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 #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 #define CLONE_IO 0x80000000 /* Clone io context */
35 #define SCHED_NORMAL 0
39 /* SCHED_ISO: reserved but not implemented yet */
41 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 #define SCHED_RESET_ON_FORK 0x40000000
50 #include <asm/param.h> /* for HZ */
52 #include <linux/capability.h>
53 #include <linux/threads.h>
54 #include <linux/kernel.h>
55 #include <linux/types.h>
56 #include <linux/timex.h>
57 #include <linux/jiffies.h>
58 #include <linux/rbtree.h>
59 #include <linux/thread_info.h>
60 #include <linux/cpumask.h>
61 #include <linux/errno.h>
62 #include <linux/nodemask.h>
63 #include <linux/mm_types.h>
65 #include <asm/system.h>
67 #include <asm/ptrace.h>
68 #include <asm/cputime.h>
70 #include <linux/smp.h>
71 #include <linux/sem.h>
72 #include <linux/signal.h>
73 #include <linux/path.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/kobject.h>
92 #include <linux/latencytop.h>
93 #include <linux/cred.h>
95 #include <asm/processor.h>
98 struct futex_pi_state
;
99 struct robust_list_head
;
103 struct perf_event_context
;
106 * List of flags we want to share for kernel threads,
107 * if only because they are not used by them anyway.
109 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
112 * These are the constant used to fake the fixed-point load-average
113 * counting. Some notes:
114 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
115 * a load-average precision of 10 bits integer + 11 bits fractional
116 * - if you want to count load-averages more often, you need more
117 * precision, or rounding will get you. With 2-second counting freq,
118 * the EXP_n values would be 1981, 2034 and 2043 if still using only
121 extern unsigned long avenrun
[]; /* Load averages */
122 extern void get_avenrun(unsigned long *loads
, unsigned long offset
, int shift
);
124 #define FSHIFT 11 /* nr of bits of precision */
125 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
126 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
127 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
128 #define EXP_5 2014 /* 1/exp(5sec/5min) */
129 #define EXP_15 2037 /* 1/exp(5sec/15min) */
131 #define CALC_LOAD(load,exp,n) \
133 load += n*(FIXED_1-exp); \
136 extern unsigned long total_forks
;
137 extern int nr_threads
;
138 DECLARE_PER_CPU(unsigned long, process_counts
);
139 extern int nr_processes(void);
140 extern unsigned long nr_running(void);
141 extern unsigned long nr_uninterruptible(void);
142 extern unsigned long nr_iowait(void);
143 extern void calc_global_load(void);
144 extern u64
cpu_nr_migrations(int cpu
);
146 extern unsigned long get_parent_ip(unsigned long addr
);
151 #ifdef CONFIG_SCHED_DEBUG
152 extern void proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
);
153 extern void proc_sched_set_task(struct task_struct
*p
);
155 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
);
158 proc_sched_show_task(struct task_struct
*p
, struct seq_file
*m
)
161 static inline void proc_sched_set_task(struct task_struct
*p
)
165 print_cfs_rq(struct seq_file
*m
, int cpu
, struct cfs_rq
*cfs_rq
)
170 extern unsigned long long time_sync_thresh
;
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
195 /* Convenience macros for the sake of set_task_state */
196 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
197 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
198 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
200 /* Convenience macros for the sake of wake_up */
201 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
202 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
204 /* get_task_state() */
205 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
206 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
209 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
210 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
211 #define task_is_stopped_or_traced(task) \
212 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
213 #define task_contributes_to_load(task) \
214 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
215 (task->flags & PF_FREEZING) == 0)
217 #define __set_task_state(tsk, state_value) \
218 do { (tsk)->state = (state_value); } while (0)
219 #define set_task_state(tsk, state_value) \
220 set_mb((tsk)->state, (state_value))
223 * set_current_state() includes a barrier so that the write of current->state
224 * is correctly serialised wrt the caller's subsequent test of whether to
227 * set_current_state(TASK_UNINTERRUPTIBLE);
228 * if (do_i_need_to_sleep())
231 * If the caller does not need such serialisation then use __set_current_state()
233 #define __set_current_state(state_value) \
234 do { current->state = (state_value); } while (0)
235 #define set_current_state(state_value) \
236 set_mb(current->state, (state_value))
238 /* Task command name length */
239 #define TASK_COMM_LEN 16
241 #include <linux/spinlock.h>
244 * This serializes "schedule()" and also protects
245 * the run-queue from deletions/modifications (but
246 * _adding_ to the beginning of the run-queue has
249 extern rwlock_t tasklist_lock
;
250 extern spinlock_t mmlist_lock
;
254 extern void sched_init(void);
255 extern void sched_init_smp(void);
256 extern asmlinkage
void schedule_tail(struct task_struct
*prev
);
257 extern void init_idle(struct task_struct
*idle
, int cpu
);
258 extern void init_idle_bootup_task(struct task_struct
*idle
);
260 extern int runqueue_is_locked(int cpu
);
261 extern void task_rq_unlock_wait(struct task_struct
*p
);
263 extern cpumask_var_t nohz_cpu_mask
;
264 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
265 extern int select_nohz_load_balancer(int cpu
);
266 extern int get_nohz_load_balancer(void);
268 static inline int select_nohz_load_balancer(int cpu
)
275 * Only dump TASK_* tasks. (0 for all tasks)
277 extern void show_state_filter(unsigned long state_filter
);
279 static inline void show_state(void)
281 show_state_filter(0);
284 extern void show_regs(struct pt_regs
*);
287 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
288 * task), SP is the stack pointer of the first frame that should be shown in the back
289 * trace (or NULL if the entire call-chain of the task should be shown).
291 extern void show_stack(struct task_struct
*task
, unsigned long *sp
);
293 void io_schedule(void);
294 long io_schedule_timeout(long timeout
);
296 extern void cpu_init (void);
297 extern void trap_init(void);
298 extern void update_process_times(int user
);
299 extern void scheduler_tick(void);
301 extern void sched_show_task(struct task_struct
*p
);
303 #ifdef CONFIG_DETECT_SOFTLOCKUP
304 extern void softlockup_tick(void);
305 extern void touch_softlockup_watchdog(void);
306 extern void touch_all_softlockup_watchdogs(void);
307 extern int proc_dosoftlockup_thresh(struct ctl_table
*table
, int write
,
308 struct file
*filp
, void __user
*buffer
,
309 size_t *lenp
, loff_t
*ppos
);
310 extern unsigned int softlockup_panic
;
311 extern int softlockup_thresh
;
313 static inline void softlockup_tick(void)
316 static inline void touch_softlockup_watchdog(void)
319 static inline void touch_all_softlockup_watchdogs(void)
324 #ifdef CONFIG_DETECT_HUNG_TASK
325 extern unsigned int sysctl_hung_task_panic
;
326 extern unsigned long sysctl_hung_task_check_count
;
327 extern unsigned long sysctl_hung_task_timeout_secs
;
328 extern unsigned long sysctl_hung_task_warnings
;
329 extern int proc_dohung_task_timeout_secs(struct ctl_table
*table
, int write
,
330 struct file
*filp
, void __user
*buffer
,
331 size_t *lenp
, loff_t
*ppos
);
334 /* Attach to any functions which should be ignored in wchan output. */
335 #define __sched __attribute__((__section__(".sched.text")))
337 /* Linker adds these: start and end of __sched functions */
338 extern char __sched_text_start
[], __sched_text_end
[];
340 /* Is this address in the __sched functions? */
341 extern int in_sched_functions(unsigned long addr
);
343 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
344 extern signed long schedule_timeout(signed long timeout
);
345 extern signed long schedule_timeout_interruptible(signed long timeout
);
346 extern signed long schedule_timeout_killable(signed long timeout
);
347 extern signed long schedule_timeout_uninterruptible(signed long timeout
);
348 asmlinkage
void __schedule(void);
349 asmlinkage
void schedule(void);
350 extern int mutex_spin_on_owner(struct mutex
*lock
, struct thread_info
*owner
);
353 struct user_namespace
;
356 * Default maximum number of active map areas, this limits the number of vmas
357 * per mm struct. Users can overwrite this number by sysctl but there is a
360 * When a program's coredump is generated as ELF format, a section is created
361 * per a vma. In ELF, the number of sections is represented in unsigned short.
362 * This means the number of sections should be smaller than 65535 at coredump.
363 * Because the kernel adds some informative sections to a image of program at
364 * generating coredump, we need some margin. The number of extra sections is
365 * 1-3 now and depends on arch. We use "5" as safe margin, here.
367 #define MAPCOUNT_ELF_CORE_MARGIN (5)
368 #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
370 extern int sysctl_max_map_count
;
372 #include <linux/aio.h>
375 arch_get_unmapped_area(struct file
*, unsigned long, unsigned long,
376 unsigned long, unsigned long);
378 arch_get_unmapped_area_topdown(struct file
*filp
, unsigned long addr
,
379 unsigned long len
, unsigned long pgoff
,
380 unsigned long flags
);
381 extern void arch_unmap_area(struct mm_struct
*, unsigned long);
382 extern void arch_unmap_area_topdown(struct mm_struct
*, unsigned long);
384 #if USE_SPLIT_PTLOCKS
386 * The mm counters are not protected by its page_table_lock,
387 * so must be incremented atomically.
389 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
390 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
391 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
392 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
393 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
395 #else /* !USE_SPLIT_PTLOCKS */
397 * The mm counters are protected by its page_table_lock,
398 * so can be incremented directly.
400 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
401 #define get_mm_counter(mm, member) ((mm)->_##member)
402 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
403 #define inc_mm_counter(mm, member) (mm)->_##member++
404 #define dec_mm_counter(mm, member) (mm)->_##member--
406 #endif /* !USE_SPLIT_PTLOCKS */
408 #define get_mm_rss(mm) \
409 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
410 #define update_hiwater_rss(mm) do { \
411 unsigned long _rss = get_mm_rss(mm); \
412 if ((mm)->hiwater_rss < _rss) \
413 (mm)->hiwater_rss = _rss; \
415 #define update_hiwater_vm(mm) do { \
416 if ((mm)->hiwater_vm < (mm)->total_vm) \
417 (mm)->hiwater_vm = (mm)->total_vm; \
420 static inline unsigned long get_mm_hiwater_rss(struct mm_struct
*mm
)
422 return max(mm
->hiwater_rss
, get_mm_rss(mm
));
425 static inline unsigned long get_mm_hiwater_vm(struct mm_struct
*mm
)
427 return max(mm
->hiwater_vm
, mm
->total_vm
);
430 extern void set_dumpable(struct mm_struct
*mm
, int value
);
431 extern int get_dumpable(struct mm_struct
*mm
);
435 #define MMF_DUMPABLE 0 /* core dump is permitted */
436 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
438 #define MMF_DUMPABLE_BITS 2
439 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
441 /* coredump filter bits */
442 #define MMF_DUMP_ANON_PRIVATE 2
443 #define MMF_DUMP_ANON_SHARED 3
444 #define MMF_DUMP_MAPPED_PRIVATE 4
445 #define MMF_DUMP_MAPPED_SHARED 5
446 #define MMF_DUMP_ELF_HEADERS 6
447 #define MMF_DUMP_HUGETLB_PRIVATE 7
448 #define MMF_DUMP_HUGETLB_SHARED 8
450 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
451 #define MMF_DUMP_FILTER_BITS 7
452 #define MMF_DUMP_FILTER_MASK \
453 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
454 #define MMF_DUMP_FILTER_DEFAULT \
455 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
456 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
458 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
459 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
461 # define MMF_DUMP_MASK_DEFAULT_ELF 0
463 /* leave room for more dump flags */
464 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
466 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
468 struct sighand_struct
{
470 struct k_sigaction action
[_NSIG
];
472 wait_queue_head_t signalfd_wqh
;
475 struct pacct_struct
{
478 unsigned long ac_mem
;
479 cputime_t ac_utime
, ac_stime
;
480 unsigned long ac_minflt
, ac_majflt
;
484 * struct task_cputime - collected CPU time counts
485 * @utime: time spent in user mode, in &cputime_t units
486 * @stime: time spent in kernel mode, in &cputime_t units
487 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
489 * This structure groups together three kinds of CPU time that are
490 * tracked for threads and thread groups. Most things considering
491 * CPU time want to group these counts together and treat all three
492 * of them in parallel.
494 struct task_cputime
{
497 unsigned long long sum_exec_runtime
;
499 /* Alternate field names when used to cache expirations. */
500 #define prof_exp stime
501 #define virt_exp utime
502 #define sched_exp sum_exec_runtime
504 #define INIT_CPUTIME \
505 (struct task_cputime) { \
506 .utime = cputime_zero, \
507 .stime = cputime_zero, \
508 .sum_exec_runtime = 0, \
512 * Disable preemption until the scheduler is running.
513 * Reset by start_kernel()->sched_init()->init_idle().
515 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
516 * before the scheduler is active -- see should_resched().
518 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
521 * struct thread_group_cputimer - thread group interval timer counts
522 * @cputime: thread group interval timers.
523 * @running: non-zero when there are timers running and
524 * @cputime receives updates.
525 * @lock: lock for fields in this struct.
527 * This structure contains the version of task_cputime, above, that is
528 * used for thread group CPU timer calculations.
530 struct thread_group_cputimer
{
531 struct task_cputime cputime
;
537 * NOTE! "signal_struct" does not have it's own
538 * locking, because a shared signal_struct always
539 * implies a shared sighand_struct, so locking
540 * sighand_struct is always a proper superset of
541 * the locking of signal_struct.
543 struct signal_struct
{
547 wait_queue_head_t wait_chldexit
; /* for wait4() */
549 /* current thread group signal load-balancing target: */
550 struct task_struct
*curr_target
;
552 /* shared signal handling: */
553 struct sigpending shared_pending
;
555 /* thread group exit support */
558 * - notify group_exit_task when ->count is equal to notify_count
559 * - everyone except group_exit_task is stopped during signal delivery
560 * of fatal signals, group_exit_task processes the signal.
563 struct task_struct
*group_exit_task
;
565 /* thread group stop support, overloads group_exit_code too */
566 int group_stop_count
;
567 unsigned int flags
; /* see SIGNAL_* flags below */
569 /* POSIX.1b Interval Timers */
570 struct list_head posix_timers
;
572 /* ITIMER_REAL timer for the process */
573 struct hrtimer real_timer
;
574 struct pid
*leader_pid
;
575 ktime_t it_real_incr
;
577 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
578 cputime_t it_prof_expires
, it_virt_expires
;
579 cputime_t it_prof_incr
, it_virt_incr
;
582 * Thread group totals for process CPU timers.
583 * See thread_group_cputimer(), et al, for details.
585 struct thread_group_cputimer cputimer
;
587 /* Earliest-expiration cache. */
588 struct task_cputime cputime_expires
;
590 struct list_head cpu_timers
[3];
592 struct pid
*tty_old_pgrp
;
594 /* boolean value for session group leader */
597 struct tty_struct
*tty
; /* NULL if no tty */
600 * Cumulative resource counters for dead threads in the group,
601 * and for reaped dead child processes forked by this group.
602 * Live threads maintain their own counters and add to these
603 * in __exit_signal, except for the group leader.
605 cputime_t utime
, stime
, cutime
, cstime
;
608 unsigned long nvcsw
, nivcsw
, cnvcsw
, cnivcsw
;
609 unsigned long min_flt
, maj_flt
, cmin_flt
, cmaj_flt
;
610 unsigned long inblock
, oublock
, cinblock
, coublock
;
611 struct task_io_accounting ioac
;
614 * Cumulative ns of schedule CPU time fo dead threads in the
615 * group, not including a zombie group leader, (This only differs
616 * from jiffies_to_ns(utime + stime) if sched_clock uses something
617 * other than jiffies.)
619 unsigned long long sum_sched_runtime
;
622 * We don't bother to synchronize most readers of this at all,
623 * because there is no reader checking a limit that actually needs
624 * to get both rlim_cur and rlim_max atomically, and either one
625 * alone is a single word that can safely be read normally.
626 * getrlimit/setrlimit use task_lock(current->group_leader) to
627 * protect this instead of the siglock, because they really
628 * have no need to disable irqs.
630 struct rlimit rlim
[RLIM_NLIMITS
];
632 #ifdef CONFIG_BSD_PROCESS_ACCT
633 struct pacct_struct pacct
; /* per-process accounting information */
635 #ifdef CONFIG_TASKSTATS
636 struct taskstats
*stats
;
640 struct tty_audit_buf
*tty_audit_buf
;
643 int oom_adj
; /* OOM kill score adjustment (bit shift) */
646 /* Context switch must be unlocked if interrupts are to be enabled */
647 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
648 # define __ARCH_WANT_UNLOCKED_CTXSW
652 * Bits in flags field of signal_struct.
654 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
655 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
656 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
657 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
659 * Pending notifications to parent.
661 #define SIGNAL_CLD_STOPPED 0x00000010
662 #define SIGNAL_CLD_CONTINUED 0x00000020
663 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
665 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
667 /* If true, all threads except ->group_exit_task have pending SIGKILL */
668 static inline int signal_group_exit(const struct signal_struct
*sig
)
670 return (sig
->flags
& SIGNAL_GROUP_EXIT
) ||
671 (sig
->group_exit_task
!= NULL
);
675 * Some day this will be a full-fledged user tracking system..
678 atomic_t __count
; /* reference count */
679 atomic_t processes
; /* How many processes does this user have? */
680 atomic_t files
; /* How many open files does this user have? */
681 atomic_t sigpending
; /* How many pending signals does this user have? */
682 #ifdef CONFIG_INOTIFY_USER
683 atomic_t inotify_watches
; /* How many inotify watches does this user have? */
684 atomic_t inotify_devs
; /* How many inotify devs does this user have opened? */
687 atomic_t epoll_watches
; /* The number of file descriptors currently watched */
689 #ifdef CONFIG_POSIX_MQUEUE
690 /* protected by mq_lock */
691 unsigned long mq_bytes
; /* How many bytes can be allocated to mqueue? */
693 unsigned long locked_shm
; /* How many pages of mlocked shm ? */
696 struct key
*uid_keyring
; /* UID specific keyring */
697 struct key
*session_keyring
; /* UID's default session keyring */
700 /* Hash table maintenance information */
701 struct hlist_node uidhash_node
;
703 struct user_namespace
*user_ns
;
705 #ifdef CONFIG_USER_SCHED
706 struct task_group
*tg
;
709 struct delayed_work work
;
713 #ifdef CONFIG_PERF_EVENTS
714 atomic_long_t locked_vm
;
718 extern int uids_sysfs_init(void);
720 extern struct user_struct
*find_user(uid_t
);
722 extern struct user_struct root_user
;
723 #define INIT_USER (&root_user)
726 struct backing_dev_info
;
727 struct reclaim_state
;
729 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
731 /* cumulative counters */
732 unsigned long pcount
; /* # of times run on this cpu */
733 unsigned long long run_delay
; /* time spent waiting on a runqueue */
736 unsigned long long last_arrival
,/* when we last ran on a cpu */
737 last_queued
; /* when we were last queued to run */
738 #ifdef CONFIG_SCHEDSTATS
740 unsigned int bkl_count
;
743 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
745 #ifdef CONFIG_TASK_DELAY_ACCT
746 struct task_delay_info
{
748 unsigned int flags
; /* Private per-task flags */
750 /* For each stat XXX, add following, aligned appropriately
752 * struct timespec XXX_start, XXX_end;
756 * Atomicity of updates to XXX_delay, XXX_count protected by
757 * single lock above (split into XXX_lock if contention is an issue).
761 * XXX_count is incremented on every XXX operation, the delay
762 * associated with the operation is added to XXX_delay.
763 * XXX_delay contains the accumulated delay time in nanoseconds.
765 struct timespec blkio_start
, blkio_end
; /* Shared by blkio, swapin */
766 u64 blkio_delay
; /* wait for sync block io completion */
767 u64 swapin_delay
; /* wait for swapin block io completion */
768 u32 blkio_count
; /* total count of the number of sync block */
769 /* io operations performed */
770 u32 swapin_count
; /* total count of the number of swapin block */
771 /* io operations performed */
773 struct timespec freepages_start
, freepages_end
;
774 u64 freepages_delay
; /* wait for memory reclaim */
775 u32 freepages_count
; /* total count of memory reclaim */
777 #endif /* CONFIG_TASK_DELAY_ACCT */
779 static inline int sched_info_on(void)
781 #ifdef CONFIG_SCHEDSTATS
783 #elif defined(CONFIG_TASK_DELAY_ACCT)
784 extern int delayacct_on
;
799 * sched-domains (multiprocessor balancing) declarations:
803 * Increase resolution of nice-level calculations:
805 #define SCHED_LOAD_SHIFT 10
806 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
808 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
811 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
812 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
813 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
814 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
815 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
816 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
817 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
818 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
819 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
820 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
821 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
823 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
825 enum powersavings_balance_level
{
826 POWERSAVINGS_BALANCE_NONE
= 0, /* No power saving load balance */
827 POWERSAVINGS_BALANCE_BASIC
, /* Fill one thread/core/package
828 * first for long running threads
830 POWERSAVINGS_BALANCE_WAKEUP
, /* Also bias task wakeups to semi-idle
831 * cpu package for power savings
833 MAX_POWERSAVINGS_BALANCE_LEVELS
836 extern int sched_mc_power_savings
, sched_smt_power_savings
;
838 static inline int sd_balance_for_mc_power(void)
840 if (sched_smt_power_savings
)
841 return SD_POWERSAVINGS_BALANCE
;
843 return SD_PREFER_SIBLING
;
846 static inline int sd_balance_for_package_power(void)
848 if (sched_mc_power_savings
| sched_smt_power_savings
)
849 return SD_POWERSAVINGS_BALANCE
;
851 return SD_PREFER_SIBLING
;
855 * Optimise SD flags for power savings:
856 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
857 * Keep default SD flags if sched_{smt,mc}_power_saving=0
860 static inline int sd_power_saving_flags(void)
862 if (sched_mc_power_savings
| sched_smt_power_savings
)
863 return SD_BALANCE_NEWIDLE
;
869 struct sched_group
*next
; /* Must be a circular list */
872 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
875 unsigned int cpu_power
;
878 * The CPUs this group covers.
880 * NOTE: this field is variable length. (Allocated dynamically
881 * by attaching extra space to the end of the structure,
882 * depending on how many CPUs the kernel has booted up with)
884 * It is also be embedded into static data structures at build
885 * time. (See 'struct static_sched_group' in kernel/sched.c)
887 unsigned long cpumask
[0];
890 static inline struct cpumask
*sched_group_cpus(struct sched_group
*sg
)
892 return to_cpumask(sg
->cpumask
);
895 enum sched_domain_level
{
905 struct sched_domain_attr
{
906 int relax_domain_level
;
909 #define SD_ATTR_INIT (struct sched_domain_attr) { \
910 .relax_domain_level = -1, \
913 struct sched_domain
{
914 /* These fields must be setup */
915 struct sched_domain
*parent
; /* top domain must be null terminated */
916 struct sched_domain
*child
; /* bottom domain must be null terminated */
917 struct sched_group
*groups
; /* the balancing groups of the domain */
918 unsigned long min_interval
; /* Minimum balance interval ms */
919 unsigned long max_interval
; /* Maximum balance interval ms */
920 unsigned int busy_factor
; /* less balancing by factor if busy */
921 unsigned int imbalance_pct
; /* No balance until over watermark */
922 unsigned int cache_nice_tries
; /* Leave cache hot tasks for # tries */
923 unsigned int busy_idx
;
924 unsigned int idle_idx
;
925 unsigned int newidle_idx
;
926 unsigned int wake_idx
;
927 unsigned int forkexec_idx
;
928 unsigned int smt_gain
;
929 int flags
; /* See SD_* */
930 enum sched_domain_level level
;
932 /* Runtime fields. */
933 unsigned long last_balance
; /* init to jiffies. units in jiffies */
934 unsigned int balance_interval
; /* initialise to 1. units in ms. */
935 unsigned int nr_balance_failed
; /* initialise to 0 */
939 #ifdef CONFIG_SCHEDSTATS
940 /* load_balance() stats */
941 unsigned int lb_count
[CPU_MAX_IDLE_TYPES
];
942 unsigned int lb_failed
[CPU_MAX_IDLE_TYPES
];
943 unsigned int lb_balanced
[CPU_MAX_IDLE_TYPES
];
944 unsigned int lb_imbalance
[CPU_MAX_IDLE_TYPES
];
945 unsigned int lb_gained
[CPU_MAX_IDLE_TYPES
];
946 unsigned int lb_hot_gained
[CPU_MAX_IDLE_TYPES
];
947 unsigned int lb_nobusyg
[CPU_MAX_IDLE_TYPES
];
948 unsigned int lb_nobusyq
[CPU_MAX_IDLE_TYPES
];
950 /* Active load balancing */
951 unsigned int alb_count
;
952 unsigned int alb_failed
;
953 unsigned int alb_pushed
;
955 /* SD_BALANCE_EXEC stats */
956 unsigned int sbe_count
;
957 unsigned int sbe_balanced
;
958 unsigned int sbe_pushed
;
960 /* SD_BALANCE_FORK stats */
961 unsigned int sbf_count
;
962 unsigned int sbf_balanced
;
963 unsigned int sbf_pushed
;
965 /* try_to_wake_up() stats */
966 unsigned int ttwu_wake_remote
;
967 unsigned int ttwu_move_affine
;
968 unsigned int ttwu_move_balance
;
970 #ifdef CONFIG_SCHED_DEBUG
975 * Span of all CPUs in this domain.
977 * NOTE: this field is variable length. (Allocated dynamically
978 * by attaching extra space to the end of the structure,
979 * depending on how many CPUs the kernel has booted up with)
981 * It is also be embedded into static data structures at build
982 * time. (See 'struct static_sched_domain' in kernel/sched.c)
984 unsigned long span
[0];
987 static inline struct cpumask
*sched_domain_span(struct sched_domain
*sd
)
989 return to_cpumask(sd
->span
);
992 extern void partition_sched_domains(int ndoms_new
, struct cpumask
*doms_new
,
993 struct sched_domain_attr
*dattr_new
);
995 /* Test a flag in parent sched domain */
996 static inline int test_sd_parent(struct sched_domain
*sd
, int flag
)
998 if (sd
->parent
&& (sd
->parent
->flags
& flag
))
1004 unsigned long default_scale_freq_power(struct sched_domain
*sd
, int cpu
);
1005 unsigned long default_scale_smt_power(struct sched_domain
*sd
, int cpu
);
1007 #else /* CONFIG_SMP */
1009 struct sched_domain_attr
;
1012 partition_sched_domains(int ndoms_new
, struct cpumask
*doms_new
,
1013 struct sched_domain_attr
*dattr_new
)
1016 #endif /* !CONFIG_SMP */
1019 struct io_context
; /* See blkdev.h */
1022 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1023 extern void prefetch_stack(struct task_struct
*t
);
1025 static inline void prefetch_stack(struct task_struct
*t
) { }
1028 struct audit_context
; /* See audit.c */
1030 struct pipe_inode_info
;
1031 struct uts_namespace
;
1034 struct sched_domain
;
1039 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1040 #define WF_FORK 0x02 /* child wakeup after fork */
1042 struct sched_class
{
1043 const struct sched_class
*next
;
1045 void (*enqueue_task
) (struct rq
*rq
, struct task_struct
*p
, int wakeup
);
1046 void (*dequeue_task
) (struct rq
*rq
, struct task_struct
*p
, int sleep
);
1047 void (*yield_task
) (struct rq
*rq
);
1049 void (*check_preempt_curr
) (struct rq
*rq
, struct task_struct
*p
, int flags
);
1051 struct task_struct
* (*pick_next_task
) (struct rq
*rq
);
1052 void (*put_prev_task
) (struct rq
*rq
, struct task_struct
*p
);
1055 int (*select_task_rq
)(struct task_struct
*p
, int sd_flag
, int flags
);
1057 unsigned long (*load_balance
) (struct rq
*this_rq
, int this_cpu
,
1058 struct rq
*busiest
, unsigned long max_load_move
,
1059 struct sched_domain
*sd
, enum cpu_idle_type idle
,
1060 int *all_pinned
, int *this_best_prio
);
1062 int (*move_one_task
) (struct rq
*this_rq
, int this_cpu
,
1063 struct rq
*busiest
, struct sched_domain
*sd
,
1064 enum cpu_idle_type idle
);
1065 void (*pre_schedule
) (struct rq
*this_rq
, struct task_struct
*task
);
1066 void (*post_schedule
) (struct rq
*this_rq
);
1067 void (*task_wake_up
) (struct rq
*this_rq
, struct task_struct
*task
);
1069 void (*set_cpus_allowed
)(struct task_struct
*p
,
1070 const struct cpumask
*newmask
);
1072 void (*rq_online
)(struct rq
*rq
);
1073 void (*rq_offline
)(struct rq
*rq
);
1076 void (*set_curr_task
) (struct rq
*rq
);
1077 void (*task_tick
) (struct rq
*rq
, struct task_struct
*p
, int queued
);
1078 void (*task_new
) (struct rq
*rq
, struct task_struct
*p
);
1080 void (*switched_from
) (struct rq
*this_rq
, struct task_struct
*task
,
1082 void (*switched_to
) (struct rq
*this_rq
, struct task_struct
*task
,
1084 void (*prio_changed
) (struct rq
*this_rq
, struct task_struct
*task
,
1085 int oldprio
, int running
);
1087 unsigned int (*get_rr_interval
) (struct task_struct
*task
);
1089 #ifdef CONFIG_FAIR_GROUP_SCHED
1090 void (*moved_group
) (struct task_struct
*p
);
1094 struct load_weight
{
1095 unsigned long weight
, inv_weight
;
1099 * CFS stats for a schedulable entity (task, task-group etc)
1101 * Current field usage histogram:
1108 struct sched_entity
{
1109 struct load_weight load
; /* for load-balancing */
1110 struct rb_node run_node
;
1111 struct list_head group_node
;
1115 u64 sum_exec_runtime
;
1117 u64 prev_sum_exec_runtime
;
1129 #ifdef CONFIG_SCHEDSTATS
1139 s64 sum_sleep_runtime
;
1146 u64 nr_migrations_cold
;
1147 u64 nr_failed_migrations_affine
;
1148 u64 nr_failed_migrations_running
;
1149 u64 nr_failed_migrations_hot
;
1150 u64 nr_forced_migrations
;
1151 u64 nr_forced2_migrations
;
1154 u64 nr_wakeups_sync
;
1155 u64 nr_wakeups_migrate
;
1156 u64 nr_wakeups_local
;
1157 u64 nr_wakeups_remote
;
1158 u64 nr_wakeups_affine
;
1159 u64 nr_wakeups_affine_attempts
;
1160 u64 nr_wakeups_passive
;
1161 u64 nr_wakeups_idle
;
1164 #ifdef CONFIG_FAIR_GROUP_SCHED
1165 struct sched_entity
*parent
;
1166 /* rq on which this entity is (to be) queued: */
1167 struct cfs_rq
*cfs_rq
;
1168 /* rq "owned" by this entity/group: */
1169 struct cfs_rq
*my_q
;
1173 struct sched_rt_entity
{
1174 struct list_head run_list
;
1175 unsigned long timeout
;
1176 unsigned int time_slice
;
1177 int nr_cpus_allowed
;
1179 struct sched_rt_entity
*back
;
1180 #ifdef CONFIG_RT_GROUP_SCHED
1181 struct sched_rt_entity
*parent
;
1182 /* rq on which this entity is (to be) queued: */
1183 struct rt_rq
*rt_rq
;
1184 /* rq "owned" by this entity/group: */
1191 struct task_struct
{
1192 volatile long state
; /* -1 unrunnable, 0 runnable, >0 stopped */
1195 unsigned int flags
; /* per process flags, defined below */
1196 unsigned int ptrace
;
1198 int lock_depth
; /* BKL lock depth */
1201 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1206 int prio
, static_prio
, normal_prio
;
1207 unsigned int rt_priority
;
1208 const struct sched_class
*sched_class
;
1209 struct sched_entity se
;
1210 struct sched_rt_entity rt
;
1212 #ifdef CONFIG_PREEMPT_NOTIFIERS
1213 /* list of struct preempt_notifier: */
1214 struct hlist_head preempt_notifiers
;
1218 * fpu_counter contains the number of consecutive context switches
1219 * that the FPU is used. If this is over a threshold, the lazy fpu
1220 * saving becomes unlazy to save the trap. This is an unsigned char
1221 * so that after 256 times the counter wraps and the behavior turns
1222 * lazy again; this to deal with bursty apps that only use FPU for
1225 unsigned char fpu_counter
;
1226 #ifdef CONFIG_BLK_DEV_IO_TRACE
1227 unsigned int btrace_seq
;
1230 unsigned int policy
;
1231 cpumask_t cpus_allowed
;
1233 #ifdef CONFIG_TREE_PREEMPT_RCU
1234 int rcu_read_lock_nesting
;
1235 char rcu_read_unlock_special
;
1236 struct rcu_node
*rcu_blocked_node
;
1237 struct list_head rcu_node_entry
;
1238 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1240 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1241 struct sched_info sched_info
;
1244 struct list_head tasks
;
1245 struct plist_node pushable_tasks
;
1247 struct mm_struct
*mm
, *active_mm
;
1250 struct linux_binfmt
*binfmt
;
1252 int exit_code
, exit_signal
;
1253 int pdeath_signal
; /* The signal sent when the parent dies */
1255 unsigned int personality
;
1256 unsigned did_exec
:1;
1257 unsigned in_execve
:1; /* Tell the LSMs that the process is doing an
1259 unsigned in_iowait
:1;
1262 /* Revert to default priority/policy when forking */
1263 unsigned sched_reset_on_fork
:1;
1268 #ifdef CONFIG_CC_STACKPROTECTOR
1269 /* Canary value for the -fstack-protector gcc feature */
1270 unsigned long stack_canary
;
1274 * pointers to (original) parent process, youngest child, younger sibling,
1275 * older sibling, respectively. (p->father can be replaced with
1276 * p->real_parent->pid)
1278 struct task_struct
*real_parent
; /* real parent process */
1279 struct task_struct
*parent
; /* recipient of SIGCHLD, wait4() reports */
1281 * children/sibling forms the list of my natural children
1283 struct list_head children
; /* list of my children */
1284 struct list_head sibling
; /* linkage in my parent's children list */
1285 struct task_struct
*group_leader
; /* threadgroup leader */
1288 * ptraced is the list of tasks this task is using ptrace on.
1289 * This includes both natural children and PTRACE_ATTACH targets.
1290 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1292 struct list_head ptraced
;
1293 struct list_head ptrace_entry
;
1296 * This is the tracer handle for the ptrace BTS extension.
1297 * This field actually belongs to the ptracer task.
1299 struct bts_context
*bts
;
1301 /* PID/PID hash table linkage. */
1302 struct pid_link pids
[PIDTYPE_MAX
];
1303 struct list_head thread_group
;
1305 struct completion
*vfork_done
; /* for vfork() */
1306 int __user
*set_child_tid
; /* CLONE_CHILD_SETTID */
1307 int __user
*clear_child_tid
; /* CLONE_CHILD_CLEARTID */
1309 cputime_t utime
, stime
, utimescaled
, stimescaled
;
1311 cputime_t prev_utime
, prev_stime
;
1312 unsigned long nvcsw
, nivcsw
; /* context switch counts */
1313 struct timespec start_time
; /* monotonic time */
1314 struct timespec real_start_time
; /* boot based time */
1315 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1316 unsigned long min_flt
, maj_flt
;
1318 struct task_cputime cputime_expires
;
1319 struct list_head cpu_timers
[3];
1321 /* process credentials */
1322 const struct cred
*real_cred
; /* objective and real subjective task
1323 * credentials (COW) */
1324 const struct cred
*cred
; /* effective (overridable) subjective task
1325 * credentials (COW) */
1326 struct mutex cred_guard_mutex
; /* guard against foreign influences on
1327 * credential calculations
1328 * (notably. ptrace) */
1329 struct cred
*replacement_session_keyring
; /* for KEYCTL_SESSION_TO_PARENT */
1331 char comm
[TASK_COMM_LEN
]; /* executable name excluding path
1332 - access with [gs]et_task_comm (which lock
1333 it with task_lock())
1334 - initialized normally by flush_old_exec */
1335 /* file system info */
1336 int link_count
, total_link_count
;
1337 #ifdef CONFIG_SYSVIPC
1339 struct sysv_sem sysvsem
;
1341 #ifdef CONFIG_DETECT_HUNG_TASK
1342 /* hung task detection */
1343 unsigned long last_switch_count
;
1345 /* CPU-specific state of this task */
1346 struct thread_struct thread
;
1347 /* filesystem information */
1348 struct fs_struct
*fs
;
1349 /* open file information */
1350 struct files_struct
*files
;
1352 struct nsproxy
*nsproxy
;
1353 /* signal handlers */
1354 struct signal_struct
*signal
;
1355 struct sighand_struct
*sighand
;
1357 sigset_t blocked
, real_blocked
;
1358 sigset_t saved_sigmask
; /* restored if set_restore_sigmask() was used */
1359 struct sigpending pending
;
1361 unsigned long sas_ss_sp
;
1363 int (*notifier
)(void *priv
);
1364 void *notifier_data
;
1365 sigset_t
*notifier_mask
;
1366 struct audit_context
*audit_context
;
1367 #ifdef CONFIG_AUDITSYSCALL
1369 unsigned int sessionid
;
1373 /* Thread group tracking */
1376 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1378 spinlock_t alloc_lock
;
1380 #ifdef CONFIG_GENERIC_HARDIRQS
1381 /* IRQ handler threads */
1382 struct irqaction
*irqaction
;
1385 /* Protection of the PI data structures: */
1388 #ifdef CONFIG_RT_MUTEXES
1389 /* PI waiters blocked on a rt_mutex held by this task */
1390 struct plist_head pi_waiters
;
1391 /* Deadlock detection and priority inheritance handling */
1392 struct rt_mutex_waiter
*pi_blocked_on
;
1395 #ifdef CONFIG_DEBUG_MUTEXES
1396 /* mutex deadlock detection */
1397 struct mutex_waiter
*blocked_on
;
1399 #ifdef CONFIG_TRACE_IRQFLAGS
1400 unsigned int irq_events
;
1401 int hardirqs_enabled
;
1402 unsigned long hardirq_enable_ip
;
1403 unsigned int hardirq_enable_event
;
1404 unsigned long hardirq_disable_ip
;
1405 unsigned int hardirq_disable_event
;
1406 int softirqs_enabled
;
1407 unsigned long softirq_disable_ip
;
1408 unsigned int softirq_disable_event
;
1409 unsigned long softirq_enable_ip
;
1410 unsigned int softirq_enable_event
;
1411 int hardirq_context
;
1412 int softirq_context
;
1414 #ifdef CONFIG_LOCKDEP
1415 # define MAX_LOCK_DEPTH 48UL
1418 unsigned int lockdep_recursion
;
1419 struct held_lock held_locks
[MAX_LOCK_DEPTH
];
1420 gfp_t lockdep_reclaim_gfp
;
1423 /* journalling filesystem info */
1426 /* stacked block device info */
1427 struct bio
*bio_list
, **bio_tail
;
1430 struct reclaim_state
*reclaim_state
;
1432 struct backing_dev_info
*backing_dev_info
;
1434 struct io_context
*io_context
;
1436 unsigned long ptrace_message
;
1437 siginfo_t
*last_siginfo
; /* For ptrace use. */
1438 struct task_io_accounting ioac
;
1439 #if defined(CONFIG_TASK_XACCT)
1440 u64 acct_rss_mem1
; /* accumulated rss usage */
1441 u64 acct_vm_mem1
; /* accumulated virtual memory usage */
1442 cputime_t acct_timexpd
; /* stime + utime since last update */
1444 #ifdef CONFIG_CPUSETS
1445 nodemask_t mems_allowed
; /* Protected by alloc_lock */
1446 int cpuset_mem_spread_rotor
;
1448 #ifdef CONFIG_CGROUPS
1449 /* Control Group info protected by css_set_lock */
1450 struct css_set
*cgroups
;
1451 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1452 struct list_head cg_list
;
1455 struct robust_list_head __user
*robust_list
;
1456 #ifdef CONFIG_COMPAT
1457 struct compat_robust_list_head __user
*compat_robust_list
;
1459 struct list_head pi_state_list
;
1460 struct futex_pi_state
*pi_state_cache
;
1462 #ifdef CONFIG_PERF_EVENTS
1463 struct perf_event_context
*perf_event_ctxp
;
1464 struct mutex perf_event_mutex
;
1465 struct list_head perf_event_list
;
1468 struct mempolicy
*mempolicy
; /* Protected by alloc_lock */
1471 atomic_t fs_excl
; /* holding fs exclusive resources */
1472 struct rcu_head rcu
;
1475 * cache last used pipe for splice
1477 struct pipe_inode_info
*splice_pipe
;
1478 #ifdef CONFIG_TASK_DELAY_ACCT
1479 struct task_delay_info
*delays
;
1481 #ifdef CONFIG_FAULT_INJECTION
1484 struct prop_local_single dirties
;
1485 #ifdef CONFIG_LATENCYTOP
1486 int latency_record_count
;
1487 struct latency_record latency_record
[LT_SAVECOUNT
];
1490 * time slack values; these are used to round up poll() and
1491 * select() etc timeout values. These are in nanoseconds.
1493 unsigned long timer_slack_ns
;
1494 unsigned long default_timer_slack_ns
;
1496 struct list_head
*scm_work_list
;
1497 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1498 /* Index of current stored adress in ret_stack */
1500 /* Stack of return addresses for return function tracing */
1501 struct ftrace_ret_stack
*ret_stack
;
1502 /* time stamp for last schedule */
1503 unsigned long long ftrace_timestamp
;
1505 * Number of functions that haven't been traced
1506 * because of depth overrun.
1508 atomic_t trace_overrun
;
1509 /* Pause for the tracing */
1510 atomic_t tracing_graph_pause
;
1512 #ifdef CONFIG_TRACING
1513 /* state flags for use by tracers */
1514 unsigned long trace
;
1515 /* bitmask of trace recursion */
1516 unsigned long trace_recursion
;
1517 #endif /* CONFIG_TRACING */
1520 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1521 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1524 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1525 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1526 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1527 * values are inverted: lower p->prio value means higher priority.
1529 * The MAX_USER_RT_PRIO value allows the actual maximum
1530 * RT priority to be separate from the value exported to
1531 * user-space. This allows kernel threads to set their
1532 * priority to a value higher than any user task. Note:
1533 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1536 #define MAX_USER_RT_PRIO 100
1537 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1539 #define MAX_PRIO (MAX_RT_PRIO + 40)
1540 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1542 static inline int rt_prio(int prio
)
1544 if (unlikely(prio
< MAX_RT_PRIO
))
1549 static inline int rt_task(struct task_struct
*p
)
1551 return rt_prio(p
->prio
);
1554 static inline struct pid
*task_pid(struct task_struct
*task
)
1556 return task
->pids
[PIDTYPE_PID
].pid
;
1559 static inline struct pid
*task_tgid(struct task_struct
*task
)
1561 return task
->group_leader
->pids
[PIDTYPE_PID
].pid
;
1565 * Without tasklist or rcu lock it is not safe to dereference
1566 * the result of task_pgrp/task_session even if task == current,
1567 * we can race with another thread doing sys_setsid/sys_setpgid.
1569 static inline struct pid
*task_pgrp(struct task_struct
*task
)
1571 return task
->group_leader
->pids
[PIDTYPE_PGID
].pid
;
1574 static inline struct pid
*task_session(struct task_struct
*task
)
1576 return task
->group_leader
->pids
[PIDTYPE_SID
].pid
;
1579 struct pid_namespace
;
1582 * the helpers to get the task's different pids as they are seen
1583 * from various namespaces
1585 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1586 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1588 * task_xid_nr_ns() : id seen from the ns specified;
1590 * set_task_vxid() : assigns a virtual id to a task;
1592 * see also pid_nr() etc in include/linux/pid.h
1594 pid_t
__task_pid_nr_ns(struct task_struct
*task
, enum pid_type type
,
1595 struct pid_namespace
*ns
);
1597 static inline pid_t
task_pid_nr(struct task_struct
*tsk
)
1602 static inline pid_t
task_pid_nr_ns(struct task_struct
*tsk
,
1603 struct pid_namespace
*ns
)
1605 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, ns
);
1608 static inline pid_t
task_pid_vnr(struct task_struct
*tsk
)
1610 return __task_pid_nr_ns(tsk
, PIDTYPE_PID
, NULL
);
1614 static inline pid_t
task_tgid_nr(struct task_struct
*tsk
)
1619 pid_t
task_tgid_nr_ns(struct task_struct
*tsk
, struct pid_namespace
*ns
);
1621 static inline pid_t
task_tgid_vnr(struct task_struct
*tsk
)
1623 return pid_vnr(task_tgid(tsk
));
1627 static inline pid_t
task_pgrp_nr_ns(struct task_struct
*tsk
,
1628 struct pid_namespace
*ns
)
1630 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, ns
);
1633 static inline pid_t
task_pgrp_vnr(struct task_struct
*tsk
)
1635 return __task_pid_nr_ns(tsk
, PIDTYPE_PGID
, NULL
);
1639 static inline pid_t
task_session_nr_ns(struct task_struct
*tsk
,
1640 struct pid_namespace
*ns
)
1642 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, ns
);
1645 static inline pid_t
task_session_vnr(struct task_struct
*tsk
)
1647 return __task_pid_nr_ns(tsk
, PIDTYPE_SID
, NULL
);
1650 /* obsolete, do not use */
1651 static inline pid_t
task_pgrp_nr(struct task_struct
*tsk
)
1653 return task_pgrp_nr_ns(tsk
, &init_pid_ns
);
1657 * pid_alive - check that a task structure is not stale
1658 * @p: Task structure to be checked.
1660 * Test if a process is not yet dead (at most zombie state)
1661 * If pid_alive fails, then pointers within the task structure
1662 * can be stale and must not be dereferenced.
1664 static inline int pid_alive(struct task_struct
*p
)
1666 return p
->pids
[PIDTYPE_PID
].pid
!= NULL
;
1670 * is_global_init - check if a task structure is init
1671 * @tsk: Task structure to be checked.
1673 * Check if a task structure is the first user space task the kernel created.
1675 static inline int is_global_init(struct task_struct
*tsk
)
1677 return tsk
->pid
== 1;
1681 * is_container_init:
1682 * check whether in the task is init in its own pid namespace.
1684 extern int is_container_init(struct task_struct
*tsk
);
1686 extern struct pid
*cad_pid
;
1688 extern void free_task(struct task_struct
*tsk
);
1689 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1691 extern void __put_task_struct(struct task_struct
*t
);
1693 static inline void put_task_struct(struct task_struct
*t
)
1695 if (atomic_dec_and_test(&t
->usage
))
1696 __put_task_struct(t
);
1699 extern cputime_t
task_utime(struct task_struct
*p
);
1700 extern cputime_t
task_stime(struct task_struct
*p
);
1701 extern cputime_t
task_gtime(struct task_struct
*p
);
1706 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1707 /* Not implemented yet, only for 486*/
1708 #define PF_STARTING 0x00000002 /* being created */
1709 #define PF_EXITING 0x00000004 /* getting shut down */
1710 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1711 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1712 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1713 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1714 #define PF_DUMPCORE 0x00000200 /* dumped core */
1715 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1716 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1717 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1718 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1719 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1720 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1721 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1722 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1723 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1724 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1725 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1726 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1727 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1728 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1729 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1730 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1731 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1732 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1733 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1734 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1735 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1738 * Only the _current_ task can read/write to tsk->flags, but other
1739 * tasks can access tsk->flags in readonly mode for example
1740 * with tsk_used_math (like during threaded core dumping).
1741 * There is however an exception to this rule during ptrace
1742 * or during fork: the ptracer task is allowed to write to the
1743 * child->flags of its traced child (same goes for fork, the parent
1744 * can write to the child->flags), because we're guaranteed the
1745 * child is not running and in turn not changing child->flags
1746 * at the same time the parent does it.
1748 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1749 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1750 #define clear_used_math() clear_stopped_child_used_math(current)
1751 #define set_used_math() set_stopped_child_used_math(current)
1752 #define conditional_stopped_child_used_math(condition, child) \
1753 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1754 #define conditional_used_math(condition) \
1755 conditional_stopped_child_used_math(condition, current)
1756 #define copy_to_stopped_child_used_math(child) \
1757 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1758 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1759 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1760 #define used_math() tsk_used_math(current)
1762 #ifdef CONFIG_TREE_PREEMPT_RCU
1764 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1765 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1767 static inline void rcu_copy_process(struct task_struct
*p
)
1769 p
->rcu_read_lock_nesting
= 0;
1770 p
->rcu_read_unlock_special
= 0;
1771 p
->rcu_blocked_node
= NULL
;
1772 INIT_LIST_HEAD(&p
->rcu_node_entry
);
1777 static inline void rcu_copy_process(struct task_struct
*p
)
1784 extern int set_cpus_allowed_ptr(struct task_struct
*p
,
1785 const struct cpumask
*new_mask
);
1787 static inline int set_cpus_allowed_ptr(struct task_struct
*p
,
1788 const struct cpumask
*new_mask
)
1790 if (!cpumask_test_cpu(0, new_mask
))
1795 static inline int set_cpus_allowed(struct task_struct
*p
, cpumask_t new_mask
)
1797 return set_cpus_allowed_ptr(p
, &new_mask
);
1801 * Architectures can set this to 1 if they have specified
1802 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1803 * but then during bootup it turns out that sched_clock()
1804 * is reliable after all:
1806 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1807 extern int sched_clock_stable
;
1810 extern unsigned long long sched_clock(void);
1812 extern void sched_clock_init(void);
1813 extern u64
sched_clock_cpu(int cpu
);
1815 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1816 static inline void sched_clock_tick(void)
1820 static inline void sched_clock_idle_sleep_event(void)
1824 static inline void sched_clock_idle_wakeup_event(u64 delta_ns
)
1828 extern void sched_clock_tick(void);
1829 extern void sched_clock_idle_sleep_event(void);
1830 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1834 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1835 * clock constructed from sched_clock():
1837 extern unsigned long long cpu_clock(int cpu
);
1839 extern unsigned long long
1840 task_sched_runtime(struct task_struct
*task
);
1841 extern unsigned long long thread_group_sched_runtime(struct task_struct
*task
);
1843 /* sched_exec is called by processes performing an exec */
1845 extern void sched_exec(void);
1847 #define sched_exec() {}
1850 extern void sched_clock_idle_sleep_event(void);
1851 extern void sched_clock_idle_wakeup_event(u64 delta_ns
);
1853 #ifdef CONFIG_HOTPLUG_CPU
1854 extern void idle_task_exit(void);
1856 static inline void idle_task_exit(void) {}
1859 extern void sched_idle_next(void);
1861 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1862 extern void wake_up_idle_cpu(int cpu
);
1864 static inline void wake_up_idle_cpu(int cpu
) { }
1867 extern unsigned int sysctl_sched_latency
;
1868 extern unsigned int sysctl_sched_min_granularity
;
1869 extern unsigned int sysctl_sched_wakeup_granularity
;
1870 extern unsigned int sysctl_sched_shares_ratelimit
;
1871 extern unsigned int sysctl_sched_shares_thresh
;
1872 extern unsigned int sysctl_sched_child_runs_first
;
1873 #ifdef CONFIG_SCHED_DEBUG
1874 extern unsigned int sysctl_sched_features
;
1875 extern unsigned int sysctl_sched_migration_cost
;
1876 extern unsigned int sysctl_sched_nr_migrate
;
1877 extern unsigned int sysctl_sched_time_avg
;
1878 extern unsigned int sysctl_timer_migration
;
1880 int sched_nr_latency_handler(struct ctl_table
*table
, int write
,
1881 struct file
*file
, void __user
*buffer
, size_t *length
,
1884 #ifdef CONFIG_SCHED_DEBUG
1885 static inline unsigned int get_sysctl_timer_migration(void)
1887 return sysctl_timer_migration
;
1890 static inline unsigned int get_sysctl_timer_migration(void)
1895 extern unsigned int sysctl_sched_rt_period
;
1896 extern int sysctl_sched_rt_runtime
;
1898 int sched_rt_handler(struct ctl_table
*table
, int write
,
1899 struct file
*filp
, void __user
*buffer
, size_t *lenp
,
1902 extern unsigned int sysctl_sched_compat_yield
;
1904 #ifdef CONFIG_RT_MUTEXES
1905 extern int rt_mutex_getprio(struct task_struct
*p
);
1906 extern void rt_mutex_setprio(struct task_struct
*p
, int prio
);
1907 extern void rt_mutex_adjust_pi(struct task_struct
*p
);
1909 static inline int rt_mutex_getprio(struct task_struct
*p
)
1911 return p
->normal_prio
;
1913 # define rt_mutex_adjust_pi(p) do { } while (0)
1916 extern void set_user_nice(struct task_struct
*p
, long nice
);
1917 extern int task_prio(const struct task_struct
*p
);
1918 extern int task_nice(const struct task_struct
*p
);
1919 extern int can_nice(const struct task_struct
*p
, const int nice
);
1920 extern int task_curr(const struct task_struct
*p
);
1921 extern int idle_cpu(int cpu
);
1922 extern int sched_setscheduler(struct task_struct
*, int, struct sched_param
*);
1923 extern int sched_setscheduler_nocheck(struct task_struct
*, int,
1924 struct sched_param
*);
1925 extern struct task_struct
*idle_task(int cpu
);
1926 extern struct task_struct
*curr_task(int cpu
);
1927 extern void set_curr_task(int cpu
, struct task_struct
*p
);
1932 * The default (Linux) execution domain.
1934 extern struct exec_domain default_exec_domain
;
1936 union thread_union
{
1937 struct thread_info thread_info
;
1938 unsigned long stack
[THREAD_SIZE
/sizeof(long)];
1941 #ifndef __HAVE_ARCH_KSTACK_END
1942 static inline int kstack_end(void *addr
)
1944 /* Reliable end of stack detection:
1945 * Some APM bios versions misalign the stack
1947 return !(((unsigned long)addr
+sizeof(void*)-1) & (THREAD_SIZE
-sizeof(void*)));
1951 extern union thread_union init_thread_union
;
1952 extern struct task_struct init_task
;
1954 extern struct mm_struct init_mm
;
1956 extern struct pid_namespace init_pid_ns
;
1959 * find a task by one of its numerical ids
1961 * find_task_by_pid_ns():
1962 * finds a task by its pid in the specified namespace
1963 * find_task_by_vpid():
1964 * finds a task by its virtual pid
1966 * see also find_vpid() etc in include/linux/pid.h
1969 extern struct task_struct
*find_task_by_vpid(pid_t nr
);
1970 extern struct task_struct
*find_task_by_pid_ns(pid_t nr
,
1971 struct pid_namespace
*ns
);
1973 extern void __set_special_pids(struct pid
*pid
);
1975 /* per-UID process charging. */
1976 extern struct user_struct
* alloc_uid(struct user_namespace
*, uid_t
);
1977 static inline struct user_struct
*get_uid(struct user_struct
*u
)
1979 atomic_inc(&u
->__count
);
1982 extern void free_uid(struct user_struct
*);
1983 extern void release_uids(struct user_namespace
*ns
);
1985 #include <asm/current.h>
1987 extern void do_timer(unsigned long ticks
);
1989 extern int wake_up_state(struct task_struct
*tsk
, unsigned int state
);
1990 extern int wake_up_process(struct task_struct
*tsk
);
1991 extern void wake_up_new_task(struct task_struct
*tsk
,
1992 unsigned long clone_flags
);
1994 extern void kick_process(struct task_struct
*tsk
);
1996 static inline void kick_process(struct task_struct
*tsk
) { }
1998 extern void sched_fork(struct task_struct
*p
, int clone_flags
);
1999 extern void sched_dead(struct task_struct
*p
);
2001 extern void proc_caches_init(void);
2002 extern void flush_signals(struct task_struct
*);
2003 extern void __flush_signals(struct task_struct
*);
2004 extern void ignore_signals(struct task_struct
*);
2005 extern void flush_signal_handlers(struct task_struct
*, int force_default
);
2006 extern int dequeue_signal(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
);
2008 static inline int dequeue_signal_lock(struct task_struct
*tsk
, sigset_t
*mask
, siginfo_t
*info
)
2010 unsigned long flags
;
2013 spin_lock_irqsave(&tsk
->sighand
->siglock
, flags
);
2014 ret
= dequeue_signal(tsk
, mask
, info
);
2015 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, flags
);
2020 extern void block_all_signals(int (*notifier
)(void *priv
), void *priv
,
2022 extern void unblock_all_signals(void);
2023 extern void release_task(struct task_struct
* p
);
2024 extern int send_sig_info(int, struct siginfo
*, struct task_struct
*);
2025 extern int force_sigsegv(int, struct task_struct
*);
2026 extern int force_sig_info(int, struct siginfo
*, struct task_struct
*);
2027 extern int __kill_pgrp_info(int sig
, struct siginfo
*info
, struct pid
*pgrp
);
2028 extern int kill_pid_info(int sig
, struct siginfo
*info
, struct pid
*pid
);
2029 extern int kill_pid_info_as_uid(int, struct siginfo
*, struct pid
*, uid_t
, uid_t
, u32
);
2030 extern int kill_pgrp(struct pid
*pid
, int sig
, int priv
);
2031 extern int kill_pid(struct pid
*pid
, int sig
, int priv
);
2032 extern int kill_proc_info(int, struct siginfo
*, pid_t
);
2033 extern int do_notify_parent(struct task_struct
*, int);
2034 extern void force_sig(int, struct task_struct
*);
2035 extern void force_sig_specific(int, struct task_struct
*);
2036 extern int send_sig(int, struct task_struct
*, int);
2037 extern void zap_other_threads(struct task_struct
*p
);
2038 extern struct sigqueue
*sigqueue_alloc(void);
2039 extern void sigqueue_free(struct sigqueue
*);
2040 extern int send_sigqueue(struct sigqueue
*, struct task_struct
*, int group
);
2041 extern int do_sigaction(int, struct k_sigaction
*, struct k_sigaction
*);
2042 extern int do_sigaltstack(const stack_t __user
*, stack_t __user
*, unsigned long);
2044 static inline int kill_cad_pid(int sig
, int priv
)
2046 return kill_pid(cad_pid
, sig
, priv
);
2049 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2050 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2051 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2052 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2054 static inline int is_si_special(const struct siginfo
*info
)
2056 return info
<= SEND_SIG_FORCED
;
2059 /* True if we are on the alternate signal stack. */
2061 static inline int on_sig_stack(unsigned long sp
)
2063 return (sp
- current
->sas_ss_sp
< current
->sas_ss_size
);
2066 static inline int sas_ss_flags(unsigned long sp
)
2068 return (current
->sas_ss_size
== 0 ? SS_DISABLE
2069 : on_sig_stack(sp
) ? SS_ONSTACK
: 0);
2073 * Routines for handling mm_structs
2075 extern struct mm_struct
* mm_alloc(void);
2077 /* mmdrop drops the mm and the page tables */
2078 extern void __mmdrop(struct mm_struct
*);
2079 static inline void mmdrop(struct mm_struct
* mm
)
2081 if (unlikely(atomic_dec_and_test(&mm
->mm_count
)))
2085 /* mmput gets rid of the mappings and all user-space */
2086 extern void mmput(struct mm_struct
*);
2087 /* Grab a reference to a task's mm, if it is not already going away */
2088 extern struct mm_struct
*get_task_mm(struct task_struct
*task
);
2089 /* Remove the current tasks stale references to the old mm_struct */
2090 extern void mm_release(struct task_struct
*, struct mm_struct
*);
2091 /* Allocate a new mm structure and copy contents from tsk->mm */
2092 extern struct mm_struct
*dup_mm(struct task_struct
*tsk
);
2094 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2095 struct task_struct
*, struct pt_regs
*);
2096 extern void flush_thread(void);
2097 extern void exit_thread(void);
2099 extern void exit_files(struct task_struct
*);
2100 extern void __cleanup_signal(struct signal_struct
*);
2101 extern void __cleanup_sighand(struct sighand_struct
*);
2103 extern void exit_itimers(struct signal_struct
*);
2104 extern void flush_itimer_signals(void);
2106 extern NORET_TYPE
void do_group_exit(int);
2108 extern void daemonize(const char *, ...);
2109 extern int allow_signal(int);
2110 extern int disallow_signal(int);
2112 extern int do_execve(char *, char __user
* __user
*, char __user
* __user
*, struct pt_regs
*);
2113 extern long do_fork(unsigned long, unsigned long, struct pt_regs
*, unsigned long, int __user
*, int __user
*);
2114 struct task_struct
*fork_idle(int);
2116 extern void set_task_comm(struct task_struct
*tsk
, char *from
);
2117 extern char *get_task_comm(char *to
, struct task_struct
*tsk
);
2120 extern void wait_task_context_switch(struct task_struct
*p
);
2121 extern unsigned long wait_task_inactive(struct task_struct
*, long match_state
);
2123 static inline void wait_task_context_switch(struct task_struct
*p
) {}
2124 static inline unsigned long wait_task_inactive(struct task_struct
*p
,
2131 #define next_task(p) \
2132 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2134 #define for_each_process(p) \
2135 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2137 extern bool current_is_single_threaded(void);
2140 * Careful: do_each_thread/while_each_thread is a double loop so
2141 * 'break' will not work as expected - use goto instead.
2143 #define do_each_thread(g, t) \
2144 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2146 #define while_each_thread(g, t) \
2147 while ((t = next_thread(t)) != g)
2149 /* de_thread depends on thread_group_leader not being a pid based check */
2150 #define thread_group_leader(p) (p == p->group_leader)
2152 /* Do to the insanities of de_thread it is possible for a process
2153 * to have the pid of the thread group leader without actually being
2154 * the thread group leader. For iteration through the pids in proc
2155 * all we care about is that we have a task with the appropriate
2156 * pid, we don't actually care if we have the right task.
2158 static inline int has_group_leader_pid(struct task_struct
*p
)
2160 return p
->pid
== p
->tgid
;
2164 int same_thread_group(struct task_struct
*p1
, struct task_struct
*p2
)
2166 return p1
->tgid
== p2
->tgid
;
2169 static inline struct task_struct
*next_thread(const struct task_struct
*p
)
2171 return list_entry_rcu(p
->thread_group
.next
,
2172 struct task_struct
, thread_group
);
2175 static inline int thread_group_empty(struct task_struct
*p
)
2177 return list_empty(&p
->thread_group
);
2180 #define delay_group_leader(p) \
2181 (thread_group_leader(p) && !thread_group_empty(p))
2183 static inline int task_detached(struct task_struct
*p
)
2185 return p
->exit_signal
== -1;
2189 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2190 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2191 * pins the final release of task.io_context. Also protects ->cpuset and
2192 * ->cgroup.subsys[].
2194 * Nests both inside and outside of read_lock(&tasklist_lock).
2195 * It must not be nested with write_lock_irq(&tasklist_lock),
2196 * neither inside nor outside.
2198 static inline void task_lock(struct task_struct
*p
)
2200 spin_lock(&p
->alloc_lock
);
2203 static inline void task_unlock(struct task_struct
*p
)
2205 spin_unlock(&p
->alloc_lock
);
2208 extern struct sighand_struct
*lock_task_sighand(struct task_struct
*tsk
,
2209 unsigned long *flags
);
2211 static inline void unlock_task_sighand(struct task_struct
*tsk
,
2212 unsigned long *flags
)
2214 spin_unlock_irqrestore(&tsk
->sighand
->siglock
, *flags
);
2217 #ifndef __HAVE_THREAD_FUNCTIONS
2219 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2220 #define task_stack_page(task) ((task)->stack)
2222 static inline void setup_thread_stack(struct task_struct
*p
, struct task_struct
*org
)
2224 *task_thread_info(p
) = *task_thread_info(org
);
2225 task_thread_info(p
)->task
= p
;
2228 static inline unsigned long *end_of_stack(struct task_struct
*p
)
2230 return (unsigned long *)(task_thread_info(p
) + 1);
2235 static inline int object_is_on_stack(void *obj
)
2237 void *stack
= task_stack_page(current
);
2239 return (obj
>= stack
) && (obj
< (stack
+ THREAD_SIZE
));
2242 extern void thread_info_cache_init(void);
2244 #ifdef CONFIG_DEBUG_STACK_USAGE
2245 static inline unsigned long stack_not_used(struct task_struct
*p
)
2247 unsigned long *n
= end_of_stack(p
);
2249 do { /* Skip over canary */
2253 return (unsigned long)n
- (unsigned long)end_of_stack(p
);
2257 /* set thread flags in other task's structures
2258 * - see asm/thread_info.h for TIF_xxxx flags available
2260 static inline void set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2262 set_ti_thread_flag(task_thread_info(tsk
), flag
);
2265 static inline void clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2267 clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2270 static inline int test_and_set_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2272 return test_and_set_ti_thread_flag(task_thread_info(tsk
), flag
);
2275 static inline int test_and_clear_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2277 return test_and_clear_ti_thread_flag(task_thread_info(tsk
), flag
);
2280 static inline int test_tsk_thread_flag(struct task_struct
*tsk
, int flag
)
2282 return test_ti_thread_flag(task_thread_info(tsk
), flag
);
2285 static inline void set_tsk_need_resched(struct task_struct
*tsk
)
2287 set_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2290 static inline void clear_tsk_need_resched(struct task_struct
*tsk
)
2292 clear_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
);
2295 static inline int test_tsk_need_resched(struct task_struct
*tsk
)
2297 return unlikely(test_tsk_thread_flag(tsk
,TIF_NEED_RESCHED
));
2300 static inline int restart_syscall(void)
2302 set_tsk_thread_flag(current
, TIF_SIGPENDING
);
2303 return -ERESTARTNOINTR
;
2306 static inline int signal_pending(struct task_struct
*p
)
2308 return unlikely(test_tsk_thread_flag(p
,TIF_SIGPENDING
));
2311 extern int __fatal_signal_pending(struct task_struct
*p
);
2313 static inline int fatal_signal_pending(struct task_struct
*p
)
2315 return signal_pending(p
) && __fatal_signal_pending(p
);
2318 static inline int signal_pending_state(long state
, struct task_struct
*p
)
2320 if (!(state
& (TASK_INTERRUPTIBLE
| TASK_WAKEKILL
)))
2322 if (!signal_pending(p
))
2325 return (state
& TASK_INTERRUPTIBLE
) || __fatal_signal_pending(p
);
2328 static inline int need_resched(void)
2330 return unlikely(test_thread_flag(TIF_NEED_RESCHED
));
2334 * cond_resched() and cond_resched_lock(): latency reduction via
2335 * explicit rescheduling in places that are safe. The return
2336 * value indicates whether a reschedule was done in fact.
2337 * cond_resched_lock() will drop the spinlock before scheduling,
2338 * cond_resched_softirq() will enable bhs before scheduling.
2340 extern int _cond_resched(void);
2342 #define cond_resched() ({ \
2343 __might_sleep(__FILE__, __LINE__, 0); \
2347 extern int __cond_resched_lock(spinlock_t
*lock
);
2349 #ifdef CONFIG_PREEMPT
2350 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2352 #define PREEMPT_LOCK_OFFSET 0
2355 #define cond_resched_lock(lock) ({ \
2356 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2357 __cond_resched_lock(lock); \
2360 extern int __cond_resched_softirq(void);
2362 #define cond_resched_softirq() ({ \
2363 __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2364 __cond_resched_softirq(); \
2368 * Does a critical section need to be broken due to another
2369 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2370 * but a general need for low latency)
2372 static inline int spin_needbreak(spinlock_t
*lock
)
2374 #ifdef CONFIG_PREEMPT
2375 return spin_is_contended(lock
);
2382 * Thread group CPU time accounting.
2384 void thread_group_cputime(struct task_struct
*tsk
, struct task_cputime
*times
);
2385 void thread_group_cputimer(struct task_struct
*tsk
, struct task_cputime
*times
);
2387 static inline void thread_group_cputime_init(struct signal_struct
*sig
)
2389 sig
->cputimer
.cputime
= INIT_CPUTIME
;
2390 spin_lock_init(&sig
->cputimer
.lock
);
2391 sig
->cputimer
.running
= 0;
2394 static inline void thread_group_cputime_free(struct signal_struct
*sig
)
2399 * Reevaluate whether the task has signals pending delivery.
2400 * Wake the task if so.
2401 * This is required every time the blocked sigset_t changes.
2402 * callers must hold sighand->siglock.
2404 extern void recalc_sigpending_and_wake(struct task_struct
*t
);
2405 extern void recalc_sigpending(void);
2407 extern void signal_wake_up(struct task_struct
*t
, int resume_stopped
);
2410 * Wrappers for p->thread_info->cpu access. No-op on UP.
2414 static inline unsigned int task_cpu(const struct task_struct
*p
)
2416 return task_thread_info(p
)->cpu
;
2419 extern void set_task_cpu(struct task_struct
*p
, unsigned int cpu
);
2423 static inline unsigned int task_cpu(const struct task_struct
*p
)
2428 static inline void set_task_cpu(struct task_struct
*p
, unsigned int cpu
)
2432 #endif /* CONFIG_SMP */
2434 extern void arch_pick_mmap_layout(struct mm_struct
*mm
);
2436 #ifdef CONFIG_TRACING
2438 __trace_special(void *__tr
, void *__data
,
2439 unsigned long arg1
, unsigned long arg2
, unsigned long arg3
);
2442 __trace_special(void *__tr
, void *__data
,
2443 unsigned long arg1
, unsigned long arg2
, unsigned long arg3
)
2448 extern long sched_setaffinity(pid_t pid
, const struct cpumask
*new_mask
);
2449 extern long sched_getaffinity(pid_t pid
, struct cpumask
*mask
);
2451 extern void normalize_rt_tasks(void);
2453 #ifdef CONFIG_GROUP_SCHED
2455 extern struct task_group init_task_group
;
2456 #ifdef CONFIG_USER_SCHED
2457 extern struct task_group root_task_group
;
2458 extern void set_tg_uid(struct user_struct
*user
);
2461 extern struct task_group
*sched_create_group(struct task_group
*parent
);
2462 extern void sched_destroy_group(struct task_group
*tg
);
2463 extern void sched_move_task(struct task_struct
*tsk
);
2464 #ifdef CONFIG_FAIR_GROUP_SCHED
2465 extern int sched_group_set_shares(struct task_group
*tg
, unsigned long shares
);
2466 extern unsigned long sched_group_shares(struct task_group
*tg
);
2468 #ifdef CONFIG_RT_GROUP_SCHED
2469 extern int sched_group_set_rt_runtime(struct task_group
*tg
,
2470 long rt_runtime_us
);
2471 extern long sched_group_rt_runtime(struct task_group
*tg
);
2472 extern int sched_group_set_rt_period(struct task_group
*tg
,
2474 extern long sched_group_rt_period(struct task_group
*tg
);
2475 extern int sched_rt_can_attach(struct task_group
*tg
, struct task_struct
*tsk
);
2479 extern int task_can_switch_user(struct user_struct
*up
,
2480 struct task_struct
*tsk
);
2482 #ifdef CONFIG_TASK_XACCT
2483 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2485 tsk
->ioac
.rchar
+= amt
;
2488 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2490 tsk
->ioac
.wchar
+= amt
;
2493 static inline void inc_syscr(struct task_struct
*tsk
)
2498 static inline void inc_syscw(struct task_struct
*tsk
)
2503 static inline void add_rchar(struct task_struct
*tsk
, ssize_t amt
)
2507 static inline void add_wchar(struct task_struct
*tsk
, ssize_t amt
)
2511 static inline void inc_syscr(struct task_struct
*tsk
)
2515 static inline void inc_syscw(struct task_struct
*tsk
)
2520 #ifndef TASK_SIZE_OF
2521 #define TASK_SIZE_OF(tsk) TASK_SIZE
2525 * Call the function if the target task is executing on a CPU right now:
2527 extern void task_oncpu_function_call(struct task_struct
*p
,
2528 void (*func
) (void *info
), void *info
);
2531 #ifdef CONFIG_MM_OWNER
2532 extern void mm_update_next_owner(struct mm_struct
*mm
);
2533 extern void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
);
2535 static inline void mm_update_next_owner(struct mm_struct
*mm
)
2539 static inline void mm_init_owner(struct mm_struct
*mm
, struct task_struct
*p
)
2542 #endif /* CONFIG_MM_OWNER */
2544 #define TASK_STATE_TO_CHAR_STR "RSDTtZX"
2546 #endif /* __KERNEL__ */