Bug fixes
[cbs-scheduler.git] / include / linux / sched.h
blob24af3cf503bbffc619bfabec6fde7039af96446a
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
4 /*
5 * cloning flags:
6 */
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 */
33 * Scheduling policies
35 #define SCHED_NORMAL 0
36 #define SCHED_FIFO 1
37 #define SCHED_RR 2
38 #define SCHED_BATCH 3
39 /* SCHED_ISO: reserved but not implemented yet */
40 #define SCHED_IDLE 5
41 #define SCHED_CBS 6
43 #ifdef __KERNEL__
46 #include <asm/param.h> /* for HZ */
48 #include <linux/capability.h>
49 #include <linux/threads.h>
50 #include <linux/kernel.h>
51 #include <linux/types.h>
52 #include <linux/timex.h>
53 #include <linux/jiffies.h>
54 #include <linux/rbtree.h>
55 #include <linux/thread_info.h>
56 #include <linux/cpumask.h>
57 #include <linux/errno.h>
58 #include <linux/nodemask.h>
59 #include <linux/mm_types.h>
61 #include <asm/system.h>
62 #include <asm/page.h>
63 #include <asm/ptrace.h>
64 #include <asm/cputime.h>
66 #include <linux/smp.h>
67 #include <linux/sem.h>
68 #include <linux/signal.h>
69 #include <linux/fs_struct.h>
70 #include <linux/compiler.h>
71 #include <linux/completion.h>
72 #include <linux/perf_counter.h>
73 #include <linux/pid.h>
74 #include <linux/percpu.h>
75 #include <linux/topology.h>
76 #include <linux/proportions.h>
77 #include <linux/seccomp.h>
78 #include <linux/rcupdate.h>
79 #include <linux/rtmutex.h>
81 #include <linux/time.h>
82 #include <linux/param.h>
83 #include <linux/resource.h>
84 #include <linux/timer.h>
85 #include <linux/hrtimer.h>
86 #include <linux/task_io_accounting.h>
87 #include <linux/kobject.h>
88 #include <linux/latencytop.h>
89 #include <linux/cred.h>
91 #include <asm/processor.h>
93 struct sched_param {
94 int sched_priority;
95 int sched_ss_low_priority;
96 struct timespec sched_ss_repl_period;
97 struct timespec sched_ss_init_budget;
98 int sched_ss_max_repl;
102 #ifdef CONFIG_PREEMPT
103 extern int kernel_preemption;
104 #else
105 # define kernel_preemption 0
106 #endif
107 #ifdef CONFIG_PREEMPT_VOLUNTARY
108 extern int voluntary_preemption;
109 #else
110 # define voluntary_preemption 0
111 #endif
112 #ifdef CONFIG_PREEMPT_SOFTIRQS
113 extern int softirq_preemption;
114 #else
115 # define softirq_preemption 0
116 #endif
118 #ifdef CONFIG_PREEMPT_HARDIRQS
119 extern int hardirq_preemption;
120 #else
121 # define hardirq_preemption 0
122 #endif
124 struct mem_cgroup;
125 struct exec_domain;
126 struct futex_pi_state;
127 struct robust_list_head;
128 struct bio;
129 struct bts_tracer;
132 * List of flags we want to share for kernel threads,
133 * if only because they are not used by them anyway.
135 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
138 * These are the constant used to fake the fixed-point load-average
139 * counting. Some notes:
140 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
141 * a load-average precision of 10 bits integer + 11 bits fractional
142 * - if you want to count load-averages more often, you need more
143 * precision, or rounding will get you. With 2-second counting freq,
144 * the EXP_n values would be 1981, 2034 and 2043 if still using only
145 * 11 bit fractions.
147 extern unsigned long avenrun[]; /* Load averages */
148 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
150 #define FSHIFT 11 /* nr of bits of precision */
151 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
152 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
153 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
154 #define EXP_5 2014 /* 1/exp(5sec/5min) */
155 #define EXP_15 2037 /* 1/exp(5sec/15min) */
157 #define CALC_LOAD(load,exp,n) \
158 load *= exp; \
159 load += n*(FIXED_1-exp); \
160 load >>= FSHIFT;
162 extern unsigned long total_forks;
163 extern int nr_threads;
164 DECLARE_PER_CPU(unsigned long, process_counts);
165 extern int nr_processes(void);
166 extern unsigned long nr_running(void);
167 extern unsigned long nr_uninterruptible(void);
168 extern unsigned long nr_iowait(void);
169 extern u64 cpu_nr_migrations(int cpu);
170 extern void calc_global_load(void);
172 extern unsigned long get_parent_ip(unsigned long addr);
174 struct seq_file;
175 struct cfs_rq;
176 struct task_group;
177 #ifdef CONFIG_SCHED_DEBUG
178 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
179 extern void proc_sched_set_task(struct task_struct *p);
180 extern void
181 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
182 #else
183 static inline void
184 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
187 static inline void proc_sched_set_task(struct task_struct *p)
190 static inline void
191 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
194 #endif
196 extern unsigned long long time_sync_thresh;
197 extern struct semaphore kernel_sem;
200 * Task state bitmask. NOTE! These bits are also
201 * encoded in fs/proc/array.c: get_task_state().
203 * We have two separate sets of flags: task->state
204 * is about runnability, while task->exit_state are
205 * about the task exiting. Confusing, but this way
206 * modifying one set can't modify the other one by
207 * mistake.
209 #define TASK_RUNNING 0
210 #define TASK_RUNNING_MUTEX 1
211 #define TASK_INTERRUPTIBLE 2
212 #define TASK_UNINTERRUPTIBLE 4
213 #define __TASK_STOPPED 8
214 #define __TASK_TRACED 16
215 /* in tsk->exit_state */
216 #define EXIT_ZOMBIE 32
217 #define EXIT_DEAD 64
218 /* in tsk->state again */
219 #define TASK_DEAD 128
220 #define TASK_WAKEKILL 256
222 /* Convenience macros for the sake of set_task_state */
223 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
224 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
225 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
227 /* Convenience macros for the sake of wake_up */
228 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
229 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
231 /* get_task_state() */
232 #define TASK_REPORT (TASK_RUNNING | TASK_RUNNING_MUTEX | \
233 TASK_INTERRUPTIBLE | \
234 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
235 __TASK_TRACED)
237 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
238 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
239 #define task_is_stopped_or_traced(task) \
240 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
241 #define task_contributes_to_load(task) \
242 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
243 (task->flags & PF_FROZEN) == 0)
245 #define __set_task_state(tsk, state_value) \
246 do { (tsk)->state = (state_value); } while (0)
247 #define set_task_state(tsk, state_value) \
248 set_mb((tsk)->state, (state_value))
250 // #define PREEMPT_DIRECT
252 #ifdef CONFIG_X86_LOCAL_APIC
253 extern void nmi_show_all_regs(void);
254 #else
255 # define nmi_show_all_regs() do { } while (0)
256 #endif
258 #include <linux/smp.h>
259 #include <linux/sem.h>
260 #include <linux/signal.h>
261 #include <linux/securebits.h>
262 #include <linux/fs_struct.h>
263 #include <linux/compiler.h>
264 #include <linux/completion.h>
265 #include <linux/pid.h>
266 #include <linux/percpu.h>
267 #include <linux/topology.h>
268 #include <linux/seccomp.h>
270 struct exec_domain;
273 * set_current_state() includes a barrier so that the write of current->state
274 * is correctly serialised wrt the caller's subsequent test of whether to
275 * actually sleep:
277 * set_current_state(TASK_UNINTERRUPTIBLE);
278 * if (do_i_need_to_sleep())
279 * schedule();
281 * If the caller does not need such serialisation then use __set_current_state()
283 #define __set_current_state(state_value) \
284 do { current->state = (state_value); } while (0)
285 #define set_current_state(state_value) \
286 set_mb(current->state, (state_value))
288 /* Task command name length */
289 #define TASK_COMM_LEN 16
291 #include <linux/spinlock.h>
294 * This serializes "schedule()" and also protects
295 * the run-queue from deletions/modifications (but
296 * _adding_ to the beginning of the run-queue has
297 * a separate lock).
299 extern rwlock_t tasklist_lock;
300 extern spinlock_t mmlist_lock;
302 struct task_struct;
304 extern void sched_init(void);
305 extern void sched_init_smp(void);
306 extern asmlinkage void schedule_tail(struct task_struct *prev);
307 extern void init_idle(struct task_struct *idle, int cpu);
308 extern void init_idle_bootup_task(struct task_struct *idle);
310 extern int runqueue_is_locked(void);
311 extern void task_rq_unlock_wait(struct task_struct *p);
313 extern cpumask_var_t nohz_cpu_mask;
314 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
315 extern int select_nohz_load_balancer(int cpu);
316 #else
317 static inline int select_nohz_load_balancer(int cpu)
319 return 0;
321 #endif
324 * Only dump TASK_* tasks. (0 for all tasks)
326 extern void show_state_filter(unsigned long state_filter);
328 static inline void show_state(void)
330 show_state_filter(0);
333 extern void show_regs(struct pt_regs *);
336 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
337 * task), SP is the stack pointer of the first frame that should be shown in the back
338 * trace (or NULL if the entire call-chain of the task should be shown).
340 extern void show_stack(struct task_struct *task, unsigned long *sp);
342 void io_schedule(void);
343 long io_schedule_timeout(long timeout);
345 extern void cpu_init (void);
346 extern void trap_init(void);
347 extern void update_process_times(int user);
348 extern void scheduler_tick(void);
350 extern void sched_show_task(struct task_struct *p);
352 #ifdef CONFIG_GENERIC_HARDIRQS
353 extern int debug_direct_keyboard;
354 #else
355 # define debug_direct_keyboard 0
356 #endif
358 #ifdef CONFIG_DETECT_SOFTLOCKUP
359 extern void softlockup_tick(void);
360 extern void touch_softlockup_watchdog(void);
361 extern void touch_all_softlockup_watchdogs(void);
362 extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
363 struct file *filp, void __user *buffer,
364 size_t *lenp, loff_t *ppos);
365 extern unsigned int softlockup_panic;
366 extern int softlockup_thresh;
367 #else
368 static inline void softlockup_tick(void)
371 static inline void touch_softlockup_watchdog(void)
374 static inline void touch_all_softlockup_watchdogs(void)
377 #endif
379 #ifdef CONFIG_DETECT_HUNG_TASK
380 extern unsigned int sysctl_hung_task_panic;
381 extern unsigned long sysctl_hung_task_check_count;
382 extern unsigned long sysctl_hung_task_timeout_secs;
383 extern unsigned long sysctl_hung_task_warnings;
384 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
385 struct file *filp, void __user *buffer,
386 size_t *lenp, loff_t *ppos);
387 #endif
389 /* Attach to any functions which should be ignored in wchan output. */
390 #define __sched __attribute__((__section__(".sched.text")))
392 /* Linker adds these: start and end of __sched functions */
393 extern char __sched_text_start[], __sched_text_end[];
395 /* Is this address in the __sched functions? */
396 extern int in_sched_functions(unsigned long addr);
398 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
399 extern signed long schedule_timeout(signed long timeout);
400 extern signed long schedule_timeout_interruptible(signed long timeout);
401 extern signed long schedule_timeout_killable(signed long timeout);
402 extern signed long schedule_timeout_uninterruptible(signed long timeout);
403 asmlinkage void __schedule(void);
404 asmlinkage void schedule(void);
405 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
407 * This one can be called with interrupts disabled, only
408 * to be used by lowlevel arch code!
410 asmlinkage void __sched __schedule(void);
412 struct nsproxy;
413 struct user_namespace;
415 /* Maximum number of active map areas.. This is a random (large) number */
416 #define DEFAULT_MAX_MAP_COUNT 65536
418 extern int sysctl_max_map_count;
420 #include <linux/aio.h>
422 extern unsigned long
423 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
424 unsigned long, unsigned long);
425 extern unsigned long
426 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
427 unsigned long len, unsigned long pgoff,
428 unsigned long flags);
429 extern void arch_unmap_area(struct mm_struct *, unsigned long);
430 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
432 #if USE_SPLIT_PTLOCKS
434 * The mm counters are not protected by its page_table_lock,
435 * so must be incremented atomically.
437 #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
438 #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
439 #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
440 #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
441 #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
443 #else /* !USE_SPLIT_PTLOCKS */
445 * The mm counters are protected by its page_table_lock,
446 * so can be incremented directly.
448 #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
449 #define get_mm_counter(mm, member) ((mm)->_##member)
450 #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
451 #define inc_mm_counter(mm, member) (mm)->_##member++
452 #define dec_mm_counter(mm, member) (mm)->_##member--
454 #endif /* !USE_SPLIT_PTLOCKS */
456 #define get_mm_rss(mm) \
457 (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
458 #define update_hiwater_rss(mm) do { \
459 unsigned long _rss = get_mm_rss(mm); \
460 if ((mm)->hiwater_rss < _rss) \
461 (mm)->hiwater_rss = _rss; \
462 } while (0)
463 #define update_hiwater_vm(mm) do { \
464 if ((mm)->hiwater_vm < (mm)->total_vm) \
465 (mm)->hiwater_vm = (mm)->total_vm; \
466 } while (0)
468 #define get_mm_hiwater_rss(mm) max((mm)->hiwater_rss, get_mm_rss(mm))
469 #define get_mm_hiwater_vm(mm) max((mm)->hiwater_vm, (mm)->total_vm)
471 extern void set_dumpable(struct mm_struct *mm, int value);
472 extern int get_dumpable(struct mm_struct *mm);
474 /* mm flags */
475 /* dumpable bits */
476 #define MMF_DUMPABLE 0 /* core dump is permitted */
477 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
478 #define MMF_DUMPABLE_BITS 2
480 /* coredump filter bits */
481 #define MMF_DUMP_ANON_PRIVATE 2
482 #define MMF_DUMP_ANON_SHARED 3
483 #define MMF_DUMP_MAPPED_PRIVATE 4
484 #define MMF_DUMP_MAPPED_SHARED 5
485 #define MMF_DUMP_ELF_HEADERS 6
486 #define MMF_DUMP_HUGETLB_PRIVATE 7
487 #define MMF_DUMP_HUGETLB_SHARED 8
488 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
489 #define MMF_DUMP_FILTER_BITS 7
490 #define MMF_DUMP_FILTER_MASK \
491 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
492 #define MMF_DUMP_FILTER_DEFAULT \
493 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
494 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
496 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
497 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
498 #else
499 # define MMF_DUMP_MASK_DEFAULT_ELF 0
500 #endif
502 struct sighand_struct {
503 atomic_t count;
504 struct k_sigaction action[_NSIG];
505 spinlock_t siglock;
506 wait_queue_head_t signalfd_wqh;
509 struct pacct_struct {
510 int ac_flag;
511 long ac_exitcode;
512 unsigned long ac_mem;
513 cputime_t ac_utime, ac_stime;
514 unsigned long ac_minflt, ac_majflt;
518 * struct task_cputime - collected CPU time counts
519 * @utime: time spent in user mode, in &cputime_t units
520 * @stime: time spent in kernel mode, in &cputime_t units
521 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
523 * This structure groups together three kinds of CPU time that are
524 * tracked for threads and thread groups. Most things considering
525 * CPU time want to group these counts together and treat all three
526 * of them in parallel.
528 struct task_cputime {
529 cputime_t utime;
530 cputime_t stime;
531 unsigned long long sum_exec_runtime;
533 /* Alternate field names when used to cache expirations. */
534 #define prof_exp stime
535 #define virt_exp utime
536 #define sched_exp sum_exec_runtime
538 #define INIT_CPUTIME \
539 (struct task_cputime) { \
540 .utime = cputime_zero, \
541 .stime = cputime_zero, \
542 .sum_exec_runtime = 0, \
546 * struct thread_group_cputimer - thread group interval timer counts
547 * @cputime: thread group interval timers.
548 * @running: non-zero when there are timers running and
549 * @cputime receives updates.
550 * @lock: lock for fields in this struct.
552 * This structure contains the version of task_cputime, above, that is
553 * used for thread group CPU timer calculations.
555 struct thread_group_cputimer {
556 struct task_cputime cputime;
557 int running;
558 raw_spinlock_t lock;
562 * NOTE! "signal_struct" does not have it's own
563 * locking, because a shared signal_struct always
564 * implies a shared sighand_struct, so locking
565 * sighand_struct is always a proper superset of
566 * the locking of signal_struct.
568 struct signal_struct {
569 atomic_t count;
570 atomic_t live;
572 wait_queue_head_t wait_chldexit; /* for wait4() */
574 /* current thread group signal load-balancing target: */
575 struct task_struct *curr_target;
577 /* shared signal handling: */
578 struct sigpending shared_pending;
580 /* thread group exit support */
581 int group_exit_code;
582 /* overloaded:
583 * - notify group_exit_task when ->count is equal to notify_count
584 * - everyone except group_exit_task is stopped during signal delivery
585 * of fatal signals, group_exit_task processes the signal.
587 int notify_count;
588 struct task_struct *group_exit_task;
590 /* thread group stop support, overloads group_exit_code too */
591 int group_stop_count;
592 unsigned int flags; /* see SIGNAL_* flags below */
594 /* POSIX.1b Interval Timers */
595 struct list_head posix_timers;
597 /* ITIMER_REAL timer for the process */
598 struct hrtimer real_timer;
599 struct pid *leader_pid;
600 ktime_t it_real_incr;
602 /* ITIMER_PROF and ITIMER_VIRTUAL timers for the process */
603 cputime_t it_prof_expires, it_virt_expires;
604 cputime_t it_prof_incr, it_virt_incr;
607 * Thread group totals for process CPU timers.
608 * See thread_group_cputimer(), et al, for details.
610 struct thread_group_cputimer cputimer;
612 /* Earliest-expiration cache. */
613 struct task_cputime cputime_expires;
615 struct list_head cpu_timers[3];
617 /* job control IDs */
620 * pgrp and session fields are deprecated.
621 * use the task_session_Xnr and task_pgrp_Xnr routines below
624 union {
625 pid_t pgrp __deprecated;
626 pid_t __pgrp;
629 struct pid *tty_old_pgrp;
631 union {
632 pid_t session __deprecated;
633 pid_t __session;
636 /* boolean value for session group leader */
637 int leader;
639 struct tty_struct *tty; /* NULL if no tty */
642 * Cumulative resource counters for dead threads in the group,
643 * and for reaped dead child processes forked by this group.
644 * Live threads maintain their own counters and add to these
645 * in __exit_signal, except for the group leader.
647 cputime_t utime, stime, cutime, cstime;
648 cputime_t gtime;
649 cputime_t cgtime;
650 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
651 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
652 unsigned long inblock, oublock, cinblock, coublock;
653 struct task_io_accounting ioac;
656 * Cumulative ns of schedule CPU time fo dead threads in the
657 * group, not including a zombie group leader, (This only differs
658 * from jiffies_to_ns(utime + stime) if sched_clock uses something
659 * other than jiffies.)
661 unsigned long long sum_sched_runtime;
664 * We don't bother to synchronize most readers of this at all,
665 * because there is no reader checking a limit that actually needs
666 * to get both rlim_cur and rlim_max atomically, and either one
667 * alone is a single word that can safely be read normally.
668 * getrlimit/setrlimit use task_lock(current->group_leader) to
669 * protect this instead of the siglock, because they really
670 * have no need to disable irqs.
672 struct rlimit rlim[RLIM_NLIMITS];
674 #ifdef CONFIG_BSD_PROCESS_ACCT
675 struct pacct_struct pacct; /* per-process accounting information */
676 #endif
677 #ifdef CONFIG_TASKSTATS
678 struct taskstats *stats;
679 #endif
680 #ifdef CONFIG_AUDIT
681 unsigned audit_tty;
682 struct tty_audit_buf *tty_audit_buf;
683 #endif
686 /* Context switch must be unlocked if interrupts are to be enabled */
687 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
688 # define __ARCH_WANT_UNLOCKED_CTXSW
689 #endif
692 * Bits in flags field of signal_struct.
694 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
695 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
696 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
697 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
699 * Pending notifications to parent.
701 #define SIGNAL_CLD_STOPPED 0x00000010
702 #define SIGNAL_CLD_CONTINUED 0x00000020
703 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
705 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
707 /* If true, all threads except ->group_exit_task have pending SIGKILL */
708 static inline int signal_group_exit(const struct signal_struct *sig)
710 return (sig->flags & SIGNAL_GROUP_EXIT) ||
711 (sig->group_exit_task != NULL);
715 * Some day this will be a full-fledged user tracking system..
717 struct user_struct {
718 atomic_t __count; /* reference count */
719 atomic_t processes; /* How many processes does this user have? */
720 atomic_t files; /* How many open files does this user have? */
721 atomic_t sigpending; /* How many pending signals does this user have? */
722 #ifdef CONFIG_INOTIFY_USER
723 atomic_t inotify_watches; /* How many inotify watches does this user have? */
724 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
725 #endif
726 #ifdef CONFIG_EPOLL
727 atomic_t epoll_watches; /* The number of file descriptors currently watched */
728 #endif
729 #ifdef CONFIG_POSIX_MQUEUE
730 /* protected by mq_lock */
731 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
732 #endif
733 unsigned long locked_shm; /* How many pages of mlocked shm ? */
735 #ifdef CONFIG_KEYS
736 struct key *uid_keyring; /* UID specific keyring */
737 struct key *session_keyring; /* UID's default session keyring */
738 #endif
740 /* Hash table maintenance information */
741 struct hlist_node uidhash_node;
742 uid_t uid;
743 struct user_namespace *user_ns;
745 #ifdef CONFIG_USER_SCHED
746 struct task_group *tg;
747 #ifdef CONFIG_SYSFS
748 struct kobject kobj;
749 struct work_struct work;
750 #endif
751 #endif
754 extern int uids_sysfs_init(void);
756 extern struct user_struct *find_user(uid_t);
758 extern struct user_struct root_user;
759 #define INIT_USER (&root_user)
762 struct backing_dev_info;
763 struct reclaim_state;
765 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
766 struct sched_info {
767 /* cumulative counters */
768 unsigned long pcount; /* # of times run on this cpu */
769 unsigned long long run_delay; /* time spent waiting on a runqueue */
771 /* timestamps */
772 unsigned long long last_arrival,/* when we last ran on a cpu */
773 last_queued; /* when we were last queued to run */
774 #ifdef CONFIG_SCHEDSTATS
775 /* BKL stats */
776 unsigned int bkl_count;
777 #endif
779 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
781 #ifdef CONFIG_TASK_DELAY_ACCT
782 struct task_delay_info {
783 spinlock_t lock;
784 unsigned int flags; /* Private per-task flags */
786 /* For each stat XXX, add following, aligned appropriately
788 * struct timespec XXX_start, XXX_end;
789 * u64 XXX_delay;
790 * u32 XXX_count;
792 * Atomicity of updates to XXX_delay, XXX_count protected by
793 * single lock above (split into XXX_lock if contention is an issue).
797 * XXX_count is incremented on every XXX operation, the delay
798 * associated with the operation is added to XXX_delay.
799 * XXX_delay contains the accumulated delay time in nanoseconds.
801 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
802 u64 blkio_delay; /* wait for sync block io completion */
803 u64 swapin_delay; /* wait for swapin block io completion */
804 u32 blkio_count; /* total count of the number of sync block */
805 /* io operations performed */
806 u32 swapin_count; /* total count of the number of swapin block */
807 /* io operations performed */
809 struct timespec freepages_start, freepages_end;
810 u64 freepages_delay; /* wait for memory reclaim */
811 u32 freepages_count; /* total count of memory reclaim */
813 #endif /* CONFIG_TASK_DELAY_ACCT */
815 static inline int sched_info_on(void)
817 #ifdef CONFIG_SCHEDSTATS
818 return 1;
819 #elif defined(CONFIG_TASK_DELAY_ACCT)
820 extern int delayacct_on;
821 return delayacct_on;
822 #else
823 return 0;
824 #endif
827 enum cpu_idle_type {
828 CPU_IDLE,
829 CPU_NOT_IDLE,
830 CPU_NEWLY_IDLE,
831 CPU_MAX_IDLE_TYPES
835 * sched-domains (multiprocessor balancing) declarations:
839 * Increase resolution of nice-level calculations:
841 #define SCHED_LOAD_SHIFT 10
842 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
844 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
846 #ifdef CONFIG_SMP
847 #define SD_LOAD_BALANCE 1 /* Do load balancing on this domain. */
848 #define SD_BALANCE_NEWIDLE 2 /* Balance when about to become idle */
849 #define SD_BALANCE_EXEC 4 /* Balance on exec */
850 #define SD_BALANCE_FORK 8 /* Balance on fork, clone */
851 #define SD_WAKE_IDLE 16 /* Wake to idle CPU on task wakeup */
852 #define SD_WAKE_AFFINE 32 /* Wake task to waking CPU */
853 #define SD_WAKE_BALANCE 64 /* Perform balancing at task wakeup */
854 #define SD_SHARE_CPUPOWER 128 /* Domain members share cpu power */
855 #define SD_POWERSAVINGS_BALANCE 256 /* Balance for power savings */
856 #define SD_SHARE_PKG_RESOURCES 512 /* Domain members share cpu pkg resources */
857 #define SD_SERIALIZE 1024 /* Only a single load balancing instance */
858 #define SD_WAKE_IDLE_FAR 2048 /* Gain latency sacrificing cache hit */
860 enum powersavings_balance_level {
861 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
862 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
863 * first for long running threads
865 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
866 * cpu package for power savings
868 MAX_POWERSAVINGS_BALANCE_LEVELS
871 extern int sched_mc_power_savings, sched_smt_power_savings;
873 static inline int sd_balance_for_mc_power(void)
875 if (sched_smt_power_savings)
876 return SD_POWERSAVINGS_BALANCE;
878 return 0;
881 static inline int sd_balance_for_package_power(void)
883 if (sched_mc_power_savings | sched_smt_power_savings)
884 return SD_POWERSAVINGS_BALANCE;
886 return 0;
890 * Optimise SD flags for power savings:
891 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
892 * Keep default SD flags if sched_{smt,mc}_power_saving=0
895 static inline int sd_power_saving_flags(void)
897 if (sched_mc_power_savings | sched_smt_power_savings)
898 return SD_BALANCE_NEWIDLE;
900 return 0;
903 struct sched_group {
904 struct sched_group *next; /* Must be a circular list */
907 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
908 * single CPU. This is read only (except for setup, hotplug CPU).
909 * Note : Never change cpu_power without recompute its reciprocal
911 unsigned int __cpu_power;
913 * reciprocal value of cpu_power to avoid expensive divides
914 * (see include/linux/reciprocal_div.h)
916 u32 reciprocal_cpu_power;
918 unsigned long cpumask[];
921 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
923 return to_cpumask(sg->cpumask);
926 enum sched_domain_level {
927 SD_LV_NONE = 0,
928 SD_LV_SIBLING,
929 SD_LV_MC,
930 SD_LV_CPU,
931 SD_LV_NODE,
932 SD_LV_ALLNODES,
933 SD_LV_MAX
936 struct sched_domain_attr {
937 int relax_domain_level;
940 #define SD_ATTR_INIT (struct sched_domain_attr) { \
941 .relax_domain_level = -1, \
944 struct sched_domain {
945 /* These fields must be setup */
946 struct sched_domain *parent; /* top domain must be null terminated */
947 struct sched_domain *child; /* bottom domain must be null terminated */
948 struct sched_group *groups; /* the balancing groups of the domain */
949 unsigned long min_interval; /* Minimum balance interval ms */
950 unsigned long max_interval; /* Maximum balance interval ms */
951 unsigned int busy_factor; /* less balancing by factor if busy */
952 unsigned int imbalance_pct; /* No balance until over watermark */
953 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
954 unsigned int busy_idx;
955 unsigned int idle_idx;
956 unsigned int newidle_idx;
957 unsigned int wake_idx;
958 unsigned int forkexec_idx;
959 int flags; /* See SD_* */
960 enum sched_domain_level level;
962 /* Runtime fields. */
963 unsigned long last_balance; /* init to jiffies. units in jiffies */
964 unsigned int balance_interval; /* initialise to 1. units in ms. */
965 unsigned int nr_balance_failed; /* initialise to 0 */
967 u64 last_update;
969 #ifdef CONFIG_SCHEDSTATS
970 /* load_balance() stats */
971 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
972 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
973 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
974 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
975 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
976 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
977 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
978 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
980 /* Active load balancing */
981 unsigned int alb_count;
982 unsigned int alb_failed;
983 unsigned int alb_pushed;
985 /* SD_BALANCE_EXEC stats */
986 unsigned int sbe_count;
987 unsigned int sbe_balanced;
988 unsigned int sbe_pushed;
990 /* SD_BALANCE_FORK stats */
991 unsigned int sbf_count;
992 unsigned int sbf_balanced;
993 unsigned int sbf_pushed;
995 /* try_to_wake_up() stats */
996 unsigned int ttwu_wake_remote;
997 unsigned int ttwu_move_affine;
998 unsigned int ttwu_move_balance;
999 #endif
1000 #ifdef CONFIG_SCHED_DEBUG
1001 char *name;
1002 #endif
1004 /* span of all CPUs in this domain */
1005 unsigned long span[];
1008 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1010 return to_cpumask(sd->span);
1013 extern void partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
1014 struct sched_domain_attr *dattr_new);
1016 /* Test a flag in parent sched domain */
1017 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1019 if (sd->parent && (sd->parent->flags & flag))
1020 return 1;
1022 return 0;
1025 #else /* CONFIG_SMP */
1027 struct sched_domain_attr;
1029 static inline void
1030 partition_sched_domains(int ndoms_new, struct cpumask *doms_new,
1031 struct sched_domain_attr *dattr_new)
1034 #endif /* !CONFIG_SMP */
1036 struct io_context; /* See blkdev.h */
1039 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1040 extern void prefetch_stack(struct task_struct *t);
1041 #else
1042 static inline void prefetch_stack(struct task_struct *t) { }
1043 #endif
1045 struct audit_context; /* See audit.c */
1046 struct mempolicy;
1047 struct pipe_inode_info;
1048 struct uts_namespace;
1050 struct rq;
1051 struct sched_domain;
1053 struct sched_class {
1054 const struct sched_class *next;
1056 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1057 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1058 void (*yield_task) (struct rq *rq);
1060 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int sync);
1062 struct task_struct * (*pick_next_task) (struct rq *rq);
1063 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1065 #ifdef CONFIG_SMP
1066 int (*select_task_rq)(struct task_struct *p, int sync);
1068 unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1069 struct rq *busiest, unsigned long max_load_move,
1070 struct sched_domain *sd, enum cpu_idle_type idle,
1071 int *all_pinned, int *this_best_prio);
1073 int (*move_one_task) (struct rq *this_rq, int this_cpu,
1074 struct rq *busiest, struct sched_domain *sd,
1075 enum cpu_idle_type idle);
1076 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1077 int (*needs_post_schedule) (struct rq *this_rq);
1078 void (*post_schedule) (struct rq *this_rq);
1079 void (*task_wake_up) (struct rq *this_rq, struct task_struct *task);
1081 void (*set_cpus_allowed)(struct task_struct *p,
1082 const struct cpumask *newmask);
1084 void (*rq_online)(struct rq *rq);
1085 void (*rq_offline)(struct rq *rq);
1086 #endif
1088 void (*set_curr_task) (struct rq *rq);
1089 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1090 void (*task_new) (struct rq *rq, struct task_struct *p);
1092 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1093 int running);
1094 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1095 int running);
1096 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1097 int oldprio, int running);
1099 #ifdef CONFIG_FAIR_GROUP_SCHED
1100 void (*moved_group) (struct task_struct *p);
1101 #endif
1104 struct load_weight {
1105 unsigned long weight, inv_weight;
1109 * CFS stats for a schedulable entity (task, task-group etc)
1111 * Current field usage histogram:
1113 * 4 se->block_start
1114 * 4 se->run_node
1115 * 4 se->sleep_start
1116 * 6 se->load.weight
1118 struct sched_entity {
1119 struct load_weight load; /* for load-balancing */
1120 struct rb_node run_node;
1121 struct list_head group_node;
1122 unsigned int on_rq;
1124 u64 exec_start;
1125 u64 sum_exec_runtime;
1126 u64 vruntime;
1127 u64 prev_sum_exec_runtime;
1129 u64 last_wakeup;
1130 u64 avg_overlap;
1132 u64 nr_migrations;
1134 u64 start_runtime;
1135 u64 avg_wakeup;
1137 #ifdef CONFIG_SCHEDSTATS
1138 u64 wait_start;
1139 u64 wait_max;
1140 u64 wait_count;
1141 u64 wait_sum;
1143 u64 sleep_start;
1144 u64 sleep_max;
1145 s64 sum_sleep_runtime;
1147 u64 block_start;
1148 u64 block_max;
1149 u64 exec_max;
1150 u64 slice_max;
1152 u64 nr_migrations_cold;
1153 u64 nr_failed_migrations_affine;
1154 u64 nr_failed_migrations_running;
1155 u64 nr_failed_migrations_hot;
1156 u64 nr_forced_migrations;
1157 u64 nr_forced2_migrations;
1159 u64 nr_wakeups;
1160 u64 nr_wakeups_sync;
1161 u64 nr_wakeups_migrate;
1162 u64 nr_wakeups_local;
1163 u64 nr_wakeups_remote;
1164 u64 nr_wakeups_affine;
1165 u64 nr_wakeups_affine_attempts;
1166 u64 nr_wakeups_passive;
1167 u64 nr_wakeups_idle;
1168 #endif
1170 #ifdef CONFIG_FAIR_GROUP_SCHED
1171 struct sched_entity *parent;
1172 /* rq on which this entity is (to be) queued: */
1173 struct cfs_rq *cfs_rq;
1174 /* rq "owned" by this entity/group: */
1175 struct cfs_rq *my_q;
1176 #endif
1179 struct sched_rt_entity {
1180 struct list_head run_list;
1181 unsigned long timeout;
1182 unsigned int time_slice;
1183 int nr_cpus_allowed;
1185 struct sched_rt_entity *back;
1186 #ifdef CONFIG_RT_GROUP_SCHED
1187 struct sched_rt_entity *parent;
1188 /* rq on which this entity is (to be) queued: */
1189 struct rt_rq *rt_rq;
1190 /* rq "owned" by this entity/group: */
1191 struct rt_rq *my_q;
1192 #endif
1195 #ifdef CONFIG_SCHED_CBS
1196 struct sched_cbs_entity { /*TODO: maybe merge with sched_entity? */
1197 struct rb_node run_node;
1198 int on_rq;
1199 u64 max_budget;
1200 u64 period;
1201 s64 budget;
1202 u64 deadline;
1203 u64 exec_start;
1205 #endif
1207 struct task_struct {
1208 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1209 void *stack;
1210 atomic_t usage;
1211 unsigned int flags; /* per process flags, defined below */
1212 unsigned int ptrace;
1214 int lock_depth; /* BKL lock depth */
1216 #ifdef CONFIG_SMP
1217 int oncpu;
1218 #endif
1220 int prio, static_prio, normal_prio;
1221 unsigned int rt_priority;
1222 const struct sched_class *sched_class;
1223 struct sched_entity se;
1224 struct sched_rt_entity rt;
1226 #ifdef CONFIG_SCHED_CBS
1227 struct sched_cbs_entity cbs_se;
1228 #endif
1230 #ifdef CONFIG_PREEMPT_NOTIFIERS
1231 /* list of struct preempt_notifier: */
1232 struct hlist_head preempt_notifiers;
1233 #endif
1236 * fpu_counter contains the number of consecutive context switches
1237 * that the FPU is used. If this is over a threshold, the lazy fpu
1238 * saving becomes unlazy to save the trap. This is an unsigned char
1239 * so that after 256 times the counter wraps and the behavior turns
1240 * lazy again; this to deal with bursty apps that only use FPU for
1241 * a short time
1243 unsigned char fpu_counter;
1244 s8 oomkilladj; /* OOM kill score adjustment (bit shift). */
1245 #ifdef CONFIG_BLK_DEV_IO_TRACE
1246 unsigned int btrace_seq;
1247 #endif
1249 unsigned int policy;
1250 cpumask_t cpus_allowed;
1252 #ifdef CONFIG_PREEMPT_RCU
1253 int rcu_read_lock_nesting;
1254 int rcu_flipctr_idx;
1255 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1257 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1258 struct sched_info sched_info;
1259 #endif
1261 struct list_head tasks;
1262 struct plist_node pushable_tasks;
1264 struct mm_struct *mm, *active_mm;
1266 /* task state */
1267 struct linux_binfmt *binfmt;
1268 int exit_state;
1269 int exit_code, exit_signal;
1270 int pdeath_signal; /* The signal sent when the parent dies */
1271 /* ??? */
1272 unsigned int personality;
1273 unsigned did_exec:1;
1274 pid_t pid;
1275 pid_t tgid;
1277 /* Canary value for the -fstack-protector gcc feature */
1278 unsigned long stack_canary;
1281 * pointers to (original) parent process, youngest child, younger sibling,
1282 * older sibling, respectively. (p->father can be replaced with
1283 * p->real_parent->pid)
1285 struct task_struct *real_parent; /* real parent process */
1286 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1288 * children/sibling forms the list of my natural children
1290 struct list_head children; /* list of my children */
1291 struct list_head sibling; /* linkage in my parent's children list */
1292 struct task_struct *group_leader; /* threadgroup leader */
1295 * ptraced is the list of tasks this task is using ptrace on.
1296 * This includes both natural children and PTRACE_ATTACH targets.
1297 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1299 struct list_head ptraced;
1300 struct list_head ptrace_entry;
1302 #ifdef CONFIG_X86_PTRACE_BTS
1304 * This is the tracer handle for the ptrace BTS extension.
1305 * This field actually belongs to the ptracer task.
1307 struct bts_tracer *bts;
1309 * The buffer to hold the BTS data.
1311 void *bts_buffer;
1312 size_t bts_size;
1313 #endif /* CONFIG_X86_PTRACE_BTS */
1315 /* PID/PID hash table linkage. */
1316 struct pid_link pids[PIDTYPE_MAX];
1317 struct list_head thread_group;
1319 struct completion *vfork_done; /* for vfork() */
1320 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1321 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1323 cputime_t utime, stime, utimescaled, stimescaled;
1324 cputime_t gtime;
1325 cputime_t prev_utime, prev_stime;
1326 unsigned long nvcsw, nivcsw; /* context switch counts */
1327 struct timespec start_time; /* monotonic time */
1328 struct timespec real_start_time; /* boot based time */
1329 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1330 unsigned long min_flt, maj_flt;
1332 struct task_cputime cputime_expires;
1333 struct list_head cpu_timers[3];
1335 struct task_struct* posix_timer_list;
1337 /* process credentials */
1338 const struct cred *real_cred; /* objective and real subjective task
1339 * credentials (COW) */
1340 const struct cred *cred; /* effective (overridable) subjective task
1341 * credentials (COW) */
1342 struct mutex cred_exec_mutex; /* execve vs ptrace cred calculation mutex */
1344 char comm[TASK_COMM_LEN]; /* executable name excluding path
1345 - access with [gs]et_task_comm (which lock
1346 it with task_lock())
1347 - initialized normally by flush_old_exec */
1348 /* file system info */
1349 int link_count, total_link_count;
1350 #ifdef CONFIG_SYSVIPC
1351 /* ipc stuff */
1352 struct sysv_sem sysvsem;
1353 #endif
1354 #ifdef CONFIG_DETECT_HUNG_TASK
1355 /* hung task detection */
1356 unsigned long last_switch_count;
1357 #endif
1358 /* CPU-specific state of this task */
1359 struct thread_struct thread;
1360 /* filesystem information */
1361 struct fs_struct *fs;
1362 /* open file information */
1363 struct files_struct *files;
1364 /* namespaces */
1365 struct nsproxy *nsproxy;
1366 /* signal handlers */
1367 struct signal_struct *signal;
1368 struct sighand_struct *sighand;
1369 struct sigqueue *sigqueue_cache;
1371 sigset_t blocked, real_blocked;
1372 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1373 struct sigpending pending;
1375 unsigned long sas_ss_sp;
1376 size_t sas_ss_size;
1377 int (*notifier)(void *priv);
1378 void *notifier_data;
1379 sigset_t *notifier_mask;
1380 struct audit_context *audit_context;
1381 #ifdef CONFIG_AUDITSYSCALL
1382 uid_t loginuid;
1383 unsigned int sessionid;
1384 #endif
1385 seccomp_t seccomp;
1387 /* Thread group tracking */
1388 u32 parent_exec_id;
1389 u32 self_exec_id;
1390 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings */
1391 spinlock_t alloc_lock;
1393 /* Protection of the PI data structures: */
1394 raw_spinlock_t pi_lock;
1396 #ifdef CONFIG_RT_MUTEXES
1397 /* PI waiters blocked on a rt_mutex held by this task */
1398 struct plist_head pi_waiters;
1399 /* Deadlock detection and priority inheritance handling */
1400 struct rt_mutex_waiter *pi_blocked_on;
1401 #endif
1403 #ifdef CONFIG_DEBUG_MUTEXES
1404 /* mutex deadlock detection */
1405 struct mutex_waiter *blocked_on;
1406 #endif
1407 int pagefault_disabled;
1408 #ifdef CONFIG_TRACE_IRQFLAGS
1409 unsigned int irq_events;
1410 int hardirqs_enabled;
1411 unsigned long hardirq_enable_ip;
1412 unsigned int hardirq_enable_event;
1413 unsigned long hardirq_disable_ip;
1414 unsigned int hardirq_disable_event;
1415 int softirqs_enabled;
1416 unsigned long softirq_disable_ip;
1417 unsigned int softirq_disable_event;
1418 unsigned long softirq_enable_ip;
1419 unsigned int softirq_enable_event;
1420 int hardirq_context;
1421 int softirq_context;
1422 #endif
1423 #ifdef CONFIG_LOCKDEP
1424 # define MAX_LOCK_DEPTH 48UL
1425 u64 curr_chain_key;
1426 int lockdep_depth;
1427 unsigned int lockdep_recursion;
1428 struct held_lock held_locks[MAX_LOCK_DEPTH];
1429 gfp_t lockdep_reclaim_gfp;
1430 #endif
1432 /* realtime bits */
1434 #define MAX_PREEMPT_TRACE 25
1435 #define MAX_LOCK_STACK MAX_PREEMPT_TRACE
1436 #ifdef CONFIG_DEBUG_PREEMPT
1437 atomic_t lock_count;
1438 # ifdef CONFIG_PREEMPT_RT
1439 struct rt_mutex *owned_lock[MAX_LOCK_STACK];
1440 # endif
1441 #endif
1442 #ifdef CONFIG_DETECT_SOFTLOCKUP
1443 unsigned long softlockup_count; /* Count to keep track how long the
1444 * thread is in the kernel without
1445 * sleeping.
1447 #endif
1448 #ifdef CONFIG_DEBUG_RT_MUTEXES
1449 void *last_kernel_lock;
1450 #endif
1452 /* journalling filesystem info */
1453 void *journal_info;
1455 /* stacked block device info */
1456 struct bio *bio_list, **bio_tail;
1458 /* VM state */
1459 struct reclaim_state *reclaim_state;
1461 struct backing_dev_info *backing_dev_info;
1463 struct io_context *io_context;
1465 unsigned long ptrace_message;
1466 siginfo_t *last_siginfo; /* For ptrace use. */
1467 struct task_io_accounting ioac;
1468 #if defined(CONFIG_TASK_XACCT)
1469 u64 acct_rss_mem1; /* accumulated rss usage */
1470 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1471 cputime_t acct_timexpd; /* stime + utime since last update */
1472 #endif
1473 #ifdef CONFIG_CPUSETS
1474 nodemask_t mems_allowed;
1475 int cpuset_mems_generation;
1476 int cpuset_mem_spread_rotor;
1477 #endif
1478 #ifdef CONFIG_CGROUPS
1479 /* Control Group info protected by css_set_lock */
1480 struct css_set *cgroups;
1481 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1482 struct list_head cg_list;
1483 #endif
1484 #ifdef CONFIG_FUTEX
1485 struct robust_list_head __user *robust_list;
1486 #ifdef CONFIG_COMPAT
1487 struct compat_robust_list_head __user *compat_robust_list;
1488 #endif
1489 struct list_head pi_state_list;
1490 struct futex_pi_state *pi_state_cache;
1491 struct task_struct *futex_wakeup;
1492 #endif
1493 struct perf_counter_context perf_counter_ctx;
1494 #ifdef CONFIG_NUMA
1495 struct mempolicy *mempolicy;
1496 short il_next;
1497 #endif
1498 atomic_t fs_excl; /* holding fs exclusive resources */
1499 struct rcu_head rcu;
1502 * cache last used pipe for splice
1504 struct pipe_inode_info *splice_pipe;
1505 #ifdef CONFIG_TASK_DELAY_ACCT
1506 struct task_delay_info *delays;
1507 #endif
1508 #ifdef CONFIG_FAULT_INJECTION
1509 int make_it_fail;
1510 #endif
1511 struct prop_local_single dirties;
1512 #ifdef CONFIG_LATENCYTOP
1513 int latency_record_count;
1514 struct latency_record latency_record[LT_SAVECOUNT];
1515 #endif
1517 * time slack values; these are used to round up poll() and
1518 * select() etc timeout values. These are in nanoseconds.
1520 unsigned long timer_slack_ns;
1521 unsigned long default_timer_slack_ns;
1523 struct list_head *scm_work_list;
1524 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1525 /* Index of current stored adress in ret_stack */
1526 int curr_ret_stack;
1527 /* Stack of return addresses for return function tracing */
1528 struct ftrace_ret_stack *ret_stack;
1529 /* time stamp for last schedule */
1530 unsigned long long ftrace_timestamp;
1532 * Number of functions that haven't been traced
1533 * because of depth overrun.
1535 atomic_t trace_overrun;
1536 /* Pause for the tracing */
1537 atomic_t tracing_graph_pause;
1538 #endif
1539 #ifdef CONFIG_TRACING
1540 /* state flags for use by tracers */
1541 unsigned long trace;
1542 #endif
1543 #ifdef CONFIG_PREEMPT_RT
1545 * Temporary hack, until we find a solution to
1546 * handle printk in atomic operations.
1548 int in_printk;
1549 #endif
1552 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1553 #define tsk_cpumask(tsk) (&(tsk)->cpus_allowed)
1555 #ifdef CONFIG_PREEMPT_RT
1556 # define set_printk_might_sleep(x) do { current->in_printk = x; } while(0)
1557 #else
1558 # define set_printk_might_sleep(x) do { } while(0)
1559 #endif
1562 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1563 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1564 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1565 * values are inverted: lower p->prio value means higher priority.
1567 * The MAX_USER_RT_PRIO value allows the actual maximum
1568 * RT priority to be separate from the value exported to
1569 * user-space. This allows kernel threads to set their
1570 * priority to a value higher than any user task. Note:
1571 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1574 #define MAX_USER_RT_PRIO 100
1575 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1577 #define MAX_PRIO (MAX_RT_PRIO + 40)
1578 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1580 static inline int rt_prio(int prio)
1582 if (unlikely(prio < MAX_RT_PRIO))
1583 return 1;
1584 return 0;
1587 static inline int rt_task(struct task_struct *p)
1589 return rt_prio(p->prio);
1592 static inline void set_task_session(struct task_struct *tsk, pid_t session)
1594 tsk->signal->__session = session;
1597 static inline void set_task_pgrp(struct task_struct *tsk, pid_t pgrp)
1599 tsk->signal->__pgrp = pgrp;
1602 static inline struct pid *task_pid(struct task_struct *task)
1604 return task->pids[PIDTYPE_PID].pid;
1607 static inline struct pid *task_tgid(struct task_struct *task)
1609 return task->group_leader->pids[PIDTYPE_PID].pid;
1612 static inline struct pid *task_pgrp(struct task_struct *task)
1614 return task->group_leader->pids[PIDTYPE_PGID].pid;
1617 static inline struct pid *task_session(struct task_struct *task)
1619 return task->group_leader->pids[PIDTYPE_SID].pid;
1622 struct pid_namespace;
1625 * the helpers to get the task's different pids as they are seen
1626 * from various namespaces
1628 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1629 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1630 * current.
1631 * task_xid_nr_ns() : id seen from the ns specified;
1633 * set_task_vxid() : assigns a virtual id to a task;
1635 * see also pid_nr() etc in include/linux/pid.h
1638 static inline pid_t task_pid_nr(struct task_struct *tsk)
1640 return tsk->pid;
1643 pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1645 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1647 return pid_vnr(task_pid(tsk));
1651 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1653 return tsk->tgid;
1656 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1658 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1660 return pid_vnr(task_tgid(tsk));
1664 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1666 return tsk->signal->__pgrp;
1669 pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1671 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1673 return pid_vnr(task_pgrp(tsk));
1677 static inline pid_t task_session_nr(struct task_struct *tsk)
1679 return tsk->signal->__session;
1682 pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1684 static inline pid_t task_session_vnr(struct task_struct *tsk)
1686 return pid_vnr(task_session(tsk));
1691 * pid_alive - check that a task structure is not stale
1692 * @p: Task structure to be checked.
1694 * Test if a process is not yet dead (at most zombie state)
1695 * If pid_alive fails, then pointers within the task structure
1696 * can be stale and must not be dereferenced.
1698 static inline int pid_alive(struct task_struct *p)
1700 return p->pids[PIDTYPE_PID].pid != NULL;
1704 * is_global_init - check if a task structure is init
1705 * @tsk: Task structure to be checked.
1707 * Check if a task structure is the first user space task the kernel created.
1709 static inline int is_global_init(struct task_struct *tsk)
1711 return tsk->pid == 1;
1715 * is_container_init:
1716 * check whether in the task is init in its own pid namespace.
1718 extern int is_container_init(struct task_struct *tsk);
1720 extern struct pid *cad_pid;
1722 extern void free_task(struct task_struct *tsk);
1723 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1725 #ifdef CONFIG_PREEMPT_RT
1726 extern void __put_task_struct_cb(struct rcu_head *rhp);
1728 static inline void put_task_struct(struct task_struct *t)
1730 if (atomic_dec_and_test(&t->usage))
1731 call_rcu(&t->rcu, __put_task_struct_cb);
1733 #else
1734 extern void __put_task_struct(struct task_struct *t);
1736 static inline void put_task_struct(struct task_struct *t)
1738 if (atomic_dec_and_test(&t->usage))
1739 __put_task_struct(t);
1741 #endif
1743 extern cputime_t task_utime(struct task_struct *p);
1744 extern cputime_t task_stime(struct task_struct *p);
1745 extern cputime_t task_gtime(struct task_struct *p);
1748 * Per process flags
1750 #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1751 /* Not implemented yet, only for 486*/
1752 #define PF_STARTING 0x00000002 /* being created */
1753 #define PF_EXITING 0x00000004 /* getting shut down */
1754 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1755 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1756 #define PF_NOSCHED 0x00000020 /* Userspace does not expect scheduling */
1757 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1758 #define PF_HARDIRQ 0x00000080 /* hardirq context */
1759 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1760 #define PF_DUMPCORE 0x00000200 /* dumped core */
1761 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1762 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1763 #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1764 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1765 #define PF_KMAP 0x00004000 /* this context has a kmap */
1766 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1767 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1768 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1769 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1770 #define PF_SWAPOFF 0x00080000 /* I am in swapoff */
1771 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1772 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1773 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1774 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1775 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1776 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1777 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1778 #define PF_SOFTIRQ 0x08000000 /* softirq context */
1779 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1780 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1781 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1782 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1785 * Only the _current_ task can read/write to tsk->flags, but other
1786 * tasks can access tsk->flags in readonly mode for example
1787 * with tsk_used_math (like during threaded core dumping).
1788 * There is however an exception to this rule during ptrace
1789 * or during fork: the ptracer task is allowed to write to the
1790 * child->flags of its traced child (same goes for fork, the parent
1791 * can write to the child->flags), because we're guaranteed the
1792 * child is not running and in turn not changing child->flags
1793 * at the same time the parent does it.
1795 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1796 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1797 #define clear_used_math() clear_stopped_child_used_math(current)
1798 #define set_used_math() set_stopped_child_used_math(current)
1799 #define conditional_stopped_child_used_math(condition, child) \
1800 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1801 #define conditional_used_math(condition) \
1802 conditional_stopped_child_used_math(condition, current)
1803 #define copy_to_stopped_child_used_math(child) \
1804 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1805 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1806 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1807 #define used_math() tsk_used_math(current)
1809 #ifdef CONFIG_SMP
1810 extern int set_cpus_allowed_ptr(struct task_struct *p,
1811 const struct cpumask *new_mask);
1812 #else
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))
1817 return -EINVAL;
1818 return 0;
1820 #endif
1821 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1823 return set_cpus_allowed_ptr(p, &new_mask);
1827 * Architectures can set this to 1 if they have specified
1828 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1829 * but then during bootup it turns out that sched_clock()
1830 * is reliable after all:
1832 #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1833 extern int sched_clock_stable;
1834 #endif
1836 extern unsigned long long sched_clock(void);
1838 extern void sched_clock_init(void);
1839 extern u64 sched_clock_cpu(int cpu);
1841 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1842 static inline void sched_clock_tick(void)
1846 static inline void sched_clock_idle_sleep_event(void)
1850 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1853 #else
1854 extern void sched_clock_tick(void);
1855 extern void sched_clock_idle_sleep_event(void);
1856 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1857 #endif
1860 * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1861 * clock constructed from sched_clock():
1863 extern unsigned long long cpu_clock(int cpu);
1865 extern unsigned long long
1866 task_sched_runtime(struct task_struct *task);
1867 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1869 /* sched_exec is called by processes performing an exec */
1870 #ifdef CONFIG_SMP
1871 extern void sched_exec(void);
1872 #else
1873 #define sched_exec() {}
1874 #endif
1876 extern void sched_clock_idle_sleep_event(void);
1877 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1879 #ifdef CONFIG_HOTPLUG_CPU
1880 extern void idle_task_exit(void);
1881 #else
1882 static inline void idle_task_exit(void) {}
1883 #endif
1885 extern void sched_idle_next(void);
1887 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1888 extern void wake_up_idle_cpu(int cpu);
1889 #else
1890 static inline void wake_up_idle_cpu(int cpu) { }
1891 #endif
1893 extern unsigned int sysctl_sched_latency;
1894 extern unsigned int sysctl_sched_min_granularity;
1895 extern unsigned int sysctl_sched_wakeup_granularity;
1896 extern unsigned int sysctl_sched_shares_ratelimit;
1897 extern unsigned int sysctl_sched_shares_thresh;
1898 #ifdef CONFIG_SCHED_DEBUG
1899 extern unsigned int sysctl_sched_child_runs_first;
1900 extern unsigned int sysctl_sched_features;
1901 extern unsigned int sysctl_sched_migration_cost;
1902 extern unsigned int sysctl_sched_nr_migrate;
1904 int sched_nr_latency_handler(struct ctl_table *table, int write,
1905 struct file *file, void __user *buffer, size_t *length,
1906 loff_t *ppos);
1907 #endif
1908 extern unsigned int sysctl_sched_rt_period;
1909 extern int sysctl_sched_rt_runtime;
1911 int sched_rt_handler(struct ctl_table *table, int write,
1912 struct file *filp, void __user *buffer, size_t *lenp,
1913 loff_t *ppos);
1915 extern unsigned int sysctl_sched_compat_yield;
1917 extern void task_setprio(struct task_struct *p, int prio);
1919 #ifdef CONFIG_RT_MUTEXES
1920 extern int rt_mutex_getprio(struct task_struct *p);
1921 static inline void rt_mutex_setprio(struct task_struct *p, int prio)
1923 task_setprio(p, prio);
1925 extern void rt_mutex_adjust_pi(struct task_struct *p);
1926 #else
1927 static inline int rt_mutex_getprio(struct task_struct *p)
1929 return p->normal_prio;
1931 # define rt_mutex_adjust_pi(p) do { } while (0)
1932 #endif
1934 extern void set_user_nice(struct task_struct *p, long nice);
1935 extern int task_prio(const struct task_struct *p);
1936 extern int task_nice(const struct task_struct *p);
1937 extern int can_nice(const struct task_struct *p, const int nice);
1938 extern int task_curr(const struct task_struct *p);
1939 extern int idle_cpu(int cpu);
1940 extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1941 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1942 struct sched_param *);
1943 extern struct task_struct *idle_task(int cpu);
1944 extern struct task_struct *curr_task(int cpu);
1945 extern void set_curr_task(int cpu, struct task_struct *p);
1947 void yield(void);
1948 void __yield(void);
1951 * The default (Linux) execution domain.
1953 extern struct exec_domain default_exec_domain;
1955 union thread_union {
1956 struct thread_info thread_info;
1957 unsigned long stack[THREAD_SIZE/sizeof(long)];
1960 #ifndef __HAVE_ARCH_KSTACK_END
1961 static inline int kstack_end(void *addr)
1963 /* Reliable end of stack detection:
1964 * Some APM bios versions misalign the stack
1966 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
1968 #endif
1970 extern union thread_union init_thread_union;
1971 extern struct task_struct init_task;
1973 extern struct mm_struct init_mm;
1975 extern struct pid_namespace init_pid_ns;
1978 * find a task by one of its numerical ids
1980 * find_task_by_pid_type_ns():
1981 * it is the most generic call - it finds a task by all id,
1982 * type and namespace specified
1983 * find_task_by_pid_ns():
1984 * finds a task by its pid in the specified namespace
1985 * find_task_by_vpid():
1986 * finds a task by its virtual pid
1988 * see also find_vpid() etc in include/linux/pid.h
1991 extern struct task_struct *find_task_by_pid_type_ns(int type, int pid,
1992 struct pid_namespace *ns);
1994 extern struct task_struct *find_task_by_vpid(pid_t nr);
1995 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
1996 struct pid_namespace *ns);
1998 extern void __set_special_pids(struct pid *pid);
2000 /* per-UID process charging. */
2001 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2002 static inline struct user_struct *get_uid(struct user_struct *u)
2004 atomic_inc(&u->__count);
2005 return u;
2007 extern void free_uid(struct user_struct *);
2008 extern void release_uids(struct user_namespace *ns);
2010 #include <asm/current.h>
2012 extern void do_timer(unsigned long ticks);
2014 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2015 extern int wake_up_process(struct task_struct *tsk);
2016 extern int wake_up_process_mutex(struct task_struct * tsk);
2017 extern int wake_up_process_sync(struct task_struct * tsk);
2018 extern int wake_up_process_mutex_sync(struct task_struct * tsk);
2019 extern void wake_up_new_task(struct task_struct *tsk,
2020 unsigned long clone_flags);
2021 #ifdef CONFIG_SMP
2022 extern void kick_process(struct task_struct *tsk);
2023 #else
2024 static inline void kick_process(struct task_struct *tsk) { }
2025 #endif
2026 extern void sched_fork(struct task_struct *p, int clone_flags);
2027 extern void sched_dead(struct task_struct *p);
2029 extern void proc_caches_init(void);
2030 extern void flush_signals(struct task_struct *);
2031 extern void ignore_signals(struct task_struct *);
2032 extern void flush_signal_handlers(struct task_struct *, int force_default);
2033 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2035 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2037 unsigned long flags;
2038 int ret;
2040 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2041 ret = dequeue_signal(tsk, mask, info);
2042 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2044 return ret;
2047 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2048 sigset_t *mask);
2049 extern void unblock_all_signals(void);
2050 extern void release_task(struct task_struct * p);
2051 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2052 extern int force_sigsegv(int, struct task_struct *);
2053 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2054 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2055 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2056 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2057 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2058 extern int kill_pid(struct pid *pid, int sig, int priv);
2059 extern int kill_proc_info(int, struct siginfo *, pid_t);
2060 extern int do_notify_parent(struct task_struct *, int);
2061 extern void force_sig(int, struct task_struct *);
2062 extern void force_sig_specific(int, struct task_struct *);
2063 extern int send_sig(int, struct task_struct *, int);
2064 extern void zap_other_threads(struct task_struct *p);
2065 extern struct sigqueue *sigqueue_alloc(void);
2066 extern void sigqueue_free(struct sigqueue *);
2067 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2068 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2069 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2071 static inline int kill_cad_pid(int sig, int priv)
2073 return kill_pid(cad_pid, sig, priv);
2076 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2077 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2078 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2079 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2081 static inline int is_si_special(const struct siginfo *info)
2083 return info <= SEND_SIG_FORCED;
2086 /* True if we are on the alternate signal stack. */
2088 static inline int on_sig_stack(unsigned long sp)
2090 return (sp - current->sas_ss_sp < current->sas_ss_size);
2093 static inline int sas_ss_flags(unsigned long sp)
2095 return (current->sas_ss_size == 0 ? SS_DISABLE
2096 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2100 * Routines for handling mm_structs
2102 extern struct mm_struct * mm_alloc(void);
2104 /* mmdrop drops the mm and the page tables */
2105 extern void __mmdrop(struct mm_struct *);
2106 extern void __mmdrop_delayed(struct mm_struct *);
2108 static inline void mmdrop(struct mm_struct * mm)
2110 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2111 __mmdrop(mm);
2114 static inline void mmdrop_delayed(struct mm_struct * mm)
2116 if (atomic_dec_and_test(&mm->mm_count))
2117 __mmdrop_delayed(mm);
2120 /* mmput gets rid of the mappings and all user-space */
2121 extern void mmput(struct mm_struct *);
2122 /* Grab a reference to a task's mm, if it is not already going away */
2123 extern struct mm_struct *get_task_mm(struct task_struct *task);
2124 /* Remove the current tasks stale references to the old mm_struct */
2125 extern void mm_release(struct task_struct *, struct mm_struct *);
2126 /* Allocate a new mm structure and copy contents from tsk->mm */
2127 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2129 extern int copy_thread(int, unsigned long, unsigned long, unsigned long, struct task_struct *, struct pt_regs *);
2130 extern void flush_thread(void);
2131 extern void exit_thread(void);
2133 extern void exit_files(struct task_struct *);
2134 extern void __cleanup_signal(struct signal_struct *);
2135 extern void __cleanup_sighand(struct sighand_struct *);
2137 extern void exit_itimers(struct signal_struct *);
2138 extern void flush_itimer_signals(void);
2140 extern NORET_TYPE void do_group_exit(int);
2142 extern void daemonize(const char *, ...);
2143 extern int allow_signal(int);
2144 extern int disallow_signal(int);
2146 extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2147 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2148 struct task_struct *fork_idle(int);
2150 extern void set_task_comm(struct task_struct *tsk, char *from);
2151 extern char *get_task_comm(char *to, struct task_struct *tsk);
2153 #ifdef CONFIG_SMP
2154 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2155 #else
2156 static inline unsigned long wait_task_inactive(struct task_struct *p,
2157 long match_state)
2159 return 1;
2161 #endif
2163 #define next_task(p) list_entry(rcu_dereference((p)->tasks.next), struct task_struct, tasks)
2165 #define for_each_process(p) \
2166 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2168 extern bool is_single_threaded(struct task_struct *);
2171 * Careful: do_each_thread/while_each_thread is a double loop so
2172 * 'break' will not work as expected - use goto instead.
2174 #define do_each_thread(g, t) \
2175 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2177 #define while_each_thread(g, t) \
2178 while ((t = next_thread(t)) != g)
2180 /* de_thread depends on thread_group_leader not being a pid based check */
2181 #define thread_group_leader(p) (p == p->group_leader)
2183 /* Do to the insanities of de_thread it is possible for a process
2184 * to have the pid of the thread group leader without actually being
2185 * the thread group leader. For iteration through the pids in proc
2186 * all we care about is that we have a task with the appropriate
2187 * pid, we don't actually care if we have the right task.
2189 static inline int has_group_leader_pid(struct task_struct *p)
2191 return p->pid == p->tgid;
2194 static inline
2195 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2197 return p1->tgid == p2->tgid;
2200 static inline struct task_struct *next_thread(const struct task_struct *p)
2202 return list_entry(rcu_dereference(p->thread_group.next),
2203 struct task_struct, thread_group);
2206 static inline int thread_group_empty(struct task_struct *p)
2208 return list_empty(&p->thread_group);
2211 #define delay_group_leader(p) \
2212 (thread_group_leader(p) && !thread_group_empty(p))
2215 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2216 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2217 * pins the final release of task.io_context. Also protects ->cpuset and
2218 * ->cgroup.subsys[].
2220 * Nests both inside and outside of read_lock(&tasklist_lock).
2221 * It must not be nested with write_lock_irq(&tasklist_lock),
2222 * neither inside nor outside.
2224 static inline void task_lock(struct task_struct *p)
2226 spin_lock(&p->alloc_lock);
2229 static inline void task_unlock(struct task_struct *p)
2231 spin_unlock(&p->alloc_lock);
2234 extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2235 unsigned long *flags);
2237 static inline void unlock_task_sighand(struct task_struct *tsk,
2238 unsigned long *flags)
2240 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2243 #ifndef __HAVE_THREAD_FUNCTIONS
2245 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2246 #define task_stack_page(task) ((task)->stack)
2248 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2250 *task_thread_info(p) = *task_thread_info(org);
2251 task_thread_info(p)->task = p;
2254 static inline unsigned long *end_of_stack(struct task_struct *p)
2256 return (unsigned long *)(task_thread_info(p) + 1);
2259 #endif
2261 static inline int object_is_on_stack(void *obj)
2263 void *stack = task_stack_page(current);
2265 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2268 extern void thread_info_cache_init(void);
2270 #ifdef CONFIG_DEBUG_STACK_USAGE
2271 static inline unsigned long stack_not_used(struct task_struct *p)
2273 unsigned long *n = end_of_stack(p);
2275 do { /* Skip over canary */
2276 n++;
2277 } while (!*n);
2279 return (unsigned long)n - (unsigned long)end_of_stack(p);
2281 #endif
2283 /* set thread flags in other task's structures
2284 * - see asm/thread_info.h for TIF_xxxx flags available
2286 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2288 set_ti_thread_flag(task_thread_info(tsk), flag);
2291 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2293 clear_ti_thread_flag(task_thread_info(tsk), flag);
2296 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2298 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2301 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2303 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2306 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2308 return test_ti_thread_flag(task_thread_info(tsk), flag);
2311 static inline void set_tsk_need_resched(struct task_struct *tsk)
2313 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2316 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2318 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2321 static inline int test_tsk_need_resched(struct task_struct *tsk)
2323 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2326 static inline int signal_pending(struct task_struct *p)
2328 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2331 extern int __fatal_signal_pending(struct task_struct *p);
2333 static inline int fatal_signal_pending(struct task_struct *p)
2335 return signal_pending(p) && __fatal_signal_pending(p);
2338 static inline int signal_pending_state(long state, struct task_struct *p)
2340 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2341 return 0;
2342 if (!signal_pending(p))
2343 return 0;
2345 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2348 static inline int need_resched(void)
2350 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2354 * cond_resched() and cond_resched_lock(): latency reduction via
2355 * explicit rescheduling in places that are safe. The return
2356 * value indicates whether a reschedule was done in fact.
2357 * cond_resched_lock() will drop the spinlock before scheduling,
2358 * cond_resched_softirq() will enable bhs before scheduling.
2360 extern int _cond_resched(void);
2361 #ifdef CONFIG_PREEMPT_BKL
2362 static inline int cond_resched(void)
2364 return 0;
2366 #else
2367 static inline int cond_resched(void)
2369 return _cond_resched();
2371 #endif
2372 extern int __cond_resched_raw_spinlock(raw_spinlock_t *lock);
2373 extern int __cond_resched_spinlock(spinlock_t *spinlock);
2375 #define cond_resched_lock(lock) \
2376 PICK_SPIN_OP_RET(__cond_resched_raw_spinlock, __cond_resched_spinlock,\
2377 lock)
2379 extern int cond_resched_softirq(void);
2380 static inline int cond_resched_bkl(void)
2382 return _cond_resched();
2384 extern int cond_resched_softirq_context(void);
2385 extern int cond_resched_hardirq_context(void);
2388 * Does a critical section need to be broken due to another
2389 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2390 * but a general need for low latency)
2392 static inline int __raw_spin_needbreak(raw_spinlock_t *lock)
2394 #ifdef CONFIG_PREEMPT
2395 return spin_is_contended(lock);
2396 #else
2397 return 0;
2398 #endif
2402 * Thread group CPU time accounting.
2404 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2405 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2407 static inline void thread_group_cputime_init(struct signal_struct *sig)
2409 sig->cputimer.cputime = INIT_CPUTIME;
2410 spin_lock_init(&sig->cputimer.lock);
2411 sig->cputimer.running = 0;
2414 static inline void thread_group_cputime_free(struct signal_struct *sig)
2418 #ifdef CONFIG_PREEMPT_RT
2419 static inline int __spin_needbreak(spinlock_t *lock)
2421 struct task_struct *tsk = current;
2423 /* break if we are priority boosted */
2424 return tsk->prio < tsk->normal_prio;
2426 #else
2427 static inline int __spin_needbreak(spinlock_t *lock)
2429 /* should never be call outside of RT */
2430 BUG();
2431 return 0;
2433 #endif
2435 #define spin_needbreak(lock) \
2436 PICK_SPIN_OP_RET(__raw_spin_needbreak, __spin_needbreak, lock)
2438 static inline int softirq_need_resched(void)
2440 if (softirq_preemption && (current->flags & PF_SOFTIRQ))
2441 return need_resched();
2442 return 0;
2445 static inline int hardirq_need_resched(void)
2447 if (hardirq_preemption && (current->flags & PF_HARDIRQ))
2448 return need_resched();
2449 return 0;
2453 * Reevaluate whether the task has signals pending delivery.
2454 * Wake the task if so.
2455 * This is required every time the blocked sigset_t changes.
2456 * callers must hold sighand->siglock.
2458 extern void recalc_sigpending_and_wake(struct task_struct *t);
2459 extern void recalc_sigpending(void);
2461 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2464 * Wrappers for p->thread_info->cpu access. No-op on UP.
2466 #ifdef CONFIG_SMP
2468 static inline unsigned int task_cpu(const struct task_struct *p)
2470 return task_thread_info(p)->cpu;
2473 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2475 #else
2477 static inline unsigned int task_cpu(const struct task_struct *p)
2479 return 0;
2482 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2486 #endif /* CONFIG_SMP */
2488 extern void arch_pick_mmap_layout(struct mm_struct *mm);
2490 #ifdef CONFIG_TRACING
2491 extern void
2492 __trace_special(void *__tr, void *__data,
2493 unsigned long arg1, unsigned long arg2, unsigned long arg3);
2494 #else
2495 static inline void
2496 __trace_special(void *__tr, void *__data,
2497 unsigned long arg1, unsigned long arg2, unsigned long arg3)
2500 #endif
2502 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2503 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2505 extern void normalize_rt_tasks(void);
2507 #ifdef CONFIG_GROUP_SCHED
2509 extern struct task_group init_task_group;
2510 #ifdef CONFIG_USER_SCHED
2511 extern struct task_group root_task_group;
2512 extern void set_tg_uid(struct user_struct *user);
2513 #endif
2515 extern struct task_group *sched_create_group(struct task_group *parent);
2516 extern void sched_destroy_group(struct task_group *tg);
2517 extern void sched_move_task(struct task_struct *tsk);
2518 #ifdef CONFIG_FAIR_GROUP_SCHED
2519 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2520 extern unsigned long sched_group_shares(struct task_group *tg);
2521 #endif
2522 #ifdef CONFIG_RT_GROUP_SCHED
2523 extern int sched_group_set_rt_runtime(struct task_group *tg,
2524 long rt_runtime_us);
2525 extern long sched_group_rt_runtime(struct task_group *tg);
2526 extern int sched_group_set_rt_period(struct task_group *tg,
2527 long rt_period_us);
2528 extern long sched_group_rt_period(struct task_group *tg);
2529 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2530 #endif
2531 #endif
2533 extern int task_can_switch_user(struct user_struct *up,
2534 struct task_struct *tsk);
2536 #ifdef CONFIG_TASK_XACCT
2537 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2539 tsk->ioac.rchar += amt;
2542 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2544 tsk->ioac.wchar += amt;
2547 static inline void inc_syscr(struct task_struct *tsk)
2549 tsk->ioac.syscr++;
2552 static inline void inc_syscw(struct task_struct *tsk)
2554 tsk->ioac.syscw++;
2556 #else
2557 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2561 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2565 static inline void inc_syscr(struct task_struct *tsk)
2569 static inline void inc_syscw(struct task_struct *tsk)
2572 #endif
2574 #ifndef TASK_SIZE_OF
2575 #define TASK_SIZE_OF(tsk) TASK_SIZE
2576 #endif
2579 * Call the function if the target task is executing on a CPU right now:
2581 extern void task_oncpu_function_call(struct task_struct *p,
2582 void (*func) (void *info), void *info);
2585 #ifdef CONFIG_MM_OWNER
2586 extern void mm_update_next_owner(struct mm_struct *mm);
2587 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2588 #else
2589 static inline void mm_update_next_owner(struct mm_struct *mm)
2593 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2596 #endif /* CONFIG_MM_OWNER */
2598 #define TASK_STATE_TO_CHAR_STR "RMSDTtZX"
2600 #ifdef CONFIG_SMP
2601 static inline int task_is_current(struct task_struct *task)
2603 return task->oncpu;
2605 #endif
2607 #endif /* __KERNEL__ */
2609 #endif