:1; /* Tell the LSMs that the process is doing an
1301 unsigned in_iowait
:1;
1304 /* Revert to default priority/policy when forking */
1305 unsigned sched_reset_on_fork
:1;
1306 unsigned sched_contributes_to_load
:1;
1311 #ifdef CONFIG_CC_STACKPROTECTOR
1312 /* Canary value for the -fstack-protector gcc feature */
1313 unsigned long stack_canary
;
1317 * pointers to (original) parent process, youngest child, younger sibling,
1318 * older sibling, respectively. (p->father can be replaced with
1319 * p->real_parent->pid)
1321 struct task_struct
*real_parent
; /* real parent process */
1322 struct task_struct
*parent
; /* recipient of SIGCHLD, wait4() reports */
1324 * children/sibling forms the list of my natural children
1326 struct list_head children
; /* list of my children */
1327 struct list_head sibling
; /* linkage in my parent's children list */
1328 struct task_struct
*group_leader
; /* threadgroup leader */
1331 * ptraced is the list of tasks this task is using ptrace on.
1332 * This includes both natural children and PTRACE_ATTACH targets.
1333 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1335 struct list_head ptraced
;
1336 struct list_head ptrace_entry
;
1338 /* PID/PID hash table linkage. */
1339 struct pid_link pids
[PIDTYPE_MAX
];
1340 struct list_head thread_group
;
1342 struct completion
*vfork_done
; /* for vfork() */
1343 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1344 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1346 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1348 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1349 cputime_t prev_utime
, prev_stime
;
1351 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1352 struct timespec start_time
; /* monotonic time */
1353 struct timespec real_start_time
; /* boot based time */
1354 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1355 unsigned long min_flt
, maj_flt
;
1357 struct task_cputime cputime_expires
;
1358 struct list_head cpu_timers
[3];
1360 /* process credentials */
1361 const struct cred __rcu
*real_cred
; /* objective and real subjective task
1362 * credentials (COW) */
1363 const struct cred __rcu
*cred
; /* effective (overridable) subjective task
1364 * credentials (COW) */
1365 struct cred
*replacement_session_keyring
; /* for KEYCTL_SESSION_TO_PARENT */
1367 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1368 - access with [gs]et_task_comm (which lock
1369 it with task_lock())
1370 - initialized normally by setup_new_exec */
1371 /* file system info */
1372 int link_count
, total_link_count
;
1373 #ifdef CONFIG_SYSVIPC
1375 struct sysv_sem sysvsem
;
1377 #ifdef CONFIG_DETECT_HUNG_TASK
1378 /* hung task detection */
1379 unsigned long last_switch_count
;
1381 /* CPU-specific state of this task */
1382 struct thread_struct thread
;
1383 /* filesystem information */
1384 struct fs_struct
*fs
;
1385 /* open file information */
1386 struct files_struct
*files
;
1388 struct nsproxy
*nsproxy
;
1389 /* signal handlers */
1390 struct signal_struct
*signal
;
1391 struct sighand_struct
*sighand
;
1393 sigset_t blocked
, real_blocked
;
1394 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1395 struct sigpending pending
;
1397 unsigned long sas_ss_sp
;
1399 int (*notifier
)(void *priv
);
1400 void *notifier_data
;
1401 sigset_t
*notifier_mask
;
1402 struct audit_context
*audit_context
;
1403 #ifdef CONFIG_AUDITSYSCALL
1405 unsigned int sessionid
;
1409 /* Thread group tracking */
1412 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1414 spinlock_t alloc_lock
;
1416 #ifdef CONFIG_GENERIC_HARDIRQS
1417 /* IRQ handler threads */
1418 struct irqaction
*irqaction
;
1421 /* Protection of the PI data structures: */
1422 raw_spinlock_t pi_lock
;
1424 #ifdef CONFIG_RT_MUTEXES
1425 /* PI waiters blocked on a rt_mutex held by this task */
1426 struct plist_head pi_waiters
;
1427 /* Deadlock detection and priority inheritance handling */
1428 struct rt_mutex_waiter
*pi_blocked_on
;
1431 #ifdef CONFIG_DEBUG_MUTEXES
1432 /* mutex deadlock detection */
1433 struct mutex_waiter
*blocked_on
;
1435 #ifdef CONFIG_TRACE_IRQFLAGS
1436 unsigned int irq_events
;
1437 unsigned long hardirq_enable_ip
;
1438 unsigned long hardirq_disable_ip
;
1439 unsigned int hardirq_enable_event
;
1440 unsigned int hardirq_disable_event
;
1441 int hardirqs_enabled
;
1442 int hardirq_context
;
1443 unsigned long softirq_disable_ip
;
1444 unsigned long softirq_enable_ip
;
1445 unsigned int softirq_disable_event
;
1446 unsigned int softirq_enable_event
;
1447 int softirqs_enabled
;
1448 int softirq_context
;
1450 #ifdef CONFIG_LOCKDEP
1451 # define MAX_LOCK_DEPTH 48UL
1454 unsigned int lockdep_recursion
;
1455 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1456 gfp_t lockdep_reclaim_gfp
;
1459 /* journalling filesystem info */
1462 /* stacked block device info */
1463 struct bio_list
*bio_list
;
1466 /* stack plugging */
1467 struct blk_plug
*plug
;
1471 struct reclaim_state
*reclaim_state
;
1473 struct backing_dev_info
*backing_dev_info
;
1475 struct io_context
*io_context
;
1477 unsigned long ptrace_message
;
1478 siginfo_t
*last_siginfo
; /* For ptrace use. */
1479 struct task_io_accounting ioac
;
1480 #if defined(CONFIG_TASK_XACCT)
1481 u64 acct_rss_mem1
; /* accumulated rss usage */
1482 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1483 cputime_t acct_timexpd
; /* stime + utime since last update */
1485 #ifdef CONFIG_CPUSETS
1486 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1487 int mems_allowed_change_disable
;
1488 int cpuset_mem_spread_rotor
;
1489 int cpuset_slab_spread_rotor
;
1491 #ifdef CONFIG_CGROUPS
1492 /* Control Group info protected by css_set_lock */
1493 struct css_set __rcu
*cgroups
;
1494 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1495 struct list_head cg_list
;
1498 struct robust_list_head __user
*robust_list
;
1499 #ifdef CONFIG_COMPAT
1500 struct compat_robust_list_head __user
*compat_robust_list
;
1502 struct list_head pi_state_list
;
1503 struct futex_pi_state
*pi_state_cache
;
1505 #ifdef CONFIG_PERF_EVENTS
1506 struct perf_event_context
*perf_event_ctxp
[perf_nr_task_contexts
];
1507 struct mutex perf_event_mutex
;
1508 struct list_head perf_event_list
;
1511 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1513 short pref_node_fork
;
1515 atomic_t fs_excl
; /* holding fs exclusive resources */
1516 struct rcu_head rcu
;
1519 * cache last used pipe for splice
1521 struct pipe_inode_info
*splice_pipe
;
1522 #ifdef CONFIG_TASK_DELAY_ACCT
1523 struct task_delay_info
*delays
;
1525 #ifdef CONFIG_FAULT_INJECTION
1528 struct prop_local_single dirties
;
1529 #ifdef CONFIG_LATENCYTOP
1530 int latency_record_count
;
1531 struct latency_record latency_record
[LT_SAVECOUNT
];
1534 * time slack values; these are used to round up poll() and
1535 * select() etc timeout values. These are in nanoseconds.
1537 unsigned long timer_slack_ns
;
1538 unsigned long default_timer_slack_ns
;
1540 struct list_head
*scm_work_list
;
1541 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1542 /* Index of current stored address in ret_stack */
1544 /* Stack of return addresses for return function tracing */
1545 struct ftrace_ret_stack
*ret_stack
;
1546 /* time stamp for last schedule */
1547 unsigned long long ftrace_timestamp
;
1549 * Number of functions that haven't been traced
1550 * because of depth overrun.
1552 atomic_t trace_overrun
;
1553 /* Pause for the tracing */
1554 atomic_t tracing_graph_pause
;
1556 #ifdef CONFIG_TRACING
1557 /* state flags for use by tracers */
1558 unsigned long trace
;
1559 /* bitmask and counter of trace recursion */
1560 unsigned long trace_recursion
;
1561 #endif /* CONFIG_TRACING */
1562 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1563 struct memcg_batch_info
{
1564 int do_batch
; /* incremented when batch uncharge started */
1565 struct mem_cgroup
*memcg
; /* target memcg of uncharge */
1566 unsigned long nr_pages
; /* uncharged usage */
1567 unsigned long memsw_nr_pages
; /* uncharged mem+swap usage */
1570 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1571 atomic_t ptrace_bp_refcnt
;
1575 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1576 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1579 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1580 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1581 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1582 * values are inverted: lower p->prio value means higher priority.
1584 * The MAX_USER_RT_PRIO value allows the actual maximum
1585 * RT priority to be separate from the value exported to
1586 * user-space. This allows kernel threads to set their
1587 * priority to a value higher than any user task. Note:
1588 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1591 #define MAX_USER_RT_PRIO 100
1592 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1594 #define MAX_PRIO (MAX_RT_PRIO + 40)
1595 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1597 static inline int rt_prio(int prio
)
1599 if (unlikely(prio
< MAX_RT_PRIO
))
1604 static inline int rt_task(struct task_struct
*p
)
1606 return rt_prio(p
->prio
);
1609 static inline struct pid
*task_pid(struct task_struct
*task
)
1611 return task
->pids
[PIDTYPE_PID
].pid
;
1614 static inline struct pid
*task_tgid(struct task_struct
*task
)
1616 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1620 * Without tasklist or rcu lock it is not safe to dereference
1621 * the result of task_pgrp/task_session even if task == current,
1622 * we can race with another thread doing sys_setsid/sys_setpgid.
1624 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1626 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1629 static inline struct pid
*task_session(struct task_struct
*task
)
1631 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1634 struct pid_namespace
;
1637 * the helpers to get the task's different pids as they are seen
1638 * from various namespaces
1640 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1641 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1643 * task_xid_nr_ns() : id seen from the ns specified;
1645 * set_task_vxid() : assigns a virtual id to a task;
1647 * see also pid_nr() etc in include/linux/pid.h
1649 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1650 struct pid_namespace
*ns
);
1652 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1657 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1658 struct pid_namespace
*ns
)
1660 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1663 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1665 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1669 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1674 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1676 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1678 return pid_vnr(task_tgid(tsk
));
1682 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1683 struct pid_namespace
*ns
)
1685 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1688 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1690 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1694 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1695 struct pid_namespace
*ns
)
1697 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1700 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1702 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1705 /* obsolete, do not use */
1706 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1708 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1712 * pid_alive - check that a task structure is not stale
1713 * @p: Task structure to be checked.
1715 * Test if a process is not yet dead (at most zombie state)
1716 * If pid_alive fails, then pointers within the task structure
1717 * can be stale and must not be dereferenced.
1719 static inline int pid_alive(struct task_struct
*p
)
1721 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1725 * is_global_init - check if a task structure is init
1726 * @tsk: Task structure to be checked.
1728 * Check if a task structure is the first user space task the kernel created.
1730 static inline int is_global_init(struct task_struct
*tsk
)
1732 return tsk
->pid
== 1;
1736 * is_container_init:
1737 * check whether in the task is init in its own pid namespace.
1739 extern int is_container_init(struct task_struct
*tsk
);
1741 extern struct pid
*cad_pid
;
1743 extern void free_task(struct task_struct
*tsk
);
1744 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1746 extern void __put_task_struct(struct task_struct
*t
);
1748 static inline void put_task_struct(struct task_struct
*t
)
1750 if (atomic_dec_and_test(&t
->usage
))
1751 __put_task_struct(t
);
1754 extern void task_times(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1755 extern void thread_group_times(struct task_struct
*p
, cputime_t
*ut
, cputime_t
*st
);
1760 #define PF_STARTING 0x00000002 /* being created */
1761 #define PF_EXITING 0x00000004 /* getting shut down */
1762 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1763 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1764 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1765 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1766 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1767 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1768 #define PF_DUMPCORE 0x00000200 /* dumped core */
1769 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1770 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1771 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1772 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1773 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1774 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1775 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1776 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1777 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1778 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1779 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1780 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1781 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1782 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1783 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1784 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1785 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1786 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1787 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1788 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1791 * Only the _current_ task can read/write to tsk->flags, but other
1792 * tasks can access tsk->flags in readonly mode for example
1793 * with tsk_used_math (like during threaded core dumping).
1794 * There is however an exception to this rule during ptrace
1795 * or during fork: the ptracer task is allowed to write to the
1796 * child->flags of its traced child (same goes for fork, the parent
1797 * can write to the child->flags), because we're guaranteed the
1798 * child is not running and in turn not changing child->flags
1799 * at the same time the parent does it.
1801 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1802 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1803 #define clear_used_math() clear_stopped_child_used_math(current)
1804 #define set_used_math() set_stopped_child_used_math(current)
1805 #define conditional_stopped_child_used_math(condition, child) \
1806 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1807 #define conditional_used_math(condition) \
1808 conditional_stopped_child_used_math(condition, current)
1809 #define copy_to_stopped_child_used_math(child) \
1810 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1811 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1812 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1813 #define used_math() tsk_used_math(current)
1816 * task->group_stop flags
1818 #define GROUP_STOP_SIGMASK 0xffff /* signr of the last group stop */
1819 #define GROUP_STOP_PENDING (1 << 16) /* task should stop for group stop */
1820 #define GROUP_STOP_CONSUME (1 << 17) /* consume group stop count */
1821 #define GROUP_STOP_TRAPPING (1 << 18) /* switching from STOPPED to TRACED */
1822 #define GROUP_STOP_DEQUEUED (1 << 19) /* stop signal dequeued */
1824 extern void task_clear_group_stop_pending(struct task_struct
*task
);
1826 #ifdef CONFIG_PREEMPT_RCU
1828 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1829 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1830 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1832 static inline void rcu_copy_process(struct task_struct
*p
)
1834 p
->rcu_read_lock_nesting
= 0;
1835 p
->rcu_read_unlock_special
= 0;
1836 #ifdef CONFIG_TREE_PREEMPT_RCU
1837 p
->rcu_blocked_node
= NULL
;
1838 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1839 #ifdef CONFIG_RCU_BOOST
1840 p
->rcu_boost_mutex
= NULL
;
1841 #endif /* #ifdef CONFIG_RCU_BOOST */
1842 INIT_LIST_HEAD(&p
->rcu_node_entry
);
1847 static inline void rcu_copy_process(struct task_struct
*p
)
1854 extern void do_set_cpus_allowed(struct task_struct
*p
,
1855 const struct cpumask
*new_mask
);
1857 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
1858 const struct cpumask
*new_mask
);
1860 static inline void do_set_cpus_allowed(struct task_struct
*p
,
1861 const struct cpumask
*new_mask
)
1864 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
1865 const struct cpumask
*new_mask
)
1867 if (!cpumask_test_cpu(0, new_mask
))
1873 #ifndef CONFIG_CPUMASK_OFFSTACK
1874 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1876 return set_cpus_allowed_ptr(p
, &new_mask
);
1881 * Do not use outside of architecture code which knows its limitations.
1883 * sched_clock() has no promise of monotonicity or bounded drift between
1884 * CPUs, use (which you should not) requires disabling IRQs.
1886 * Please use one of the three interfaces below.
1888 extern unsigned long long notrace
sched_clock(void);
1890 * See the comment in kernel/sched_clock.c
1892 extern u64
cpu_clock(int cpu
);
1893 extern u64
local_clock(void);
1894 extern u64
sched_clock_cpu(int cpu
);
1897 extern void sched_clock_init(void);
1899 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1900 static inline void sched_clock_tick(void)
1904 static inline void sched_clock_idle_sleep_event(void)
1908 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
1913 * Architectures can set this to 1 if they have specified
1914 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1915 * but then during bootup it turns out that sched_clock()
1916 * is reliable after all:
1918 extern int sched_clock_stable
;
1920 extern void sched_clock_tick(void);
1921 extern void sched_clock_idle_sleep_event(void);
1922 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1925 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1927 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1928 * The reason for this explicit opt-in is not to have perf penalty with
1929 * slow sched_clocks.
1931 extern void enable_sched_clock_irqtime(void);
1932 extern void disable_sched_clock_irqtime(void);
1934 static inline void enable_sched_clock_irqtime(void) {}
1935 static inline void disable_sched_clock_irqtime(void) {}
1938 extern unsigned long long
1939 task_sched_runtime(struct task_struct
*task
);
1940 extern unsigned long long thread_group_sched_runtime(struct task_struct
*task
);
1942 /* sched_exec is called by processes performing an exec */
1944 extern void sched_exec(void);
1946 #define sched_exec() {}
1949 extern void sched_clock_idle_sleep_event(void);
1950 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1952 #ifdef CONFIG_HOTPLUG_CPU
1953 extern void idle_task_exit(void);
1955 static inline void idle_task_exit(void) {}
1958 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1959 extern void wake_up_idle_cpu(int cpu
);
1961 static inline void wake_up_idle_cpu(int cpu
) { }
1964 extern unsigned int sysctl_sched_latency
;
1965 extern unsigned int sysctl_sched_min_granularity
;
1966 extern unsigned int sysctl_sched_wakeup_granularity
;
1967 extern unsigned int sysctl_sched_child_runs_first
;
1969 enum sched_tunable_scaling
{
1970 SCHED_TUNABLESCALING_NONE
,
1971 SCHED_TUNABLESCALING_LOG
,
1972 SCHED_TUNABLESCALING_LINEAR
,
1973 SCHED_TUNABLESCALING_END
,
1975 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling
;
1977 #ifdef CONFIG_SCHED_DEBUG
1978 extern unsigned int sysctl_sched_migration_cost
;
1979 extern unsigned int sysctl_sched_nr_migrate
;
1980 extern unsigned int sysctl_sched_time_avg
;
1981 extern unsigned int sysctl_timer_migration
;
1982 extern unsigned int sysctl_sched_shares_window
;
1984 int sched_proc_update_handler(struct ctl_table
*table
, int write
,
1985 void __user
*buffer
, size_t *length
,
1988 #ifdef CONFIG_SCHED_DEBUG
1989 static inline unsigned int get_sysctl_timer_migration(void)
1991 return sysctl_timer_migration
;
1994 static inline unsigned int get_sysctl_timer_migration(void)
1999 extern unsigned int sysctl_sched_rt_period
;
2000 extern int sysctl_sched_rt_runtime
;
2002 int sched_rt_handler(struct ctl_table
*table
, int write
,
2003 void __user
*buffer
, size_t *lenp
,
2006 #ifdef CONFIG_SCHED_AUTOGROUP
2007 extern unsigned int sysctl_sched_autogroup_enabled
;
2009 extern void sched_autogroup_create_attach(struct task_struct
*p
);
2010 extern void sched_autogroup_detach(struct task_struct
*p
);
2011 extern void sched_autogroup_fork(struct signal_struct
*sig
);
2012 extern void sched_autogroup_exit(struct signal_struct
*sig
);
2013 #ifdef CONFIG_PROC_FS
2014 extern void proc_sched_autogroup_show_task(struct task_struct
*p
, struct seq_file
*m
);
2015 extern int proc_sched_autogroup_set_nice(struct task_struct
*p
, int *nice
);
2018 static inline void sched_autogroup_create_attach(struct task_struct
*p
) { }
2019 static inline void sched_autogroup_detach(struct task_struct
*p
) { }
2020 static inline void sched_autogroup_fork(struct signal_struct
*sig
) { }
2021 static inline void sched_autogroup_exit(struct signal_struct
*sig
) { }
2024 #ifdef CONFIG_RT_MUTEXES
2025 extern int rt_mutex_getprio(struct task_struct
*p
);
2026 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
2027 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
2029 static inline int rt_mutex_getprio(struct task_struct
*p
)
2031 return p
->normal_prio
;
2033 # define rt_mutex_adjust_pi(p) do { } while (0)
2036 extern bool yield_to(struct task_struct
*p
, bool preempt
);
2037 extern void set_user_nice(struct task_struct
*p
, long nice
);
2038 extern int task_prio(const struct task_struct
*p
);
2039 extern int task_nice(const struct task_struct
*p
);
2040 extern int can_nice(const struct task_struct
*p
, const int nice
);
2041 extern int task_curr(const struct task_struct
*p
);
2042 extern int idle_cpu(int cpu
);
2043 extern int sched_setscheduler(struct task_struct
*, int,
2044 const struct sched_param
*);
2045 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
2046 const struct sched_param
*);
2047 extern struct task_struct
*idle_task(int cpu
);
2048 extern struct task_struct
*curr_task(int cpu
);
2049 extern void set_curr_task(int cpu
, struct task_struct
*p
);
2054 * The default (Linux) execution domain.
2056 extern struct exec_domain default_exec_domain
;
2058 union thread_union
{
2059 struct thread_info thread_info
;
2060 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
2063 #ifndef __HAVE_ARCH_KSTACK_END
2064 static inline int kstack_end(void *addr
)
2066 /* Reliable end of stack detection:
2067 * Some APM bios versions misalign the stack
2069 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
2073 extern union thread_union init_thread_union
;
2074 extern struct task_struct init_task
;
2076 extern struct mm_struct init_mm
;
2078 extern struct pid_namespace init_pid_ns
;
2081 * find a task by one of its numerical ids
2083 * find_task_by_pid_ns():
2084 * finds a task by its pid in the specified namespace
2085 * find_task_by_vpid():
2086 * finds a task by its virtual pid
2088 * see also find_vpid() etc in include/linux/pid.h
2091 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
2092 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
2093 struct pid_namespace
*ns
);
2095 extern void __set_special_pids(struct pid
*pid
);
2097 /* per-UID process charging. */
2098 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
2099 static inline struct user_struct
*get_uid(struct user_struct
*u
)
2101 atomic_inc(&u
->__count
);
2104 extern void free_uid(struct user_struct
*);
2105 extern void release_uids(struct user_namespace
*ns
);
2107 #include <asm/current.h>
2109 extern void xtime_update(unsigned long ticks
);
2111 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
2112 extern int wake_up_process(struct task_struct
*tsk
);
2113 extern void wake_up_new_task(struct task_struct
*tsk
);
2115 extern void kick_process(struct task_struct
*tsk
);
2117 static inline void kick_process(struct task_struct
*tsk
) { }
2119 extern void sched_fork(struct task_struct
*p
);
2120 extern void sched_dead(struct task_struct
*p
);
2122 extern void proc_caches_init(void);
2123 extern void flush_signals(struct task_struct
*);
2124 extern void __flush_signals(struct task_struct
*);
2125 extern void ignore_signals(struct task_struct
*);
2126 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2127 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2129 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
2131 unsigned long flags
;
2134 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
2135 ret
= dequeue_signal(tsk
, mask
, info
);
2136 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
2141 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
2143 extern void unblock_all_signals(void);
2144 extern void release_task(struct task_struct
* p
);
2145 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2146 extern int force_sigsegv(int, struct task_struct
*);
2147 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2148 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2149 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2150 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
2151 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2152 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2153 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2154 extern int do_notify_parent(struct task_struct
*, int);
2155 extern void __wake_up_parent(struct task_struct
*p
, struct task_struct
*parent
);
2156 extern void force_sig(int, struct task_struct
*);
2157 extern int send_sig(int, struct task_struct
*, int);
2158 extern int zap_other_threads(struct task_struct
*p
);
2159 extern struct sigqueue
*sigqueue_alloc(void);
2160 extern void sigqueue_free(struct sigqueue
*);
2161 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2162 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2163 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
2165 static inline int kill_cad_pid(int sig
, int priv
)
2167 return kill_pid(cad_pid
, sig
, priv
);
2170 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2171 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2172 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2173 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2176 * True if we are on the alternate signal stack.
2178 static inline int on_sig_stack(unsigned long sp
)
2180 #ifdef CONFIG_STACK_GROWSUP
2181 return sp
>= current
->sas_ss_sp
&&
2182 sp
- current
->sas_ss_sp
< current
->sas_ss_size
;
2184 return sp
> current
->sas_ss_sp
&&
2185 sp
- current
->sas_ss_sp
<= current
->sas_ss_size
;
2189 static inline int sas_ss_flags(unsigned long sp
)
2191 return (current
->sas_ss_size
== 0 ? SS_DISABLE
2192 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
2196 * Routines for handling mm_structs
2198 extern struct mm_struct
* mm_alloc(void);
2200 /* mmdrop drops the mm and the page tables */
2201 extern void __mmdrop(struct mm_struct
*);
2202 static inline void mmdrop(struct mm_struct
* mm
)
2204 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2208 /* mmput gets rid of the mappings and all user-space */
2209 extern void mmput(struct mm_struct
*);
2210 /* Grab a reference to a task's mm, if it is not already going away */
2211 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2212 /* Remove the current tasks stale references to the old mm_struct */
2213 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2214 /* Allocate a new mm structure and copy contents from tsk->mm */
2215 extern struct mm_struct
*dup_mm(struct task_struct
*tsk
);
2217 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2218 struct task_struct
*, struct pt_regs
*);
2219 extern void flush_thread(void);
2220 extern void exit_thread(void);
2222 extern void exit_files(struct task_struct
*);
2223 extern void __cleanup_sighand(struct sighand_struct
*);
2225 extern void exit_itimers(struct signal_struct
*);
2226 extern void flush_itimer_signals(void);
2228 extern NORET_TYPE
void do_group_exit(int);
2230 extern void daemonize(const char *, ...);
2231 extern int allow_signal(int);
2232 extern int disallow_signal(int);
2234 extern int do_execve(const char *,
2235 const char __user
* const __user
*,
2236 const char __user
* const __user
*, struct pt_regs
*);
2237 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
2238 struct task_struct
*fork_idle(int);
2240 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
2241 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2244 void scheduler_ipi(void);
2245 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2247 static inline void scheduler_ipi(void) { }
2248 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2255 #define next_task(p) \
2256 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2258 #define for_each_process(p) \
2259 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2261 extern bool current_is_single_threaded(void);
2264 * Careful: do_each_thread/while_each_thread is a double loop so
2265 * 'break' will not work as expected - use goto instead.
2267 #define do_each_thread(g, t) \
2268 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2270 #define while_each_thread(g, t) \
2271 while ((t = next_thread(t)) != g)
2273 static inline int get_nr_threads(struct task_struct
*tsk
)
2275 return tsk
->signal
->nr_threads
;
2278 /* de_thread depends on thread_group_leader not being a pid based check */
2279 #define thread_group_leader(p) (p == p->group_leader)
2281 /* Do to the insanities of de_thread it is possible for a process
2282 * to have the pid of the thread group leader without actually being
2283 * the thread group leader. For iteration through the pids in proc
2284 * all we care about is that we have a task with the appropriate
2285 * pid, we don't actually care if we have the right task.
2287 static inline int has_group_leader_pid(struct task_struct
*p
)
2289 return p
->pid
== p
->tgid
;
2293 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2295 return p1
->tgid
== p2
->tgid
;
2298 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2300 return list_entry_rcu(p
->thread_group
.next
,
2301 struct task_struct
, thread_group
);
2304 static inline int thread_group_empty(struct task_struct
*p
)
2306 return list_empty(&p
->thread_group
);
2309 #define delay_group_leader(p) \
2310 (thread_group_leader(p) && !thread_group_empty(p))
2312 static inline int task_detached(struct task_struct
*p
)
2314 return p
->exit_signal
== -1;
2318 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2319 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2320 * pins the final release of task.io_context. Also protects ->cpuset and
2321 * ->cgroup.subsys[].
2323 * Nests both inside and outside of read_lock(&tasklist_lock).
2324 * It must not be nested with write_lock_irq(&tasklist_lock),
2325 * neither inside nor outside.
2327 static inline void task_lock(struct task_struct
*p
)
2329 spin_lock(&p
->alloc_lock
);
2332 static inline void task_unlock(struct task_struct
*p
)
2334 spin_unlock(&p
->alloc_lock
);
2337 extern struct sighand_struct
*__lock_task_sighand(struct task_struct
*tsk
,
2338 unsigned long *flags
);
2340 #define lock_task_sighand(tsk, flags) \
2341 ({ struct sighand_struct *__ss; \
2342 __cond_lock(&(tsk)->sighand->siglock, \
2343 (__ss = __lock_task_sighand(tsk, flags))); \
2347 static inline void unlock_task_sighand(struct task_struct *tsk,
2348 unsigned long *flags
)
2350 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2353 /* See the declaration of threadgroup_fork_lock in signal_struct. */
2354 #ifdef CONFIG_CGROUPS
2355 static inline void threadgroup_fork_read_lock(struct task_struct
*tsk
)
2357 down_read(&tsk
->signal
->threadgroup_fork_lock
);
2359 static inline void threadgroup_fork_read_unlock(struct task_struct
*tsk
)
2361 up_read(&tsk
->signal
->threadgroup_fork_lock
);
2363 static inline void threadgroup_fork_write_lock(struct task_struct
*tsk
)
2365 down_write(&tsk
->signal
->threadgroup_fork_lock
);
2367 static inline void threadgroup_fork_write_unlock(struct task_struct
*tsk
)
2369 up_write(&tsk
->signal
->threadgroup_fork_lock
);
2372 static inline void threadgroup_fork_read_lock(struct task_struct
*tsk
) {}
2373 static inline void threadgroup_fork_read_unlock(struct task_struct
*tsk
) {}
2374 static inline void threadgroup_fork_write_lock(struct task_struct
*tsk
) {}
2375 static inline void threadgroup_fork_write_unlock(struct task_struct
*tsk
) {}
2378 #ifndef __HAVE_THREAD_FUNCTIONS
2380 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2381 #define task_stack_page(task) ((task)->stack)
2383 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2385 *task_thread_info(p
) = *task_thread_info(org
);
2386 task_thread_info(p
)->task
= p
;
2389 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2391 return (unsigned long *)(task_thread_info(p
) + 1);
2396 static inline int object_is_on_stack(void *obj
)
2398 void *stack
= task_stack_page(current
);
2400 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2403 extern void thread_info_cache_init(void);
2405 #ifdef CONFIG_DEBUG_STACK_USAGE
2406 static inline unsigned long stack_not_used(struct task_struct
*p
)
2408 unsigned long *n
= end_of_stack(p
);
2410 do { /* Skip over canary */
2414 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2418 /* set thread flags in other task's structures
2419 * - see asm/thread_info.h for TIF_xxxx flags available
2421 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2423 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2426 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2428 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2431 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2433 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2436 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2438 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2441 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2443 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2446 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2448 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2451 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2453 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2456 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2458 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2461 static inline int restart_syscall(void)
2463 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2464 return -ERESTARTNOINTR
;
2467 static inline int signal_pending(struct task_struct
*p
)
2469 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2472 static inline int __fatal_signal_pending(struct task_struct
*p
)
2474 return unlikely(sigismember(&p
->pending
.signal
, SIGKILL
));
2477 static inline int fatal_signal_pending(struct task_struct
*p
)
2479 return signal_pending(p
) && __fatal_signal_pending(p
);
2482 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2484 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2486 if (!signal_pending(p
))
2489 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2492 static inline int need_resched(void)
2494 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
2498 * cond_resched() and cond_resched_lock(): latency reduction via
2499 * explicit rescheduling in places that are safe. The return
2500 * value indicates whether a reschedule was done in fact.
2501 * cond_resched_lock() will drop the spinlock before scheduling,
2502 * cond_resched_softirq() will enable bhs before scheduling.
2504 extern int _cond_resched(void);
2506 #define cond_resched() ({ \
2507 __might_sleep(__FILE__, __LINE__, 0); \
2511 extern int __cond_resched_lock(spinlock_t
*lock
);
2513 #ifdef CONFIG_PREEMPT_COUNT
2514 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2516 #define PREEMPT_LOCK_OFFSET 0
2519 #define cond_resched_lock(lock) ({ \
2520 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2521 __cond_resched_lock(lock); \
2524 extern int __cond_resched_softirq(void);
2526 #define cond_resched_softirq() ({ \
2527 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2528 __cond_resched_softirq(); \
2532 * Does a critical section need to be broken due to another
2533 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2534 * but a general need for low latency)
2536 static inline int spin_needbreak(spinlock_t
*lock
)
2538 #ifdef CONFIG_PREEMPT
2539 return spin_is_contended(lock
);
2546 * Thread group CPU time accounting.
2548 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2549 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2551 static inline void thread_group_cputime_init(struct signal_struct
*sig
)
2553 spin_lock_init(&sig
->cputimer
.lock
);
2557 * Reevaluate whether the task has signals pending delivery.
2558 * Wake the task if so.
2559 * This is required every time the blocked sigset_t changes.
2560 * callers must hold sighand->siglock.
2562 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2563 extern void recalc_sigpending(void);
2565 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
2568 * Wrappers for p->thread_info->cpu access. No-op on UP.
2572 static inline unsigned int task_cpu(const struct task_struct
*p
)
2574 return task_thread_info(p
)->cpu
;
2577 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2581 static inline unsigned int task_cpu(const struct task_struct
*p
)
2586 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2590 #endif /* CONFIG_SMP */
2592 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
2593 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
2595 extern void normalize_rt_tasks(void);
2597 #ifdef CONFIG_CGROUP_SCHED
2599 extern struct task_group root_task_group
;
2601 extern struct task_group
*sched_create_group(struct task_group
*parent
);
2602 extern void sched_destroy_group(struct task_group
*tg
);
2603 extern void sched_move_task(struct task_struct
*tsk
);
2604 #ifdef CONFIG_FAIR_GROUP_SCHED
2605 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
2606 extern unsigned long sched_group_shares(struct task_group
*tg
);
2608 #ifdef CONFIG_RT_GROUP_SCHED
2609 extern int sched_group_set_rt_runtime(struct task_group
*tg
,
2610 long rt_runtime_us
);
2611 extern long sched_group_rt_runtime(struct task_group
*tg
);
2612 extern int sched_group_set_rt_period(struct task_group
*tg
,
2614 extern long sched_group_rt_period(struct task_group
*tg
);
2615 extern int sched_rt_can_attach(struct task_group
*tg
, struct task_struct
*tsk
);
2619 extern int task_can_switch_user(struct user_struct
*up
,
2620 struct task_struct
*tsk
);
2622 #ifdef CONFIG_TASK_XACCT
2623 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2625 tsk
->ioac
.rchar
+= amt
;
2628 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2630 tsk
->ioac
.wchar
+= amt
;
2633 static inline void inc_syscr(struct task_struct
*tsk
)
2638 static inline void inc_syscw(struct task_struct
*tsk
)
2643 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2647 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2651 static inline void inc_syscr(struct task_struct
*tsk
)
2655 static inline void inc_syscw(struct task_struct
*tsk
)
2660 #ifndef TASK_SIZE_OF
2661 #define TASK_SIZE_OF(tsk) TASK_SIZE
2664 #ifdef CONFIG_MM_OWNER
2665 extern void mm_update_next_owner(struct mm_struct
*mm
);
2666 extern void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
);
2668 static inline void mm_update_next_owner(struct mm_struct
*mm
)
2672 static inline void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
)
2675 #endif /* CONFIG_MM_OWNER */
2677 static inline unsigned long task_rlimit(const struct task_struct
*tsk
,
2680 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_cur
);
2683 static inline unsigned long task_rlimit_max(const struct task_struct
*tsk
,
2686 return ACCESS_ONCE(tsk
->signal
->rlim
[limit
].rlim_max
);
2689 static inline unsigned long rlimit(unsigned int limit
)
2691 return task_rlimit(current
, limit
);
2694 static inline unsigned long rlimit_max(unsigned int limit
)
2696 return task_rlimit_max(current
, limit
);
2699 #endif /* __KERNEL__ */