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;
103 extern int disable_nx;
104 extern int print_fatal_signals;
107 * List of flags we want to share for kernel threads,
108 * if only because they are not used by them anyway.
110 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
113 * These are the constant used to fake the fixed-point load-average
114 * counting. Some notes:
115 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
116 * a load-average precision of 10 bits integer + 11 bits fractional
117 * - if you want to count load-averages more often, you need more
118 * precision, or rounding will get you. With 2-second counting freq,
119 * the EXP_n values would be 1981, 2034 and 2043 if still using only
122 extern unsigned long avenrun[]; /* Load averages */
123 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
125 #define FSHIFT 11 /* nr of bits of precision */
126 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
127 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
128 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
129 #define EXP_5 2014 /* 1/exp(5sec/5min) */
130 #define EXP_15 2037 /* 1/exp(5sec/15min) */
132 #define CALC_LOAD(load,exp,n) \
134 load += n*(FIXED_1-exp); \
137 extern unsigned long total_forks;
138 extern int nr_threads;
139 DECLARE_PER_CPU(unsigned long, process_counts);
140 extern int nr_processes(void);
141 extern unsigned long nr_running(void);
142 extern unsigned long nr_uninterruptible(void);
143 extern unsigned long nr_iowait(void);
144 extern unsigned long nr_iowait_cpu(int cpu);
145 extern unsigned long this_cpu_load(void);
148 extern void calc_global_load(unsigned long ticks);
150 extern unsigned long get_parent_ip(unsigned long addr);
155 #ifdef CONFIG_SCHED_DEBUG
156 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
157 extern void proc_sched_set_task(struct task_struct *p);
159 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
162 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
165 static inline void proc_sched_set_task(struct task_struct *p)
169 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
175 * Task state bitmask. NOTE! These bits are also
176 * encoded in fs/proc/array.c: get_task_state().
178 * We have two separate sets of flags: task->state
179 * is about runnability, while task->exit_state are
180 * about the task exiting. Confusing, but this way
181 * modifying one set can't modify the other one by
184 #define TASK_RUNNING 0
185 #define TASK_INTERRUPTIBLE 1
186 #define TASK_UNINTERRUPTIBLE 2
187 #define __TASK_STOPPED 4
188 #define __TASK_TRACED 8
189 /* in tsk->exit_state */
190 #define EXIT_ZOMBIE 16
192 /* in tsk->state again */
194 #define TASK_WAKEKILL 128
195 #define TASK_WAKING 256
196 #define TASK_STATE_MAX 512
198 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
200 extern char ___assert_task_state[1 - 2*!!(
201 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
203 /* Convenience macros for the sake of set_task_state */
204 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
205 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
206 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
208 /* Convenience macros for the sake of wake_up */
209 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
210 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
212 /* get_task_state() */
213 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
214 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
217 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
218 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
219 #define task_is_dead(task) ((task)->exit_state != 0)
220 #define task_is_stopped_or_traced(task) \
221 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
222 #define task_contributes_to_load(task) \
223 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
224 (task->flags & PF_FREEZING) == 0)
226 #define __set_task_state(tsk, state_value) \
227 do { (tsk)->state = (state_value); } while (0)
228 #define set_task_state(tsk, state_value) \
229 set_mb((tsk)->state, (state_value))
232 * set_current_state() includes a barrier so that the write of current->state
233 * is correctly serialised wrt the caller's subsequent test of whether to
236 * set_current_state(TASK_UNINTERRUPTIBLE);
237 * if (do_i_need_to_sleep())
240 * If the caller does not need such serialisation then use __set_current_state()
242 #define __set_current_state(state_value) \
243 do { current->state = (state_value); } while (0)
244 #define set_current_state(state_value) \
245 set_mb(current->state, (state_value))
247 /* Task command name length */
248 #define TASK_COMM_LEN 16
250 #include <linux/spinlock.h>
253 * This serializes "schedule()" and also protects
254 * the run-queue from deletions/modifications (but
255 * _adding_ to the beginning of the run-queue has
258 extern rwlock_t tasklist_lock;
259 extern spinlock_t mmlist_lock;
263 #ifdef CONFIG_PROVE_RCU
264 extern int lockdep_tasklist_lock_is_held(void);
265 #endif /* #ifdef CONFIG_PROVE_RCU */
267 extern void sched_init(void);
268 extern void sched_init_smp(void);
269 extern asmlinkage void schedule_tail(struct task_struct *prev);
270 extern void init_idle(struct task_struct *idle, int cpu);
271 extern void init_idle_bootup_task(struct task_struct *idle);
273 extern int runqueue_is_locked(int cpu);
275 extern cpumask_var_t nohz_cpu_mask;
276 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
277 extern void select_nohz_load_balancer(int stop_tick);
278 extern int get_nohz_timer_target(void);
280 static inline void select_nohz_load_balancer(int stop_tick) { }
284 * Only dump TASK_* tasks. (0 for all tasks)
286 extern void show_state_filter(unsigned long state_filter);
288 static inline void show_state(void)
290 show_state_filter(0);
293 extern void show_regs(struct pt_regs *);
296 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
297 * task), SP is the stack pointer of the first frame that should be shown in the back
298 * trace (or NULL if the entire call-chain of the task should be shown).
300 extern void show_stack(struct task_struct *task, unsigned long *sp);
302 void io_schedule(void);
303 long io_schedule_timeout(long timeout);
305 extern void cpu_init (void);
306 extern void trap_init(void);
307 extern void update_process_times(int user);
308 extern void scheduler_tick(void);
310 extern void sched_show_task(struct task_struct *p);
312 #ifdef CONFIG_LOCKUP_DETECTOR
313 extern void touch_softlockup_watchdog(void);
314 extern void touch_softlockup_watchdog_sync(void);
315 extern void touch_all_softlockup_watchdogs(void);
316 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
318 size_t *lenp, loff_t *ppos);
319 extern unsigned int softlockup_panic;
320 extern int softlockup_thresh;
321 void lockup_detector_init(void);
323 static inline void touch_softlockup_watchdog(void)
326 static inline void touch_softlockup_watchdog_sync(void)
329 static inline void touch_all_softlockup_watchdogs(void)
332 static inline void lockup_detector_init(void)
337 #ifdef CONFIG_DETECT_HUNG_TASK
338 extern unsigned int sysctl_hung_task_panic;
339 extern unsigned long sysctl_hung_task_check_count;
340 extern unsigned long sysctl_hung_task_timeout_secs;
341 extern unsigned long sysctl_hung_task_warnings;
342 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
344 size_t *lenp, loff_t *ppos);
346 /* Avoid need for ifdefs elsewhere in the code */
347 enum { sysctl_hung_task_timeout_secs = 0 };
350 /* Attach to any functions which should be ignored in wchan output. */
351 #define __sched __attribute__((__section__(".sched.text")))
353 /* Linker adds these: start and end of __sched functions */
354 extern char __sched_text_start[], __sched_text_end[];
356 /* Is this address in the __sched functions? */
357 extern int in_sched_functions(unsigned long addr);
359 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
360 extern signed long schedule_timeout(signed long timeout);
361 extern signed long schedule_timeout_interruptible(signed long timeout);
362 extern signed long schedule_timeout_killable(signed long timeout);
363 extern signed long schedule_timeout_uninterruptible(signed long timeout);
364 asmlinkage void schedule(void);
365 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
368 struct user_namespace;
371 * Default maximum number of active map areas, this limits the number of vmas
372 * per mm struct. Users can overwrite this number by sysctl but there is a
375 * When a program's coredump is generated as ELF format, a section is created
376 * per a vma. In ELF, the number of sections is represented in unsigned short.
377 * This means the number of sections should be smaller than 65535 at coredump.
378 * Because the kernel adds some informative sections to a image of program at
379 * generating coredump, we need some margin. The number of extra sections is
380 * 1-3 now and depends on arch. We use "5" as safe margin, here.
382 #define MAPCOUNT_ELF_CORE_MARGIN (5)
383 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
385 extern int sysctl_max_map_count;
387 #include <linux/aio.h>
390 extern void arch_pick_mmap_layout(struct mm_struct *mm);
392 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
393 unsigned long, unsigned long);
396 arch_get_unmapped_exec_area(struct file *, unsigned long, unsigned long,
397 unsigned long, unsigned long);
399 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
400 unsigned long len, unsigned long pgoff,
401 unsigned long flags);
402 extern void arch_unmap_area(struct mm_struct *, unsigned long);
403 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
405 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
409 extern void set_dumpable(struct mm_struct *mm, int value);
410 extern int get_dumpable(struct mm_struct *mm);
414 #define MMF_DUMPABLE 0 /* core dump is permitted */
415 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
417 #define MMF_DUMPABLE_BITS 2
418 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
420 /* coredump filter bits */
421 #define MMF_DUMP_ANON_PRIVATE 2
422 #define MMF_DUMP_ANON_SHARED 3
423 #define MMF_DUMP_MAPPED_PRIVATE 4
424 #define MMF_DUMP_MAPPED_SHARED 5
425 #define MMF_DUMP_ELF_HEADERS 6
426 #define MMF_DUMP_HUGETLB_PRIVATE 7
427 #define MMF_DUMP_HUGETLB_SHARED 8
429 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
430 #define MMF_DUMP_FILTER_BITS 7
431 #define MMF_DUMP_FILTER_MASK \
432 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
433 #define MMF_DUMP_FILTER_DEFAULT \
434 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
435 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
437 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
438 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
440 # define MMF_DUMP_MASK_DEFAULT_ELF 0
442 /* leave room for more dump flags */
443 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
444 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
446 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
448 struct sighand_struct {
450 struct k_sigaction action[_NSIG];
452 wait_queue_head_t signalfd_wqh;
455 struct pacct_struct {
458 unsigned long ac_mem;
459 cputime_t ac_utime, ac_stime;
460 unsigned long ac_minflt, ac_majflt;
471 * struct task_cputime - collected CPU time counts
472 * @utime: time spent in user mode, in &cputime_t units
473 * @stime: time spent in kernel mode, in &cputime_t units
474 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
476 * This structure groups together three kinds of CPU time that are
477 * tracked for threads and thread groups. Most things considering
478 * CPU time want to group these counts together and treat all three
479 * of them in parallel.
481 struct task_cputime {
484 unsigned long long sum_exec_runtime;
486 /* Alternate field names when used to cache expirations. */
487 #define prof_exp stime
488 #define virt_exp utime
489 #define sched_exp sum_exec_runtime
491 #define INIT_CPUTIME \
492 (struct task_cputime) { \
493 .utime = cputime_zero, \
494 .stime = cputime_zero, \
495 .sum_exec_runtime = 0, \
499 * Disable preemption until the scheduler is running.
500 * Reset by start_kernel()->sched_init()->init_idle().
502 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
503 * before the scheduler is active -- see should_resched().
505 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
508 * struct thread_group_cputimer - thread group interval timer counts
509 * @cputime: thread group interval timers.
510 * @running: non-zero when there are timers running and
511 * @cputime receives updates.
512 * @lock: lock for fields in this struct.
514 * This structure contains the version of task_cputime, above, that is
515 * used for thread group CPU timer calculations.
517 struct thread_group_cputimer {
518 struct task_cputime cputime;
526 * NOTE! "signal_struct" does not have it's own
527 * locking, because a shared signal_struct always
528 * implies a shared sighand_struct, so locking
529 * sighand_struct is always a proper superset of
530 * the locking of signal_struct.
532 struct signal_struct {
537 wait_queue_head_t wait_chldexit; /* for wait4() */
539 /* current thread group signal load-balancing target: */
540 struct task_struct *curr_target;
542 /* shared signal handling: */
543 struct sigpending shared_pending;
545 /* thread group exit support */
548 * - notify group_exit_task when ->count is equal to notify_count
549 * - everyone except group_exit_task is stopped during signal delivery
550 * of fatal signals, group_exit_task processes the signal.
553 struct task_struct *group_exit_task;
555 /* thread group stop support, overloads group_exit_code too */
556 int group_stop_count;
557 unsigned int flags; /* see SIGNAL_* flags below */
559 /* POSIX.1b Interval Timers */
560 struct list_head posix_timers;
562 /* ITIMER_REAL timer for the process */
563 struct hrtimer real_timer;
564 struct pid *leader_pid;
565 ktime_t it_real_incr;
568 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
569 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
570 * values are defined to 0 and 1 respectively
572 struct cpu_itimer it[2];
575 * Thread group totals for process CPU timers.
576 * See thread_group_cputimer(), et al, for details.
578 struct thread_group_cputimer cputimer;
580 /* Earliest-expiration cache. */
581 struct task_cputime cputime_expires;
583 struct list_head cpu_timers[3];
585 struct pid *tty_old_pgrp;
587 /* boolean value for session group leader */
590 struct tty_struct *tty; /* NULL if no tty */
592 #ifdef CONFIG_SCHED_AUTOGROUP
593 struct autogroup *autogroup;
596 * Cumulative resource counters for dead threads in the group,
597 * and for reaped dead child processes forked by this group.
598 * Live threads maintain their own counters and add to these
599 * in __exit_signal, except for the group leader.
601 cputime_t utime, stime, cutime, cstime;
604 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
605 cputime_t prev_utime, prev_stime;
607 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
608 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
609 unsigned long inblock, oublock, cinblock, coublock;
610 unsigned long maxrss, cmaxrss;
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) */
644 int oom_score_adj; /* OOM kill score adjustment */
645 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
646 * Only settable by CAP_SYS_RESOURCE. */
648 struct mutex cred_guard_mutex; /* guard against foreign influences on
649 * credential calculations
650 * (notably. ptrace) */
653 /* Context switch must be unlocked if interrupts are to be enabled */
654 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
655 # define __ARCH_WANT_UNLOCKED_CTXSW
659 * Bits in flags field of signal_struct.
661 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
662 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
663 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
664 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
666 * Pending notifications to parent.
668 #define SIGNAL_CLD_STOPPED 0x00000010
669 #define SIGNAL_CLD_CONTINUED 0x00000020
670 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
672 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
674 /* If true, all threads except ->group_exit_task have pending SIGKILL */
675 static inline int signal_group_exit(const struct signal_struct *sig)
677 return (sig->flags & SIGNAL_GROUP_EXIT) ||
678 (sig->group_exit_task != NULL);
682 * Some day this will be a full-fledged user tracking system..
685 atomic_t __count; /* reference count */
686 atomic_t processes; /* How many processes does this user have? */
687 atomic_t files; /* How many open files does this user have? */
688 atomic_t sigpending; /* How many pending signals does this user have? */
689 #ifdef CONFIG_INOTIFY_USER
690 atomic_t inotify_watches; /* How many inotify watches does this user have? */
691 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
693 #ifdef CONFIG_FANOTIFY
694 atomic_t fanotify_listeners;
697 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
699 #ifdef CONFIG_POSIX_MQUEUE
700 /* protected by mq_lock */
701 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
703 unsigned long locked_shm; /* How many pages of mlocked shm ? */
706 struct key *uid_keyring; /* UID specific keyring */
707 struct key *session_keyring; /* UID's default session keyring */
710 /* Hash table maintenance information */
711 struct hlist_node uidhash_node;
713 struct user_namespace *user_ns;
715 #ifdef CONFIG_PERF_EVENTS
716 atomic_long_t locked_vm;
720 extern int uids_sysfs_init(void);
722 extern struct user_struct *find_user(uid_t);
724 extern struct user_struct root_user;
725 #define INIT_USER (&root_user)
728 struct backing_dev_info;
729 struct reclaim_state;
731 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
733 /* cumulative counters */
734 unsigned long pcount; /* # of times run on this cpu */
735 unsigned long long run_delay; /* time spent waiting on a runqueue */
738 unsigned long long last_arrival,/* when we last ran on a cpu */
739 last_queued; /* when we were last queued to run */
740 #ifdef CONFIG_SCHEDSTATS
742 unsigned int bkl_count;
745 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
747 #ifdef CONFIG_TASK_DELAY_ACCT
748 struct task_delay_info {
750 unsigned int flags; /* Private per-task flags */
752 /* For each stat XXX, add following, aligned appropriately
754 * struct timespec XXX_start, XXX_end;
758 * Atomicity of updates to XXX_delay, XXX_count protected by
759 * single lock above (split into XXX_lock if contention is an issue).
763 * XXX_count is incremented on every XXX operation, the delay
764 * associated with the operation is added to XXX_delay.
765 * XXX_delay contains the accumulated delay time in nanoseconds.
767 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
768 u64 blkio_delay; /* wait for sync block io completion */
769 u64 swapin_delay; /* wait for swapin block io completion */
770 u32 blkio_count; /* total count of the number of sync block */
771 /* io operations performed */
772 u32 swapin_count; /* total count of the number of swapin block */
773 /* io operations performed */
775 struct timespec freepages_start, freepages_end;
776 u64 freepages_delay; /* wait for memory reclaim */
777 u32 freepages_count; /* total count of memory reclaim */
779 #endif /* CONFIG_TASK_DELAY_ACCT */
781 static inline int sched_info_on(void)
783 #ifdef CONFIG_SCHEDSTATS
785 #elif defined(CONFIG_TASK_DELAY_ACCT)
786 extern int delayacct_on;
801 * sched-domains (multiprocessor balancing) declarations:
805 * Increase resolution of nice-level calculations:
807 #define SCHED_LOAD_SHIFT 10
808 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
810 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
813 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
814 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
815 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
816 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
817 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
818 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
819 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
820 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
821 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
822 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
823 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
824 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
825 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
827 enum powersavings_balance_level {
828 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
829 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
830 * first for long running threads
832 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
833 * cpu package for power savings
835 MAX_POWERSAVINGS_BALANCE_LEVELS
838 extern int sched_mc_power_savings, sched_smt_power_savings;
840 static inline int sd_balance_for_mc_power(void)
842 if (sched_smt_power_savings)
843 return SD_POWERSAVINGS_BALANCE;
845 if (!sched_mc_power_savings)
846 return SD_PREFER_SIBLING;
851 static inline int sd_balance_for_package_power(void)
853 if (sched_mc_power_savings | sched_smt_power_savings)
854 return SD_POWERSAVINGS_BALANCE;
856 return SD_PREFER_SIBLING;
859 extern int __weak arch_sd_sibiling_asym_packing(void);
862 * Optimise SD flags for power savings:
863 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
864 * Keep default SD flags if sched_{smt,mc}_power_saving=0
867 static inline int sd_power_saving_flags(void)
869 if (sched_mc_power_savings | sched_smt_power_savings)
870 return SD_BALANCE_NEWIDLE;
876 struct sched_group *next; /* Must be a circular list */
879 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
882 unsigned int cpu_power, cpu_power_orig;
883 unsigned int group_weight;
886 * The CPUs this group covers.
888 * NOTE: this field is variable length. (Allocated dynamically
889 * by attaching extra space to the end of the structure,
890 * depending on how many CPUs the kernel has booted up with)
892 * It is also be embedded into static data structures at build
893 * time. (See 'struct static_sched_group' in kernel/sched.c)
895 unsigned long cpumask[0];
898 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
900 return to_cpumask(sg->cpumask);
903 enum sched_domain_level {
914 struct sched_domain_attr {
915 int relax_domain_level;
918 #define SD_ATTR_INIT (struct sched_domain_attr) { \
919 .relax_domain_level = -1, \
922 struct sched_domain {
923 /* These fields must be setup */
924 struct sched_domain *parent; /* top domain must be null terminated */
925 struct sched_domain *child; /* bottom domain must be null terminated */
926 struct sched_group *groups; /* the balancing groups of the domain */
927 unsigned long min_interval; /* Minimum balance interval ms */
928 unsigned long max_interval; /* Maximum balance interval ms */
929 unsigned int busy_factor; /* less balancing by factor if busy */
930 unsigned int imbalance_pct; /* No balance until over watermark */
931 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
932 unsigned int busy_idx;
933 unsigned int idle_idx;
934 unsigned int newidle_idx;
935 unsigned int wake_idx;
936 unsigned int forkexec_idx;
937 unsigned int smt_gain;
938 int flags; /* See SD_* */
939 enum sched_domain_level level;
941 /* Runtime fields. */
942 unsigned long last_balance; /* init to jiffies. units in jiffies */
943 unsigned int balance_interval; /* initialise to 1. units in ms. */
944 unsigned int nr_balance_failed; /* initialise to 0 */
948 #ifdef CONFIG_SCHEDSTATS
949 /* load_balance() stats */
950 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
952 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
953 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
954 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
955 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
956 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
957 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
959 /* Active load balancing */
960 unsigned int alb_count;
961 unsigned int alb_failed;
962 unsigned int alb_pushed;
964 /* SD_BALANCE_EXEC stats */
965 unsigned int sbe_count;
966 unsigned int sbe_balanced;
967 unsigned int sbe_pushed;
969 /* SD_BALANCE_FORK stats */
970 unsigned int sbf_count;
971 unsigned int sbf_balanced;
972 unsigned int sbf_pushed;
974 /* try_to_wake_up() stats */
975 unsigned int ttwu_wake_remote;
976 unsigned int ttwu_move_affine;
977 unsigned int ttwu_move_balance;
979 #ifdef CONFIG_SCHED_DEBUG
983 unsigned int span_weight;
985 * Span of all CPUs in this domain.
987 * NOTE: this field is variable length. (Allocated dynamically
988 * by attaching extra space to the end of the structure,
989 * depending on how many CPUs the kernel has booted up with)
991 * It is also be embedded into static data structures at build
992 * time. (See 'struct static_sched_domain' in kernel/sched.c)
994 unsigned long span[0];
997 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
999 return to_cpumask(sd->span);
1002 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1003 struct sched_domain_attr *dattr_new);
1005 /* Allocate an array of sched domains, for partition_sched_domains(). */
1006 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1007 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1009 /* Test a flag in parent sched domain */
1010 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1012 if (sd->parent && (sd->parent->flags & flag))
1018 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1021 #else /* CONFIG_SMP */
1023 struct sched_domain_attr;
1026 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1027 struct sched_domain_attr *dattr_new)
1030 #endif /* !CONFIG_SMP */
1033 struct io_context; /* See blkdev.h */
1036 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1037 extern void prefetch_stack(struct task_struct *t);
1039 static inline void prefetch_stack(struct task_struct *t) { }
1042 struct audit_context; /* See audit.c */
1044 struct pipe_inode_info;
1045 struct uts_namespace;
1048 struct sched_domain;
1053 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1054 #define WF_FORK 0x02 /* child wakeup after fork */
1056 #define ENQUEUE_WAKEUP 1
1057 #define ENQUEUE_WAKING 2
1058 #define ENQUEUE_HEAD 4
1060 #define DEQUEUE_SLEEP 1
1062 struct sched_class {
1063 const struct sched_class *next;
1065 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1066 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1067 void (*yield_task) (struct rq *rq);
1069 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1071 struct task_struct * (*pick_next_task) (struct rq *rq);
1072 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1075 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1076 int sd_flag, int flags);
1078 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1079 void (*post_schedule) (struct rq *this_rq);
1080 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1081 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1083 void (*set_cpus_allowed)(struct task_struct *p,
1084 const struct cpumask *newmask);
1086 void (*rq_online)(struct rq *rq);
1087 void (*rq_offline)(struct rq *rq);
1090 void (*set_curr_task) (struct rq *rq);
1091 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1092 void (*task_fork) (struct task_struct *p);
1094 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1096 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1098 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1099 int oldprio, int running);
1101 unsigned int (*get_rr_interval) (struct rq *rq,
1102 struct task_struct *task);
1104 #ifdef CONFIG_FAIR_GROUP_SCHED
1105 void (*task_move_group) (struct task_struct *p, int on_rq);
1109 struct load_weight {
1110 unsigned long weight, inv_weight;
1113 #ifdef CONFIG_SCHEDSTATS
1114 struct sched_statistics {
1124 s64 sum_sleep_runtime;
1131 u64 nr_migrations_cold;
1132 u64 nr_failed_migrations_affine;
1133 u64 nr_failed_migrations_running;
1134 u64 nr_failed_migrations_hot;
1135 u64 nr_forced_migrations;
1138 u64 nr_wakeups_sync;
1139 u64 nr_wakeups_migrate;
1140 u64 nr_wakeups_local;
1141 u64 nr_wakeups_remote;
1142 u64 nr_wakeups_affine;
1143 u64 nr_wakeups_affine_attempts;
1144 u64 nr_wakeups_passive;
1145 u64 nr_wakeups_idle;
1149 struct sched_entity {
1150 struct load_weight load; /* for load-balancing */
1151 struct rb_node run_node;
1152 struct list_head group_node;
1156 u64 sum_exec_runtime;
1158 u64 prev_sum_exec_runtime;
1162 #ifdef CONFIG_SCHEDSTATS
1163 struct sched_statistics statistics;
1166 #ifdef CONFIG_FAIR_GROUP_SCHED
1167 struct sched_entity *parent;
1168 /* rq on which this entity is (to be) queued: */
1169 struct cfs_rq *cfs_rq;
1170 /* rq "owned" by this entity/group: */
1171 struct cfs_rq *my_q;
1175 struct sched_rt_entity {
1176 struct list_head run_list;
1177 unsigned long timeout;
1178 unsigned int time_slice;
1179 int nr_cpus_allowed;
1181 struct sched_rt_entity *back;
1182 #ifdef CONFIG_RT_GROUP_SCHED
1183 struct sched_rt_entity *parent;
1184 /* rq on which this entity is (to be) queued: */
1185 struct rt_rq *rt_rq;
1186 /* rq "owned" by this entity/group: */
1193 enum perf_event_task_context {
1194 perf_invalid_context = -1,
1195 perf_hw_context = 0,
1197 perf_nr_task_contexts,
1200 struct task_struct {
1201 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1204 unsigned int flags; /* per process flags, defined below */
1205 unsigned int ptrace;
1207 int lock_depth; /* BKL lock depth */
1210 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1215 int prio, static_prio, normal_prio;
1216 unsigned int rt_priority;
1217 const struct sched_class *sched_class;
1218 struct sched_entity se;
1219 struct sched_rt_entity rt;
1221 #ifdef CONFIG_PREEMPT_NOTIFIERS
1222 /* list of struct preempt_notifier: */
1223 struct hlist_head preempt_notifiers;
1227 * fpu_counter contains the number of consecutive context switches
1228 * that the FPU is used. If this is over a threshold, the lazy fpu
1229 * saving becomes unlazy to save the trap. This is an unsigned char
1230 * so that after 256 times the counter wraps and the behavior turns
1231 * lazy again; this to deal with bursty apps that only use FPU for
1234 unsigned char fpu_counter;
1235 #ifdef CONFIG_BLK_DEV_IO_TRACE
1236 unsigned int btrace_seq;
1239 unsigned int policy;
1240 cpumask_t cpus_allowed;
1242 #ifdef CONFIG_PREEMPT_RCU
1243 int rcu_read_lock_nesting;
1244 char rcu_read_unlock_special;
1245 struct list_head rcu_node_entry;
1246 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1247 #ifdef CONFIG_TREE_PREEMPT_RCU
1248 struct rcu_node *rcu_blocked_node;
1249 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1250 #ifdef CONFIG_RCU_BOOST
1251 struct rt_mutex *rcu_boost_mutex;
1252 #endif /* #ifdef CONFIG_RCU_BOOST */
1254 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1255 struct sched_info sched_info;
1258 struct list_head tasks;
1260 struct plist_node pushable_tasks;
1263 struct mm_struct *mm, *active_mm;
1264 #if defined(SPLIT_RSS_COUNTING)
1265 struct task_rss_stat rss_stat;
1269 int exit_code, exit_signal;
1270 int pdeath_signal; /* The signal sent when the parent dies */
1272 unsigned int personality;
1273 unsigned did_exec:1;
1274 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1276 unsigned in_iowait:1;
1279 /* Revert to default priority/policy when forking */
1280 unsigned sched_reset_on_fork:1;
1285 #ifdef CONFIG_CC_STACKPROTECTOR
1286 /* Canary value for the -fstack-protector gcc feature */
1287 unsigned long stack_canary;
1291 * pointers to (original) parent process, youngest child, younger sibling,
1292 * older sibling, respectively. (p->father can be replaced with
1293 * p->real_parent->pid)
1295 struct task_struct *real_parent; /* real parent process */
1296 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1298 * children/sibling forms the list of my natural children
1300 struct list_head children; /* list of my children */
1301 struct list_head sibling; /* linkage in my parent's children list */
1302 struct task_struct *group_leader; /* threadgroup leader */
1305 * ptraced is the list of tasks this task is using ptrace on.
1306 * This includes both natural children and PTRACE_ATTACH targets.
1307 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1309 struct list_head ptraced;
1310 struct list_head ptrace_entry;
1312 /* PID/PID hash table linkage. */
1313 struct pid_link pids[PIDTYPE_MAX];
1314 struct list_head thread_group;
1316 struct completion *vfork_done; /* for vfork() */
1317 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1318 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1320 cputime_t utime, stime, utimescaled, stimescaled;
1322 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1323 cputime_t prev_utime, prev_stime;
1325 unsigned long nvcsw, nivcsw; /* context switch counts */
1326 struct timespec start_time; /* monotonic time */
1327 struct timespec real_start_time; /* boot based time */
1328 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1329 unsigned long min_flt, maj_flt;
1331 struct task_cputime cputime_expires;
1332 struct list_head cpu_timers[3];
1334 /* process credentials */
1335 const struct cred __rcu *real_cred; /* objective and real subjective task
1336 * credentials (COW) */
1337 const struct cred __rcu *cred; /* effective (overridable) subjective task
1338 * credentials (COW) */
1339 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1341 char comm[TASK_COMM_LEN]; /* executable name excluding path
1342 - access with [gs]et_task_comm (which lock
1343 it with task_lock())
1344 - initialized normally by setup_new_exec */
1345 /* file system info */
1346 int link_count, total_link_count;
1347 #ifdef CONFIG_SYSVIPC
1349 struct sysv_sem sysvsem;
1351 #ifdef CONFIG_DETECT_HUNG_TASK
1352 /* hung task detection */
1353 unsigned long last_switch_count;
1355 /* CPU-specific state of this task */
1356 struct thread_struct thread;
1357 /* filesystem information */
1358 struct fs_struct *fs;
1359 /* open file information */
1360 struct files_struct *files;
1362 struct nsproxy *nsproxy;
1363 /* signal handlers */
1364 struct signal_struct *signal;
1365 struct sighand_struct *sighand;
1367 sigset_t blocked, real_blocked;
1368 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1369 struct sigpending pending;
1371 unsigned long sas_ss_sp;
1373 int (*notifier)(void *priv);
1374 void *notifier_data;
1375 sigset_t *notifier_mask;
1376 struct audit_context *audit_context;
1377 #ifdef CONFIG_AUDITSYSCALL
1379 unsigned int sessionid;
1383 /* Thread group tracking */
1386 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1388 spinlock_t alloc_lock;
1390 #ifdef CONFIG_GENERIC_HARDIRQS
1391 /* IRQ handler threads */
1392 struct irqaction *irqaction;
1395 /* Protection of the PI data structures: */
1396 raw_spinlock_t pi_lock;
1398 #ifdef CONFIG_RT_MUTEXES
1399 /* PI waiters blocked on a rt_mutex held by this task */
1400 struct plist_head pi_waiters;
1401 /* Deadlock detection and priority inheritance handling */
1402 struct rt_mutex_waiter *pi_blocked_on;
1405 #ifdef CONFIG_DEBUG_MUTEXES
1406 /* mutex deadlock detection */
1407 struct mutex_waiter *blocked_on;
1409 #ifdef CONFIG_TRACE_IRQFLAGS
1410 unsigned int irq_events;
1411 unsigned long hardirq_enable_ip;
1412 unsigned long hardirq_disable_ip;
1413 unsigned int hardirq_enable_event;
1414 unsigned int hardirq_disable_event;
1415 int hardirqs_enabled;
1416 int hardirq_context;
1417 unsigned long softirq_disable_ip;
1418 unsigned long softirq_enable_ip;
1419 unsigned int softirq_disable_event;
1420 unsigned int softirq_enable_event;
1421 int softirqs_enabled;
1422 int softirq_context;
1424 #ifdef CONFIG_LOCKDEP
1425 # define MAX_LOCK_DEPTH 48UL
1428 unsigned int lockdep_recursion;
1429 struct held_lock held_locks[MAX_LOCK_DEPTH];
1430 gfp_t lockdep_reclaim_gfp;
1433 /* journalling filesystem info */
1436 /* stacked block device info */
1437 struct bio_list *bio_list;
1440 struct reclaim_state *reclaim_state;
1442 struct backing_dev_info *backing_dev_info;
1444 struct io_context *io_context;
1446 unsigned long ptrace_message;
1447 siginfo_t *last_siginfo; /* For ptrace use. */
1448 struct task_io_accounting ioac;
1449 #if defined(CONFIG_TASK_XACCT)
1450 u64 acct_rss_mem1; /* accumulated rss usage */
1451 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1452 cputime_t acct_timexpd; /* stime + utime since last update */
1454 #ifdef CONFIG_CPUSETS
1455 nodemask_t mems_allowed; /* Protected by alloc_lock */
1456 int mems_allowed_change_disable;
1457 int cpuset_mem_spread_rotor;
1458 int cpuset_slab_spread_rotor;
1460 #ifdef CONFIG_CGROUPS
1461 /* Control Group info protected by css_set_lock */
1462 struct css_set __rcu *cgroups;
1463 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1464 struct list_head cg_list;
1467 struct robust_list_head __user *robust_list;
1468 #ifdef CONFIG_COMPAT
1469 struct compat_robust_list_head __user *compat_robust_list;
1471 struct list_head pi_state_list;
1472 struct futex_pi_state *pi_state_cache;
1474 #ifdef CONFIG_PERF_EVENTS
1475 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1476 struct mutex perf_event_mutex;
1477 struct list_head perf_event_list;
1480 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1483 atomic_t fs_excl; /* holding fs exclusive resources */
1484 struct rcu_head rcu;
1487 * cache last used pipe for splice
1489 struct pipe_inode_info *splice_pipe;
1490 #ifdef CONFIG_TASK_DELAY_ACCT
1491 struct task_delay_info *delays;
1493 #ifdef CONFIG_FAULT_INJECTION
1496 struct prop_local_single dirties;
1497 #ifdef CONFIG_LATENCYTOP
1498 int latency_record_count;
1499 struct latency_record latency_record[LT_SAVECOUNT];
1502 * time slack values; these are used to round up poll() and
1503 * select() etc timeout values. These are in nanoseconds.
1505 unsigned long timer_slack_ns;
1506 unsigned long default_timer_slack_ns;
1508 struct list_head *scm_work_list;
1509 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1510 /* Index of current stored address in ret_stack */
1512 /* Stack of return addresses for return function tracing */
1513 struct ftrace_ret_stack *ret_stack;
1514 /* time stamp for last schedule */
1515 unsigned long long ftrace_timestamp;
1517 * Number of functions that haven't been traced
1518 * because of depth overrun.
1520 atomic_t trace_overrun;
1521 /* Pause for the tracing */
1522 atomic_t tracing_graph_pause;
1524 #ifdef CONFIG_TRACING
1525 /* state flags for use by tracers */
1526 unsigned long trace;
1527 /* bitmask of trace recursion */
1528 unsigned long trace_recursion;
1529 #endif /* CONFIG_TRACING */
1530 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1531 struct memcg_batch_info {
1532 int do_batch; /* incremented when batch uncharge started */
1533 struct mem_cgroup *memcg; /* target memcg of uncharge */
1534 unsigned long bytes; /* uncharged usage */
1535 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1540 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1541 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1544 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1545 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1546 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1547 * values are inverted: lower p->prio value means higher priority.
1549 * The MAX_USER_RT_PRIO value allows the actual maximum
1550 * RT priority to be separate from the value exported to
1551 * user-space. This allows kernel threads to set their
1552 * priority to a value higher than any user task. Note:
1553 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1556 #define MAX_USER_RT_PRIO 100
1557 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1559 #define MAX_PRIO (MAX_RT_PRIO + 40)
1560 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1562 static inline int rt_prio(int prio)
1564 if (unlikely(prio < MAX_RT_PRIO))
1569 static inline int rt_task(struct task_struct *p)
1571 return rt_prio(p->prio);
1574 static inline struct pid *task_pid(struct task_struct *task)
1576 return task->pids[PIDTYPE_PID].pid;
1579 static inline struct pid *task_tgid(struct task_struct *task)
1581 return task->group_leader->pids[PIDTYPE_PID].pid;
1585 * Without tasklist or rcu lock it is not safe to dereference
1586 * the result of task_pgrp/task_session even if task == current,
1587 * we can race with another thread doing sys_setsid/sys_setpgid.
1589 static inline struct pid *task_pgrp(struct task_struct *task)
1591 return task->group_leader->pids[PIDTYPE_PGID].pid;
1594 static inline struct pid *task_session(struct task_struct *task)
1596 return task->group_leader->pids[PIDTYPE_SID].pid;
1599 struct pid_namespace;
1602 * the helpers to get the task's different pids as they are seen
1603 * from various namespaces
1605 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1606 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1608 * task_xid_nr_ns() : id seen from the ns specified;
1610 * set_task_vxid() : assigns a virtual id to a task;
1612 * see also pid_nr() etc in include/linux/pid.h
1614 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1615 struct pid_namespace *ns);
1617 static inline pid_t task_pid_nr(struct task_struct *tsk)
1622 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1623 struct pid_namespace *ns)
1625 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1628 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1630 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1634 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1639 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1641 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1643 return pid_vnr(task_tgid(tsk));
1647 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1648 struct pid_namespace *ns)
1650 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1653 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1655 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1659 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1660 struct pid_namespace *ns)
1662 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1665 static inline pid_t task_session_vnr(struct task_struct *tsk)
1667 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1670 /* obsolete, do not use */
1671 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1673 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1677 * pid_alive - check that a task structure is not stale
1678 * @p: Task structure to be checked.
1680 * Test if a process is not yet dead (at most zombie state)
1681 * If pid_alive fails, then pointers within the task structure
1682 * can be stale and must not be dereferenced.
1684 static inline int pid_alive(struct task_struct *p)
1686 return p->pids[PIDTYPE_PID].pid != NULL;
1690 * is_global_init - check if a task structure is init
1691 * @tsk: Task structure to be checked.
1693 * Check if a task structure is the first user space task the kernel created.
1695 static inline int is_global_init(struct task_struct *tsk)
1697 return tsk->pid == 1;
1701 * is_container_init:
1702 * check whether in the task is init in its own pid namespace.
1704 extern int is_container_init(struct task_struct *tsk);
1706 extern struct pid *cad_pid;
1708 extern void free_task(struct task_struct *tsk);
1709 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1711 extern void __put_task_struct(struct task_struct *t);
1713 static inline void put_task_struct(struct task_struct *t)
1715 if (atomic_dec_and_test(&t->usage))
1716 __put_task_struct(t);
1719 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1720 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1725 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1726 #define PF_STARTING 0x00000002 /* being created */
1727 #define PF_EXITING 0x00000004 /* getting shut down */
1728 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1729 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1730 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1731 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1732 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1733 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1734 #define PF_DUMPCORE 0x00000200 /* dumped core */
1735 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1736 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1737 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1738 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1739 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1740 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1741 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1742 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1743 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1744 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1745 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1746 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1747 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1748 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1749 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1750 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1751 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1752 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1753 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1754 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1755 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1758 * Only the _current_ task can read/write to tsk->flags, but other
1759 * tasks can access tsk->flags in readonly mode for example
1760 * with tsk_used_math (like during threaded core dumping).
1761 * There is however an exception to this rule during ptrace
1762 * or during fork: the ptracer task is allowed to write to the
1763 * child->flags of its traced child (same goes for fork, the parent
1764 * can write to the child->flags), because we're guaranteed the
1765 * child is not running and in turn not changing child->flags
1766 * at the same time the parent does it.
1768 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1769 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1770 #define clear_used_math() clear_stopped_child_used_math(current)
1771 #define set_used_math() set_stopped_child_used_math(current)
1772 #define conditional_stopped_child_used_math(condition, child) \
1773 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1774 #define conditional_used_math(condition) \
1775 conditional_stopped_child_used_math(condition, current)
1776 #define copy_to_stopped_child_used_math(child) \
1777 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1778 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1779 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1780 #define used_math() tsk_used_math(current)
1782 #ifdef CONFIG_PREEMPT_RCU
1784 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1785 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1786 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1788 static inline void rcu_copy_process(struct task_struct *p)
1790 p->rcu_read_lock_nesting = 0;
1791 p->rcu_read_unlock_special = 0;
1792 #ifdef CONFIG_TREE_PREEMPT_RCU
1793 p->rcu_blocked_node = NULL;
1794 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1795 #ifdef CONFIG_RCU_BOOST
1796 p->rcu_boost_mutex = NULL;
1797 #endif /* #ifdef CONFIG_RCU_BOOST */
1798 INIT_LIST_HEAD(&p->rcu_node_entry);
1803 static inline void rcu_copy_process(struct task_struct *p)
1810 extern int set_cpus_allowed_ptr(struct task_struct *p,
1811 const struct cpumask *new_mask);
1813 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1814 const struct cpumask *new_mask)
1816 if (!cpumask_test_cpu(0, new_mask))
1822 #ifndef CONFIG_CPUMASK_OFFSTACK
1823 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1825 return set_cpus_allowed_ptr(p, &new_mask);
1830 * Do not use outside of architecture code which knows its limitations.
1832 * sched_clock() has no promise of monotonicity or bounded drift between
1833 * CPUs, use (which you should not) requires disabling IRQs.
1835 * Please use one of the three interfaces below.
1837 extern unsigned long long notrace sched_clock(void);
1839 * See the comment in kernel/sched_clock.c
1841 extern u64 cpu_clock(int cpu);
1842 extern u64 local_clock(void);
1843 extern u64 sched_clock_cpu(int cpu);
1846 extern void sched_clock_init(void);
1848 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1849 static inline void sched_clock_tick(void)
1853 static inline void sched_clock_idle_sleep_event(void)
1857 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1862 * Architectures can set this to 1 if they have specified
1863 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1864 * but then during bootup it turns out that sched_clock()
1865 * is reliable after all:
1867 extern int sched_clock_stable;
1869 extern void sched_clock_tick(void);
1870 extern void sched_clock_idle_sleep_event(void);
1871 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1874 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1876 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1877 * The reason for this explicit opt-in is not to have perf penalty with
1878 * slow sched_clocks.
1880 extern void enable_sched_clock_irqtime(void);
1881 extern void disable_sched_clock_irqtime(void);
1883 static inline void enable_sched_clock_irqtime(void) {}
1884 static inline void disable_sched_clock_irqtime(void) {}
1887 extern unsigned long long
1888 task_sched_runtime(struct task_struct *task);
1889 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1891 /* sched_exec is called by processes performing an exec */
1893 extern void sched_exec(void);
1895 #define sched_exec() {}
1898 extern void sched_clock_idle_sleep_event(void);
1899 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1901 #ifdef CONFIG_HOTPLUG_CPU
1902 extern void idle_task_exit(void);
1904 static inline void idle_task_exit(void) {}
1907 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1908 extern void wake_up_idle_cpu(int cpu);
1910 static inline void wake_up_idle_cpu(int cpu) { }
1913 extern unsigned int sysctl_sched_latency;
1914 extern unsigned int sysctl_sched_min_granularity;
1915 extern unsigned int sysctl_sched_wakeup_granularity;
1916 extern unsigned int sysctl_sched_child_runs_first;
1918 enum sched_tunable_scaling {
1919 SCHED_TUNABLESCALING_NONE,
1920 SCHED_TUNABLESCALING_LOG,
1921 SCHED_TUNABLESCALING_LINEAR,
1922 SCHED_TUNABLESCALING_END,
1924 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1926 #ifdef CONFIG_SCHED_DEBUG
1927 extern unsigned int sysctl_sched_migration_cost;
1928 extern unsigned int sysctl_sched_nr_migrate;
1929 extern unsigned int sysctl_sched_time_avg;
1930 extern unsigned int sysctl_timer_migration;
1931 extern unsigned int sysctl_sched_shares_window;
1933 int sched_proc_update_handler(struct ctl_table *table, int write,
1934 void __user *buffer, size_t *length,
1937 #ifdef CONFIG_SCHED_DEBUG
1938 static inline unsigned int get_sysctl_timer_migration(void)
1940 return sysctl_timer_migration;
1943 static inline unsigned int get_sysctl_timer_migration(void)
1948 extern unsigned int sysctl_sched_rt_period;
1949 extern int sysctl_sched_rt_runtime;
1951 int sched_rt_handler(struct ctl_table *table, int write,
1952 void __user *buffer, size_t *lenp,
1955 extern unsigned int sysctl_sched_compat_yield;
1957 #ifdef CONFIG_SCHED_AUTOGROUP
1958 extern unsigned int sysctl_sched_autogroup_enabled;
1960 extern void sched_autogroup_create_attach(struct task_struct *p);
1961 extern void sched_autogroup_detach(struct task_struct *p);
1962 extern void sched_autogroup_fork(struct signal_struct *sig);
1963 extern void sched_autogroup_exit(struct signal_struct *sig);
1964 #ifdef CONFIG_PROC_FS
1965 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1966 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1969 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1970 static inline void sched_autogroup_detach(struct task_struct *p) { }
1971 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1972 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1975 #ifdef CONFIG_RT_MUTEXES
1976 extern int rt_mutex_getprio(struct task_struct *p);
1977 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1978 extern void rt_mutex_adjust_pi(struct task_struct *p);
1980 static inline int rt_mutex_getprio(struct task_struct *p)
1982 return p->normal_prio;
1984 # define rt_mutex_adjust_pi(p) do { } while (0)
1987 extern void set_user_nice(struct task_struct *p, long nice);
1988 extern int task_prio(const struct task_struct *p);
1989 extern int task_nice(const struct task_struct *p);
1990 extern int can_nice(const struct task_struct *p, const int nice);
1991 extern int task_curr(const struct task_struct *p);
1992 extern int idle_cpu(int cpu);
1993 extern int sched_setscheduler(struct task_struct *, int,
1994 const struct sched_param *);
1995 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1996 const struct sched_param *);
1997 extern struct task_struct *idle_task(int cpu);
1998 extern struct task_struct *curr_task(int cpu);
1999 extern void set_curr_task(int cpu, struct task_struct *p);
2004 * The default (Linux) execution domain.
2006 extern struct exec_domain default_exec_domain;
2008 union thread_union {
2009 struct thread_info thread_info;
2010 unsigned long stack[THREAD_SIZE/sizeof(long)];
2013 #ifndef __HAVE_ARCH_KSTACK_END
2014 static inline int kstack_end(void *addr)
2016 /* Reliable end of stack detection:
2017 * Some APM bios versions misalign the stack
2019 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2023 extern union thread_union init_thread_union;
2024 extern struct task_struct init_task;
2026 extern struct mm_struct init_mm;
2028 extern struct pid_namespace init_pid_ns;
2031 * find a task by one of its numerical ids
2033 * find_task_by_pid_ns():
2034 * finds a task by its pid in the specified namespace
2035 * find_task_by_vpid():
2036 * finds a task by its virtual pid
2038 * see also find_vpid() etc in include/linux/pid.h
2041 extern struct task_struct *find_task_by_vpid(pid_t nr);
2042 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2043 struct pid_namespace *ns);
2045 extern void __set_special_pids(struct pid *pid);
2047 /* per-UID process charging. */
2048 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2049 static inline struct user_struct *get_uid(struct user_struct *u)
2051 atomic_inc(&u->__count);
2054 extern void free_uid(struct user_struct *);
2055 extern void release_uids(struct user_namespace *ns);
2057 #include <asm/current.h>
2059 extern void do_timer(unsigned long ticks);
2061 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2062 extern int wake_up_process(struct task_struct *tsk);
2063 extern void wake_up_new_task(struct task_struct *tsk,
2064 unsigned long clone_flags);
2066 extern void kick_process(struct task_struct *tsk);
2068 static inline void kick_process(struct task_struct *tsk) { }
2070 extern void sched_fork(struct task_struct *p, int clone_flags);
2071 extern void sched_dead(struct task_struct *p);
2073 extern void proc_caches_init(void);
2074 extern void flush_signals(struct task_struct *);
2075 extern void __flush_signals(struct task_struct *);
2076 extern void ignore_signals(struct task_struct *);
2077 extern void flush_signal_handlers(struct task_struct *, int force_default);
2078 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2080 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2082 unsigned long flags;
2085 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2086 ret = dequeue_signal(tsk, mask, info);
2087 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2092 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2094 extern void unblock_all_signals(void);
2095 extern void release_task(struct task_struct * p);
2096 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2097 extern int force_sigsegv(int, struct task_struct *);
2098 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2099 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2100 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2101 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2102 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2103 extern int kill_pid(struct pid *pid, int sig, int priv);
2104 extern int kill_proc_info(int, struct siginfo *, pid_t);
2105 extern int do_notify_parent(struct task_struct *, int);
2106 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2107 extern void force_sig(int, struct task_struct *);
2108 extern int send_sig(int, struct task_struct *, int);
2109 extern int zap_other_threads(struct task_struct *p);
2110 extern struct sigqueue *sigqueue_alloc(void);
2111 extern void sigqueue_free(struct sigqueue *);
2112 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2113 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2114 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2116 static inline int kill_cad_pid(int sig, int priv)
2118 return kill_pid(cad_pid, sig, priv);
2121 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2122 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2123 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2124 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2127 * True if we are on the alternate signal stack.
2129 static inline int on_sig_stack(unsigned long sp)
2131 #ifdef CONFIG_STACK_GROWSUP
2132 return sp >= current->sas_ss_sp &&
2133 sp - current->sas_ss_sp < current->sas_ss_size;
2135 return sp > current->sas_ss_sp &&
2136 sp - current->sas_ss_sp <= current->sas_ss_size;
2140 static inline int sas_ss_flags(unsigned long sp)
2142 return (current->sas_ss_size == 0 ? SS_DISABLE
2143 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2147 * Routines for handling mm_structs
2149 extern struct mm_struct * mm_alloc(void);
2151 /* mmdrop drops the mm and the page tables */
2152 extern void __mmdrop(struct mm_struct *);
2153 static inline void mmdrop(struct mm_struct * mm)
2155 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2159 /* mmput gets rid of the mappings and all user-space */
2160 extern void mmput(struct mm_struct *);
2161 /* Grab a reference to a task's mm, if it is not already going away */
2162 extern struct mm_struct *get_task_mm(struct task_struct *task);
2163 /* Remove the current tasks stale references to the old mm_struct */
2164 extern void mm_release(struct task_struct *, struct mm_struct *);
2165 /* Allocate a new mm structure and copy contents from tsk->mm */
2166 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2168 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2169 struct task_struct *, struct pt_regs *);
2170 extern void flush_thread(void);
2171 extern void exit_thread(void);
2173 extern void exit_files(struct task_struct *);
2174 extern void __cleanup_sighand(struct sighand_struct *);
2176 extern void exit_itimers(struct signal_struct *);
2177 extern void flush_itimer_signals(void);
2179 extern NORET_TYPE void do_group_exit(int);
2181 extern void daemonize(const char *, ...);
2182 extern int allow_signal(int);
2183 extern int disallow_signal(int);
2185 extern int do_execve(const char *,
2186 const char __user * const __user *,
2187 const char __user * const __user *, struct pt_regs *);
2188 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2189 struct task_struct *fork_idle(int);
2191 extern void set_task_comm(struct task_struct *tsk, char *from);
2192 extern char *get_task_comm(char *to, struct task_struct *tsk);
2195 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2197 static inline unsigned long wait_task_inactive(struct task_struct *p,
2204 #define next_task(p) \
2205 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2207 #define for_each_process(p) \
2208 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2210 extern bool current_is_single_threaded(void);
2213 * Careful: do_each_thread/while_each_thread is a double loop so
2214 * 'break' will not work as expected - use goto instead.
2216 #define do_each_thread(g, t) \
2217 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2219 #define while_each_thread(g, t) \
2220 while ((t = next_thread(t)) != g)
2222 static inline int get_nr_threads(struct task_struct *tsk)
2224 return tsk->signal->nr_threads;
2227 /* de_thread depends on thread_group_leader not being a pid based check */
2228 #define thread_group_leader(p) (p == p->group_leader)
2230 /* Do to the insanities of de_thread it is possible for a process
2231 * to have the pid of the thread group leader without actually being
2232 * the thread group leader. For iteration through the pids in proc
2233 * all we care about is that we have a task with the appropriate
2234 * pid, we don't actually care if we have the right task.
2236 static inline int has_group_leader_pid(struct task_struct *p)
2238 return p->pid == p->tgid;
2242 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2244 return p1->tgid == p2->tgid;
2247 static inline struct task_struct *next_thread(const struct task_struct *p)
2249 return list_entry_rcu(p->thread_group.next,
2250 struct task_struct, thread_group);
2253 static inline int thread_group_empty(struct task_struct *p)
2255 return list_empty(&p->thread_group);
2258 #define delay_group_leader(p) \
2259 (thread_group_leader(p) && !thread_group_empty(p))
2261 static inline int task_detached(struct task_struct *p)
2263 return p->exit_signal == -1;
2267 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2268 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2269 * pins the final release of task.io_context. Also protects ->cpuset and
2270 * ->cgroup.subsys[].
2272 * Nests both inside and outside of read_lock(&tasklist_lock).
2273 * It must not be nested with write_lock_irq(&tasklist_lock),
2274 * neither inside nor outside.
2276 static inline void task_lock(struct task_struct *p)
2278 spin_lock(&p->alloc_lock);
2281 static inline void task_unlock(struct task_struct *p)
2283 spin_unlock(&p->alloc_lock);
2286 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2287 unsigned long *flags);
2289 #define lock_task_sighand(tsk, flags) \
2290 ({ struct sighand_struct *__ss; \
2291 __cond_lock(&(tsk)->sighand->siglock, \
2292 (__ss = __lock_task_sighand(tsk, flags))); \
2296 static inline void unlock_task_sighand(struct task_struct *tsk,
2297 unsigned long *flags)
2299 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2302 #ifndef __HAVE_THREAD_FUNCTIONS
2304 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2305 #define task_stack_page(task) ((task)->stack)
2307 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2309 *task_thread_info(p) = *task_thread_info(org);
2310 task_thread_info(p)->task = p;
2313 static inline unsigned long *end_of_stack(struct task_struct *p)
2315 return (unsigned long *)(task_thread_info(p) + 1);
2320 static inline int object_is_on_stack(void *obj)
2322 void *stack = task_stack_page(current);
2324 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2327 extern void thread_info_cache_init(void);
2329 #ifdef CONFIG_DEBUG_STACK_USAGE
2330 static inline unsigned long stack_not_used(struct task_struct *p)
2332 unsigned long *n = end_of_stack(p);
2334 do { /* Skip over canary */
2338 return (unsigned long)n - (unsigned long)end_of_stack(p);
2342 /* set thread flags in other task's structures
2343 * - see asm/thread_info.h for TIF_xxxx flags available
2345 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2347 set_ti_thread_flag(task_thread_info(tsk), flag);
2350 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2352 clear_ti_thread_flag(task_thread_info(tsk), flag);
2355 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2357 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2360 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2362 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2365 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2367 return test_ti_thread_flag(task_thread_info(tsk), flag);
2370 static inline void set_tsk_need_resched(struct task_struct *tsk)
2372 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2375 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2377 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2380 static inline int test_tsk_need_resched(struct task_struct *tsk)
2382 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2385 static inline int restart_syscall(void)
2387 set_tsk_thread_flag(current, TIF_SIGPENDING);
2388 return -ERESTARTNOINTR;
2391 static inline int signal_pending(struct task_struct *p)
2393 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2396 static inline int __fatal_signal_pending(struct task_struct *p)
2398 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2401 static inline int fatal_signal_pending(struct task_struct *p)
2403 return signal_pending(p) && __fatal_signal_pending(p);
2406 static inline int signal_pending_state(long state, struct task_struct *p)
2408 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2410 if (!signal_pending(p))
2413 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2416 static inline int need_resched(void)
2418 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2422 * cond_resched() and cond_resched_lock(): latency reduction via
2423 * explicit rescheduling in places that are safe. The return
2424 * value indicates whether a reschedule was done in fact.
2425 * cond_resched_lock() will drop the spinlock before scheduling,
2426 * cond_resched_softirq() will enable bhs before scheduling.
2428 extern int _cond_resched(void);
2430 #define cond_resched() ({ \
2431 __might_sleep(__FILE__, __LINE__, 0); \
2435 extern int __cond_resched_lock(spinlock_t *lock);
2437 #ifdef CONFIG_PREEMPT
2438 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2440 #define PREEMPT_LOCK_OFFSET 0
2443 #define cond_resched_lock(lock) ({ \
2444 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2445 __cond_resched_lock(lock); \
2448 extern int __cond_resched_softirq(void);
2450 #define cond_resched_softirq() ({ \
2451 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2452 __cond_resched_softirq(); \
2456 * Does a critical section need to be broken due to another
2457 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2458 * but a general need for low latency)
2460 static inline int spin_needbreak(spinlock_t *lock)
2462 #ifdef CONFIG_PREEMPT
2463 return spin_is_contended(lock);
2470 * Thread group CPU time accounting.
2472 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2473 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2475 static inline void thread_group_cputime_init(struct signal_struct *sig)
2477 spin_lock_init(&sig->cputimer.lock);
2481 * Reevaluate whether the task has signals pending delivery.
2482 * Wake the task if so.
2483 * This is required every time the blocked sigset_t changes.
2484 * callers must hold sighand->siglock.
2486 extern void recalc_sigpending_and_wake(struct task_struct *t);
2487 extern void recalc_sigpending(void);
2489 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2492 * Wrappers for p->thread_info->cpu access. No-op on UP.
2496 static inline unsigned int task_cpu(const struct task_struct *p)
2498 return task_thread_info(p)->cpu;
2501 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2505 static inline unsigned int task_cpu(const struct task_struct *p)
2510 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2514 #endif /* CONFIG_SMP */
2516 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2517 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2519 extern void normalize_rt_tasks(void);
2521 #ifdef CONFIG_CGROUP_SCHED
2523 extern struct task_group root_task_group;
2525 extern struct task_group *sched_create_group(struct task_group *parent);
2526 extern void sched_destroy_group(struct task_group *tg);
2527 extern void sched_move_task(struct task_struct *tsk);
2528 #ifdef CONFIG_FAIR_GROUP_SCHED
2529 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2530 extern unsigned long sched_group_shares(struct task_group *tg);
2532 #ifdef CONFIG_RT_GROUP_SCHED
2533 extern int sched_group_set_rt_runtime(struct task_group *tg,
2534 long rt_runtime_us);
2535 extern long sched_group_rt_runtime(struct task_group *tg);
2536 extern int sched_group_set_rt_period(struct task_group *tg,
2538 extern long sched_group_rt_period(struct task_group *tg);
2539 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2543 extern int task_can_switch_user(struct user_struct *up,
2544 struct task_struct *tsk);
2546 #ifdef CONFIG_TASK_XACCT
2547 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2549 tsk->ioac.rchar += amt;
2552 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2554 tsk->ioac.wchar += amt;
2557 static inline void inc_syscr(struct task_struct *tsk)
2562 static inline void inc_syscw(struct task_struct *tsk)
2567 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2571 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2575 static inline void inc_syscr(struct task_struct *tsk)
2579 static inline void inc_syscw(struct task_struct *tsk)
2584 #ifndef TASK_SIZE_OF
2585 #define TASK_SIZE_OF(tsk) TASK_SIZE
2589 * Call the function if the target task is executing on a CPU right now:
2591 extern void task_oncpu_function_call(struct task_struct *p,
2592 void (*func) (void *info), void *info);
2595 #ifdef CONFIG_MM_OWNER
2596 extern void mm_update_next_owner(struct mm_struct *mm);
2597 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2599 static inline void mm_update_next_owner(struct mm_struct *mm)
2603 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2606 #endif /* CONFIG_MM_OWNER */
2608 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2611 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2614 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2617 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2620 static inline unsigned long rlimit(unsigned int limit)
2622 return task_rlimit(current, limit);
2625 static inline unsigned long rlimit_max(unsigned int limit)
2627 return task_rlimit_max(current, limit);
2630 #endif /* __KERNEL__ */