KEYS: Fix up comments in key management code
[linux-2.6/next.git] / include / linux / sched.h
blobd747f948b34e9608acb1575abdd9aea0e96c7204
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 /* 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 */
34 * Scheduling policies
36 #define SCHED_NORMAL 0
37 #define SCHED_FIFO 1
38 #define SCHED_RR 2
39 #define SCHED_BATCH 3
40 /* SCHED_ISO: reserved but not implemented yet */
41 #define SCHED_IDLE 5
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
45 #ifdef __KERNEL__
47 struct sched_param {
48 int sched_priority;
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>
67 #include <asm/page.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>
96 struct exec_domain;
97 struct futex_pi_state;
98 struct robust_list_head;
99 struct bio_list;
100 struct fs_struct;
101 struct perf_event_context;
104 * List of flags we want to share for kernel threads,
105 * if only because they are not used by them anyway.
107 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 * These are the constant used to fake the fixed-point load-average
111 * counting. Some notes:
112 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
113 * a load-average precision of 10 bits integer + 11 bits fractional
114 * - if you want to count load-averages more often, you need more
115 * precision, or rounding will get you. With 2-second counting freq,
116 * the EXP_n values would be 1981, 2034 and 2043 if still using only
117 * 11 bit fractions.
119 extern unsigned long avenrun[]; /* Load averages */
120 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
122 #define FSHIFT 11 /* nr of bits of precision */
123 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
124 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
125 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
126 #define EXP_5 2014 /* 1/exp(5sec/5min) */
127 #define EXP_15 2037 /* 1/exp(5sec/15min) */
129 #define CALC_LOAD(load,exp,n) \
130 load *= exp; \
131 load += n*(FIXED_1-exp); \
132 load >>= FSHIFT;
134 extern unsigned long total_forks;
135 extern int nr_threads;
136 DECLARE_PER_CPU(unsigned long, process_counts);
137 extern int nr_processes(void);
138 extern unsigned long nr_running(void);
139 extern unsigned long nr_uninterruptible(void);
140 extern unsigned long nr_iowait(void);
141 extern unsigned long nr_iowait_cpu(int cpu);
142 extern unsigned long this_cpu_load(void);
145 extern void calc_global_load(unsigned long ticks);
147 extern unsigned long get_parent_ip(unsigned long addr);
149 struct seq_file;
150 struct cfs_rq;
151 struct task_group;
152 #ifdef CONFIG_SCHED_DEBUG
153 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
154 extern void proc_sched_set_task(struct task_struct *p);
155 extern void
156 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
157 #else
158 static inline void
159 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
162 static inline void proc_sched_set_task(struct task_struct *p)
165 static inline void
166 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
169 #endif
172 * Task state bitmask. NOTE! These bits are also
173 * encoded in fs/proc/array.c: get_task_state().
175 * We have two separate sets of flags: task->state
176 * is about runnability, while task->exit_state are
177 * about the task exiting. Confusing, but this way
178 * modifying one set can't modify the other one by
179 * mistake.
181 #define TASK_RUNNING 0
182 #define TASK_INTERRUPTIBLE 1
183 #define TASK_UNINTERRUPTIBLE 2
184 #define __TASK_STOPPED 4
185 #define __TASK_TRACED 8
186 /* in tsk->exit_state */
187 #define EXIT_ZOMBIE 16
188 #define EXIT_DEAD 32
189 /* in tsk->state again */
190 #define TASK_DEAD 64
191 #define TASK_WAKEKILL 128
192 #define TASK_WAKING 256
193 #define TASK_STATE_MAX 512
195 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
197 extern char ___assert_task_state[1 - 2*!!(
198 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
200 /* Convenience macros for the sake of set_task_state */
201 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
202 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
203 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
205 /* Convenience macros for the sake of wake_up */
206 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
207 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
209 /* get_task_state() */
210 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
211 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
212 __TASK_TRACED)
214 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
215 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
216 #define task_is_dead(task) ((task)->exit_state != 0)
217 #define task_is_stopped_or_traced(task) \
218 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
219 #define task_contributes_to_load(task) \
220 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
221 (task->flags & PF_FREEZING) == 0)
223 #define __set_task_state(tsk, state_value) \
224 do { (tsk)->state = (state_value); } while (0)
225 #define set_task_state(tsk, state_value) \
226 set_mb((tsk)->state, (state_value))
229 * set_current_state() includes a barrier so that the write of current->state
230 * is correctly serialised wrt the caller's subsequent test of whether to
231 * actually sleep:
233 * set_current_state(TASK_UNINTERRUPTIBLE);
234 * if (do_i_need_to_sleep())
235 * schedule();
237 * If the caller does not need such serialisation then use __set_current_state()
239 #define __set_current_state(state_value) \
240 do { current->state = (state_value); } while (0)
241 #define set_current_state(state_value) \
242 set_mb(current->state, (state_value))
244 /* Task command name length */
245 #define TASK_COMM_LEN 16
247 #include <linux/spinlock.h>
250 * This serializes "schedule()" and also protects
251 * the run-queue from deletions/modifications (but
252 * _adding_ to the beginning of the run-queue has
253 * a separate lock).
255 extern rwlock_t tasklist_lock;
256 extern spinlock_t mmlist_lock;
258 struct task_struct;
260 #ifdef CONFIG_PROVE_RCU
261 extern int lockdep_tasklist_lock_is_held(void);
262 #endif /* #ifdef CONFIG_PROVE_RCU */
264 extern void sched_init(void);
265 extern void sched_init_smp(void);
266 extern asmlinkage void schedule_tail(struct task_struct *prev);
267 extern void init_idle(struct task_struct *idle, int cpu);
268 extern void init_idle_bootup_task(struct task_struct *idle);
270 extern int runqueue_is_locked(int cpu);
272 extern cpumask_var_t nohz_cpu_mask;
273 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
274 extern void select_nohz_load_balancer(int stop_tick);
275 extern int get_nohz_timer_target(void);
276 #else
277 static inline void select_nohz_load_balancer(int stop_tick) { }
278 #endif
281 * Only dump TASK_* tasks. (0 for all tasks)
283 extern void show_state_filter(unsigned long state_filter);
285 static inline void show_state(void)
287 show_state_filter(0);
290 extern void show_regs(struct pt_regs *);
293 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
294 * task), SP is the stack pointer of the first frame that should be shown in the back
295 * trace (or NULL if the entire call-chain of the task should be shown).
297 extern void show_stack(struct task_struct *task, unsigned long *sp);
299 void io_schedule(void);
300 long io_schedule_timeout(long timeout);
302 extern void cpu_init (void);
303 extern void trap_init(void);
304 extern void update_process_times(int user);
305 extern void scheduler_tick(void);
307 extern void sched_show_task(struct task_struct *p);
309 #ifdef CONFIG_LOCKUP_DETECTOR
310 extern void touch_softlockup_watchdog(void);
311 extern void touch_softlockup_watchdog_sync(void);
312 extern void touch_all_softlockup_watchdogs(void);
313 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
314 void __user *buffer,
315 size_t *lenp, loff_t *ppos);
316 extern unsigned int softlockup_panic;
317 extern int softlockup_thresh;
318 void lockup_detector_init(void);
319 #else
320 static inline void touch_softlockup_watchdog(void)
323 static inline void touch_softlockup_watchdog_sync(void)
326 static inline void touch_all_softlockup_watchdogs(void)
329 static inline void lockup_detector_init(void)
332 #endif
334 #ifdef CONFIG_DETECT_HUNG_TASK
335 extern unsigned int sysctl_hung_task_panic;
336 extern unsigned long sysctl_hung_task_check_count;
337 extern unsigned long sysctl_hung_task_timeout_secs;
338 extern unsigned long sysctl_hung_task_warnings;
339 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
342 #else
343 /* Avoid need for ifdefs elsewhere in the code */
344 enum { sysctl_hung_task_timeout_secs = 0 };
345 #endif
347 /* Attach to any functions which should be ignored in wchan output. */
348 #define __sched __attribute__((__section__(".sched.text")))
350 /* Linker adds these: start and end of __sched functions */
351 extern char __sched_text_start[], __sched_text_end[];
353 /* Is this address in the __sched functions? */
354 extern int in_sched_functions(unsigned long addr);
356 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
357 extern signed long schedule_timeout(signed long timeout);
358 extern signed long schedule_timeout_interruptible(signed long timeout);
359 extern signed long schedule_timeout_killable(signed long timeout);
360 extern signed long schedule_timeout_uninterruptible(signed long timeout);
361 asmlinkage void schedule(void);
362 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
364 struct nsproxy;
365 struct user_namespace;
368 * Default maximum number of active map areas, this limits the number of vmas
369 * per mm struct. Users can overwrite this number by sysctl but there is a
370 * problem.
372 * When a program's coredump is generated as ELF format, a section is created
373 * per a vma. In ELF, the number of sections is represented in unsigned short.
374 * This means the number of sections should be smaller than 65535 at coredump.
375 * Because the kernel adds some informative sections to a image of program at
376 * generating coredump, we need some margin. The number of extra sections is
377 * 1-3 now and depends on arch. We use "5" as safe margin, here.
379 #define MAPCOUNT_ELF_CORE_MARGIN (5)
380 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
382 extern int sysctl_max_map_count;
384 #include <linux/aio.h>
386 #ifdef CONFIG_MMU
387 extern void arch_pick_mmap_layout(struct mm_struct *mm);
388 extern unsigned long
389 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
390 unsigned long, unsigned long);
391 extern unsigned long
392 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
393 unsigned long len, unsigned long pgoff,
394 unsigned long flags);
395 extern void arch_unmap_area(struct mm_struct *, unsigned long);
396 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
397 #else
398 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #endif
402 extern void set_dumpable(struct mm_struct *mm, int value);
403 extern int get_dumpable(struct mm_struct *mm);
405 /* mm flags */
406 /* dumpable bits */
407 #define MMF_DUMPABLE 0 /* core dump is permitted */
408 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
410 #define MMF_DUMPABLE_BITS 2
411 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
413 /* coredump filter bits */
414 #define MMF_DUMP_ANON_PRIVATE 2
415 #define MMF_DUMP_ANON_SHARED 3
416 #define MMF_DUMP_MAPPED_PRIVATE 4
417 #define MMF_DUMP_MAPPED_SHARED 5
418 #define MMF_DUMP_ELF_HEADERS 6
419 #define MMF_DUMP_HUGETLB_PRIVATE 7
420 #define MMF_DUMP_HUGETLB_SHARED 8
422 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
423 #define MMF_DUMP_FILTER_BITS 7
424 #define MMF_DUMP_FILTER_MASK \
425 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
426 #define MMF_DUMP_FILTER_DEFAULT \
427 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
428 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
430 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
431 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
432 #else
433 # define MMF_DUMP_MASK_DEFAULT_ELF 0
434 #endif
435 /* leave room for more dump flags */
436 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
437 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
441 struct sighand_struct {
442 atomic_t count;
443 struct k_sigaction action[_NSIG];
444 spinlock_t siglock;
445 wait_queue_head_t signalfd_wqh;
448 struct pacct_struct {
449 int ac_flag;
450 long ac_exitcode;
451 unsigned long ac_mem;
452 cputime_t ac_utime, ac_stime;
453 unsigned long ac_minflt, ac_majflt;
456 struct cpu_itimer {
457 cputime_t expires;
458 cputime_t incr;
459 u32 error;
460 u32 incr_error;
464 * struct task_cputime - collected CPU time counts
465 * @utime: time spent in user mode, in &cputime_t units
466 * @stime: time spent in kernel mode, in &cputime_t units
467 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
469 * This structure groups together three kinds of CPU time that are
470 * tracked for threads and thread groups. Most things considering
471 * CPU time want to group these counts together and treat all three
472 * of them in parallel.
474 struct task_cputime {
475 cputime_t utime;
476 cputime_t stime;
477 unsigned long long sum_exec_runtime;
479 /* Alternate field names when used to cache expirations. */
480 #define prof_exp stime
481 #define virt_exp utime
482 #define sched_exp sum_exec_runtime
484 #define INIT_CPUTIME \
485 (struct task_cputime) { \
486 .utime = cputime_zero, \
487 .stime = cputime_zero, \
488 .sum_exec_runtime = 0, \
492 * Disable preemption until the scheduler is running.
493 * Reset by start_kernel()->sched_init()->init_idle().
495 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
496 * before the scheduler is active -- see should_resched().
498 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
501 * struct thread_group_cputimer - thread group interval timer counts
502 * @cputime: thread group interval timers.
503 * @running: non-zero when there are timers running and
504 * @cputime receives updates.
505 * @lock: lock for fields in this struct.
507 * This structure contains the version of task_cputime, above, that is
508 * used for thread group CPU timer calculations.
510 struct thread_group_cputimer {
511 struct task_cputime cputime;
512 int running;
513 spinlock_t lock;
516 struct autogroup;
519 * NOTE! "signal_struct" does not have it's own
520 * locking, because a shared signal_struct always
521 * implies a shared sighand_struct, so locking
522 * sighand_struct is always a proper superset of
523 * the locking of signal_struct.
525 struct signal_struct {
526 atomic_t sigcnt;
527 atomic_t live;
528 int nr_threads;
530 wait_queue_head_t wait_chldexit; /* for wait4() */
532 /* current thread group signal load-balancing target: */
533 struct task_struct *curr_target;
535 /* shared signal handling: */
536 struct sigpending shared_pending;
538 /* thread group exit support */
539 int group_exit_code;
540 /* overloaded:
541 * - notify group_exit_task when ->count is equal to notify_count
542 * - everyone except group_exit_task is stopped during signal delivery
543 * of fatal signals, group_exit_task processes the signal.
545 int notify_count;
546 struct task_struct *group_exit_task;
548 /* thread group stop support, overloads group_exit_code too */
549 int group_stop_count;
550 unsigned int flags; /* see SIGNAL_* flags below */
552 /* POSIX.1b Interval Timers */
553 struct list_head posix_timers;
555 /* ITIMER_REAL timer for the process */
556 struct hrtimer real_timer;
557 struct pid *leader_pid;
558 ktime_t it_real_incr;
561 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
562 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
563 * values are defined to 0 and 1 respectively
565 struct cpu_itimer it[2];
568 * Thread group totals for process CPU timers.
569 * See thread_group_cputimer(), et al, for details.
571 struct thread_group_cputimer cputimer;
573 /* Earliest-expiration cache. */
574 struct task_cputime cputime_expires;
576 struct list_head cpu_timers[3];
578 struct pid *tty_old_pgrp;
580 /* boolean value for session group leader */
581 int leader;
583 struct tty_struct *tty; /* NULL if no tty */
585 #ifdef CONFIG_SCHED_AUTOGROUP
586 struct autogroup *autogroup;
587 #endif
589 * Cumulative resource counters for dead threads in the group,
590 * and for reaped dead child processes forked by this group.
591 * Live threads maintain their own counters and add to these
592 * in __exit_signal, except for the group leader.
594 cputime_t utime, stime, cutime, cstime;
595 cputime_t gtime;
596 cputime_t cgtime;
597 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
598 cputime_t prev_utime, prev_stime;
599 #endif
600 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
601 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
602 unsigned long inblock, oublock, cinblock, coublock;
603 unsigned long maxrss, cmaxrss;
604 struct task_io_accounting ioac;
607 * Cumulative ns of schedule CPU time fo dead threads in the
608 * group, not including a zombie group leader, (This only differs
609 * from jiffies_to_ns(utime + stime) if sched_clock uses something
610 * other than jiffies.)
612 unsigned long long sum_sched_runtime;
615 * We don't bother to synchronize most readers of this at all,
616 * because there is no reader checking a limit that actually needs
617 * to get both rlim_cur and rlim_max atomically, and either one
618 * alone is a single word that can safely be read normally.
619 * getrlimit/setrlimit use task_lock(current->group_leader) to
620 * protect this instead of the siglock, because they really
621 * have no need to disable irqs.
623 struct rlimit rlim[RLIM_NLIMITS];
625 #ifdef CONFIG_BSD_PROCESS_ACCT
626 struct pacct_struct pacct; /* per-process accounting information */
627 #endif
628 #ifdef CONFIG_TASKSTATS
629 struct taskstats *stats;
630 #endif
631 #ifdef CONFIG_AUDIT
632 unsigned audit_tty;
633 struct tty_audit_buf *tty_audit_buf;
634 #endif
636 int oom_adj; /* OOM kill score adjustment (bit shift) */
637 int oom_score_adj; /* OOM kill score adjustment */
638 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
639 * Only settable by CAP_SYS_RESOURCE. */
641 struct mutex cred_guard_mutex; /* guard against foreign influences on
642 * credential calculations
643 * (notably. ptrace) */
646 /* Context switch must be unlocked if interrupts are to be enabled */
647 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
648 # define __ARCH_WANT_UNLOCKED_CTXSW
649 #endif
652 * Bits in flags field of signal_struct.
654 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
655 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
656 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
657 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
659 * Pending notifications to parent.
661 #define SIGNAL_CLD_STOPPED 0x00000010
662 #define SIGNAL_CLD_CONTINUED 0x00000020
663 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
665 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
667 /* If true, all threads except ->group_exit_task have pending SIGKILL */
668 static inline int signal_group_exit(const struct signal_struct *sig)
670 return (sig->flags & SIGNAL_GROUP_EXIT) ||
671 (sig->group_exit_task != NULL);
675 * Some day this will be a full-fledged user tracking system..
677 struct user_struct {
678 atomic_t __count; /* reference count */
679 atomic_t processes; /* How many processes does this user have? */
680 atomic_t files; /* How many open files does this user have? */
681 atomic_t sigpending; /* How many pending signals does this user have? */
682 #ifdef CONFIG_INOTIFY_USER
683 atomic_t inotify_watches; /* How many inotify watches does this user have? */
684 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
685 #endif
686 #ifdef CONFIG_FANOTIFY
687 atomic_t fanotify_listeners;
688 #endif
689 #ifdef CONFIG_EPOLL
690 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
691 #endif
692 #ifdef CONFIG_POSIX_MQUEUE
693 /* protected by mq_lock */
694 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
695 #endif
696 unsigned long locked_shm; /* How many pages of mlocked shm ? */
698 #ifdef CONFIG_KEYS
699 struct key *uid_keyring; /* UID specific keyring */
700 struct key *session_keyring; /* UID's default session keyring */
701 #endif
703 /* Hash table maintenance information */
704 struct hlist_node uidhash_node;
705 uid_t uid;
706 struct user_namespace *user_ns;
708 #ifdef CONFIG_PERF_EVENTS
709 atomic_long_t locked_vm;
710 #endif
713 extern int uids_sysfs_init(void);
715 extern struct user_struct *find_user(uid_t);
717 extern struct user_struct root_user;
718 #define INIT_USER (&root_user)
721 struct backing_dev_info;
722 struct reclaim_state;
724 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
725 struct sched_info {
726 /* cumulative counters */
727 unsigned long pcount; /* # of times run on this cpu */
728 unsigned long long run_delay; /* time spent waiting on a runqueue */
730 /* timestamps */
731 unsigned long long last_arrival,/* when we last ran on a cpu */
732 last_queued; /* when we were last queued to run */
733 #ifdef CONFIG_SCHEDSTATS
734 /* BKL stats */
735 unsigned int bkl_count;
736 #endif
738 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
740 #ifdef CONFIG_TASK_DELAY_ACCT
741 struct task_delay_info {
742 spinlock_t lock;
743 unsigned int flags; /* Private per-task flags */
745 /* For each stat XXX, add following, aligned appropriately
747 * struct timespec XXX_start, XXX_end;
748 * u64 XXX_delay;
749 * u32 XXX_count;
751 * Atomicity of updates to XXX_delay, XXX_count protected by
752 * single lock above (split into XXX_lock if contention is an issue).
756 * XXX_count is incremented on every XXX operation, the delay
757 * associated with the operation is added to XXX_delay.
758 * XXX_delay contains the accumulated delay time in nanoseconds.
760 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
761 u64 blkio_delay; /* wait for sync block io completion */
762 u64 swapin_delay; /* wait for swapin block io completion */
763 u32 blkio_count; /* total count of the number of sync block */
764 /* io operations performed */
765 u32 swapin_count; /* total count of the number of swapin block */
766 /* io operations performed */
768 struct timespec freepages_start, freepages_end;
769 u64 freepages_delay; /* wait for memory reclaim */
770 u32 freepages_count; /* total count of memory reclaim */
772 #endif /* CONFIG_TASK_DELAY_ACCT */
774 static inline int sched_info_on(void)
776 #ifdef CONFIG_SCHEDSTATS
777 return 1;
778 #elif defined(CONFIG_TASK_DELAY_ACCT)
779 extern int delayacct_on;
780 return delayacct_on;
781 #else
782 return 0;
783 #endif
786 enum cpu_idle_type {
787 CPU_IDLE,
788 CPU_NOT_IDLE,
789 CPU_NEWLY_IDLE,
790 CPU_MAX_IDLE_TYPES
794 * sched-domains (multiprocessor balancing) declarations:
798 * Increase resolution of nice-level calculations:
800 #define SCHED_LOAD_SHIFT 10
801 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
803 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
805 #ifdef CONFIG_SMP
806 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
807 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
808 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
809 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
810 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
811 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
812 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
813 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
814 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
815 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
816 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
817 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
818 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
820 enum powersavings_balance_level {
821 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
822 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
823 * first for long running threads
825 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
826 * cpu package for power savings
828 MAX_POWERSAVINGS_BALANCE_LEVELS
831 extern int sched_mc_power_savings, sched_smt_power_savings;
833 static inline int sd_balance_for_mc_power(void)
835 if (sched_smt_power_savings)
836 return SD_POWERSAVINGS_BALANCE;
838 if (!sched_mc_power_savings)
839 return SD_PREFER_SIBLING;
841 return 0;
844 static inline int sd_balance_for_package_power(void)
846 if (sched_mc_power_savings | sched_smt_power_savings)
847 return SD_POWERSAVINGS_BALANCE;
849 return SD_PREFER_SIBLING;
852 extern int __weak arch_sd_sibiling_asym_packing(void);
855 * Optimise SD flags for power savings:
856 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
857 * Keep default SD flags if sched_{smt,mc}_power_saving=0
860 static inline int sd_power_saving_flags(void)
862 if (sched_mc_power_savings | sched_smt_power_savings)
863 return SD_BALANCE_NEWIDLE;
865 return 0;
868 struct sched_group {
869 struct sched_group *next; /* Must be a circular list */
872 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
873 * single CPU.
875 unsigned int cpu_power, cpu_power_orig;
876 unsigned int group_weight;
879 * The CPUs this group covers.
881 * NOTE: this field is variable length. (Allocated dynamically
882 * by attaching extra space to the end of the structure,
883 * depending on how many CPUs the kernel has booted up with)
885 * It is also be embedded into static data structures at build
886 * time. (See 'struct static_sched_group' in kernel/sched.c)
888 unsigned long cpumask[0];
891 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
893 return to_cpumask(sg->cpumask);
896 enum sched_domain_level {
897 SD_LV_NONE = 0,
898 SD_LV_SIBLING,
899 SD_LV_MC,
900 SD_LV_BOOK,
901 SD_LV_CPU,
902 SD_LV_NODE,
903 SD_LV_ALLNODES,
904 SD_LV_MAX
907 struct sched_domain_attr {
908 int relax_domain_level;
911 #define SD_ATTR_INIT (struct sched_domain_attr) { \
912 .relax_domain_level = -1, \
915 struct sched_domain {
916 /* These fields must be setup */
917 struct sched_domain *parent; /* top domain must be null terminated */
918 struct sched_domain *child; /* bottom domain must be null terminated */
919 struct sched_group *groups; /* the balancing groups of the domain */
920 unsigned long min_interval; /* Minimum balance interval ms */
921 unsigned long max_interval; /* Maximum balance interval ms */
922 unsigned int busy_factor; /* less balancing by factor if busy */
923 unsigned int imbalance_pct; /* No balance until over watermark */
924 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
925 unsigned int busy_idx;
926 unsigned int idle_idx;
927 unsigned int newidle_idx;
928 unsigned int wake_idx;
929 unsigned int forkexec_idx;
930 unsigned int smt_gain;
931 int flags; /* See SD_* */
932 enum sched_domain_level level;
934 /* Runtime fields. */
935 unsigned long last_balance; /* init to jiffies. units in jiffies */
936 unsigned int balance_interval; /* initialise to 1. units in ms. */
937 unsigned int nr_balance_failed; /* initialise to 0 */
939 u64 last_update;
941 #ifdef CONFIG_SCHEDSTATS
942 /* load_balance() stats */
943 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
949 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
952 /* Active load balancing */
953 unsigned int alb_count;
954 unsigned int alb_failed;
955 unsigned int alb_pushed;
957 /* SD_BALANCE_EXEC stats */
958 unsigned int sbe_count;
959 unsigned int sbe_balanced;
960 unsigned int sbe_pushed;
962 /* SD_BALANCE_FORK stats */
963 unsigned int sbf_count;
964 unsigned int sbf_balanced;
965 unsigned int sbf_pushed;
967 /* try_to_wake_up() stats */
968 unsigned int ttwu_wake_remote;
969 unsigned int ttwu_move_affine;
970 unsigned int ttwu_move_balance;
971 #endif
972 #ifdef CONFIG_SCHED_DEBUG
973 char *name;
974 #endif
976 unsigned int span_weight;
978 * Span of all CPUs in this domain.
980 * NOTE: this field is variable length. (Allocated dynamically
981 * by attaching extra space to the end of the structure,
982 * depending on how many CPUs the kernel has booted up with)
984 * It is also be embedded into static data structures at build
985 * time. (See 'struct static_sched_domain' in kernel/sched.c)
987 unsigned long span[0];
990 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
992 return to_cpumask(sd->span);
995 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
996 struct sched_domain_attr *dattr_new);
998 /* Allocate an array of sched domains, for partition_sched_domains(). */
999 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1000 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1002 /* Test a flag in parent sched domain */
1003 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1005 if (sd->parent && (sd->parent->flags & flag))
1006 return 1;
1008 return 0;
1011 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1012 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1014 #else /* CONFIG_SMP */
1016 struct sched_domain_attr;
1018 static inline void
1019 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1020 struct sched_domain_attr *dattr_new)
1023 #endif /* !CONFIG_SMP */
1026 struct io_context; /* See blkdev.h */
1029 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1030 extern void prefetch_stack(struct task_struct *t);
1031 #else
1032 static inline void prefetch_stack(struct task_struct *t) { }
1033 #endif
1035 struct audit_context; /* See audit.c */
1036 struct mempolicy;
1037 struct pipe_inode_info;
1038 struct uts_namespace;
1040 struct rq;
1041 struct sched_domain;
1044 * wake flags
1046 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1047 #define WF_FORK 0x02 /* child wakeup after fork */
1049 #define ENQUEUE_WAKEUP 1
1050 #define ENQUEUE_WAKING 2
1051 #define ENQUEUE_HEAD 4
1053 #define DEQUEUE_SLEEP 1
1055 struct sched_class {
1056 const struct sched_class *next;
1058 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1059 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1060 void (*yield_task) (struct rq *rq);
1062 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1064 struct task_struct * (*pick_next_task) (struct rq *rq);
1065 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1067 #ifdef CONFIG_SMP
1068 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1069 int sd_flag, int flags);
1071 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1072 void (*post_schedule) (struct rq *this_rq);
1073 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1074 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1076 void (*set_cpus_allowed)(struct task_struct *p,
1077 const struct cpumask *newmask);
1079 void (*rq_online)(struct rq *rq);
1080 void (*rq_offline)(struct rq *rq);
1081 #endif
1083 void (*set_curr_task) (struct rq *rq);
1084 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1085 void (*task_fork) (struct task_struct *p);
1087 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1088 int running);
1089 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1090 int running);
1091 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1092 int oldprio, int running);
1094 unsigned int (*get_rr_interval) (struct rq *rq,
1095 struct task_struct *task);
1097 #ifdef CONFIG_FAIR_GROUP_SCHED
1098 void (*task_move_group) (struct task_struct *p, int on_rq);
1099 #endif
1102 struct load_weight {
1103 unsigned long weight, inv_weight;
1106 #ifdef CONFIG_SCHEDSTATS
1107 struct sched_statistics {
1108 u64 wait_start;
1109 u64 wait_max;
1110 u64 wait_count;
1111 u64 wait_sum;
1112 u64 iowait_count;
1113 u64 iowait_sum;
1115 u64 sleep_start;
1116 u64 sleep_max;
1117 s64 sum_sleep_runtime;
1119 u64 block_start;
1120 u64 block_max;
1121 u64 exec_max;
1122 u64 slice_max;
1124 u64 nr_migrations_cold;
1125 u64 nr_failed_migrations_affine;
1126 u64 nr_failed_migrations_running;
1127 u64 nr_failed_migrations_hot;
1128 u64 nr_forced_migrations;
1130 u64 nr_wakeups;
1131 u64 nr_wakeups_sync;
1132 u64 nr_wakeups_migrate;
1133 u64 nr_wakeups_local;
1134 u64 nr_wakeups_remote;
1135 u64 nr_wakeups_affine;
1136 u64 nr_wakeups_affine_attempts;
1137 u64 nr_wakeups_passive;
1138 u64 nr_wakeups_idle;
1140 #endif
1142 struct sched_entity {
1143 struct load_weight load; /* for load-balancing */
1144 struct rb_node run_node;
1145 struct list_head group_node;
1146 unsigned int on_rq;
1148 u64 exec_start;
1149 u64 sum_exec_runtime;
1150 u64 vruntime;
1151 u64 prev_sum_exec_runtime;
1153 u64 nr_migrations;
1155 #ifdef CONFIG_SCHEDSTATS
1156 struct sched_statistics statistics;
1157 #endif
1159 #ifdef CONFIG_FAIR_GROUP_SCHED
1160 struct sched_entity *parent;
1161 /* rq on which this entity is (to be) queued: */
1162 struct cfs_rq *cfs_rq;
1163 /* rq "owned" by this entity/group: */
1164 struct cfs_rq *my_q;
1165 #endif
1168 struct sched_rt_entity {
1169 struct list_head run_list;
1170 unsigned long timeout;
1171 unsigned int time_slice;
1172 int nr_cpus_allowed;
1174 struct sched_rt_entity *back;
1175 #ifdef CONFIG_RT_GROUP_SCHED
1176 struct sched_rt_entity *parent;
1177 /* rq on which this entity is (to be) queued: */
1178 struct rt_rq *rt_rq;
1179 /* rq "owned" by this entity/group: */
1180 struct rt_rq *my_q;
1181 #endif
1184 struct rcu_node;
1186 enum perf_event_task_context {
1187 perf_invalid_context = -1,
1188 perf_hw_context = 0,
1189 perf_sw_context,
1190 perf_nr_task_contexts,
1193 struct task_struct {
1194 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1195 void *stack;
1196 atomic_t usage;
1197 unsigned int flags; /* per process flags, defined below */
1198 unsigned int ptrace;
1200 int lock_depth; /* BKL lock depth */
1202 #ifdef CONFIG_SMP
1203 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1204 int oncpu;
1205 #endif
1206 #endif
1208 int prio, static_prio, normal_prio;
1209 unsigned int rt_priority;
1210 const struct sched_class *sched_class;
1211 struct sched_entity se;
1212 struct sched_rt_entity rt;
1214 #ifdef CONFIG_PREEMPT_NOTIFIERS
1215 /* list of struct preempt_notifier: */
1216 struct hlist_head preempt_notifiers;
1217 #endif
1220 * fpu_counter contains the number of consecutive context switches
1221 * that the FPU is used. If this is over a threshold, the lazy fpu
1222 * saving becomes unlazy to save the trap. This is an unsigned char
1223 * so that after 256 times the counter wraps and the behavior turns
1224 * lazy again; this to deal with bursty apps that only use FPU for
1225 * a short time
1227 unsigned char fpu_counter;
1228 #ifdef CONFIG_BLK_DEV_IO_TRACE
1229 unsigned int btrace_seq;
1230 #endif
1232 unsigned int policy;
1233 cpumask_t cpus_allowed;
1235 #ifdef CONFIG_PREEMPT_RCU
1236 int rcu_read_lock_nesting;
1237 char rcu_read_unlock_special;
1238 struct list_head rcu_node_entry;
1239 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1240 #ifdef CONFIG_TREE_PREEMPT_RCU
1241 struct rcu_node *rcu_blocked_node;
1242 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1243 #ifdef CONFIG_RCU_BOOST
1244 struct rt_mutex *rcu_boost_mutex;
1245 #endif /* #ifdef CONFIG_RCU_BOOST */
1247 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1248 struct sched_info sched_info;
1249 #endif
1251 struct list_head tasks;
1252 #ifdef CONFIG_SMP
1253 struct plist_node pushable_tasks;
1254 #endif
1256 struct mm_struct *mm, *active_mm;
1257 #if defined(SPLIT_RSS_COUNTING)
1258 struct task_rss_stat rss_stat;
1259 #endif
1260 /* task state */
1261 int exit_state;
1262 int exit_code, exit_signal;
1263 int pdeath_signal; /* The signal sent when the parent dies */
1264 /* ??? */
1265 unsigned int personality;
1266 unsigned did_exec:1;
1267 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1268 * execve */
1269 unsigned in_iowait:1;
1272 /* Revert to default priority/policy when forking */
1273 unsigned sched_reset_on_fork:1;
1275 pid_t pid;
1276 pid_t tgid;
1278 #ifdef CONFIG_CC_STACKPROTECTOR
1279 /* Canary value for the -fstack-protector gcc feature */
1280 unsigned long stack_canary;
1281 #endif
1284 * pointers to (original) parent process, youngest child, younger sibling,
1285 * older sibling, respectively. (p->father can be replaced with
1286 * p->real_parent->pid)
1288 struct task_struct *real_parent; /* real parent process */
1289 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1291 * children/sibling forms the list of my natural children
1293 struct list_head children; /* list of my children */
1294 struct list_head sibling; /* linkage in my parent's children list */
1295 struct task_struct *group_leader; /* threadgroup leader */
1298 * ptraced is the list of tasks this task is using ptrace on.
1299 * This includes both natural children and PTRACE_ATTACH targets.
1300 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1302 struct list_head ptraced;
1303 struct list_head ptrace_entry;
1305 /* PID/PID hash table linkage. */
1306 struct pid_link pids[PIDTYPE_MAX];
1307 struct list_head thread_group;
1309 struct completion *vfork_done; /* for vfork() */
1310 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1311 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1313 cputime_t utime, stime, utimescaled, stimescaled;
1314 cputime_t gtime;
1315 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1316 cputime_t prev_utime, prev_stime;
1317 #endif
1318 unsigned long nvcsw, nivcsw; /* context switch counts */
1319 struct timespec start_time; /* monotonic time */
1320 struct timespec real_start_time; /* boot based time */
1321 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1322 unsigned long min_flt, maj_flt;
1324 struct task_cputime cputime_expires;
1325 struct list_head cpu_timers[3];
1327 /* process credentials */
1328 const struct cred __rcu *real_cred; /* objective and real subjective task
1329 * credentials (COW) */
1330 const struct cred __rcu *cred; /* effective (overridable) subjective task
1331 * credentials (COW) */
1332 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1334 char comm[TASK_COMM_LEN]; /* executable name excluding path
1335 - access with [gs]et_task_comm (which lock
1336 it with task_lock())
1337 - initialized normally by setup_new_exec */
1338 /* file system info */
1339 int link_count, total_link_count;
1340 #ifdef CONFIG_SYSVIPC
1341 /* ipc stuff */
1342 struct sysv_sem sysvsem;
1343 #endif
1344 #ifdef CONFIG_DETECT_HUNG_TASK
1345 /* hung task detection */
1346 unsigned long last_switch_count;
1347 #endif
1348 /* CPU-specific state of this task */
1349 struct thread_struct thread;
1350 /* filesystem information */
1351 struct fs_struct *fs;
1352 /* open file information */
1353 struct files_struct *files;
1354 /* namespaces */
1355 struct nsproxy *nsproxy;
1356 /* signal handlers */
1357 struct signal_struct *signal;
1358 struct sighand_struct *sighand;
1360 sigset_t blocked, real_blocked;
1361 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1362 struct sigpending pending;
1364 unsigned long sas_ss_sp;
1365 size_t sas_ss_size;
1366 int (*notifier)(void *priv);
1367 void *notifier_data;
1368 sigset_t *notifier_mask;
1369 struct audit_context *audit_context;
1370 #ifdef CONFIG_AUDITSYSCALL
1371 uid_t loginuid;
1372 unsigned int sessionid;
1373 #endif
1374 seccomp_t seccomp;
1376 /* Thread group tracking */
1377 u32 parent_exec_id;
1378 u32 self_exec_id;
1379 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1380 * mempolicy */
1381 spinlock_t alloc_lock;
1383 #ifdef CONFIG_GENERIC_HARDIRQS
1384 /* IRQ handler threads */
1385 struct irqaction *irqaction;
1386 #endif
1388 /* Protection of the PI data structures: */
1389 raw_spinlock_t pi_lock;
1391 #ifdef CONFIG_RT_MUTEXES
1392 /* PI waiters blocked on a rt_mutex held by this task */
1393 struct plist_head pi_waiters;
1394 /* Deadlock detection and priority inheritance handling */
1395 struct rt_mutex_waiter *pi_blocked_on;
1396 #endif
1398 #ifdef CONFIG_DEBUG_MUTEXES
1399 /* mutex deadlock detection */
1400 struct mutex_waiter *blocked_on;
1401 #endif
1402 #ifdef CONFIG_TRACE_IRQFLAGS
1403 unsigned int irq_events;
1404 unsigned long hardirq_enable_ip;
1405 unsigned long hardirq_disable_ip;
1406 unsigned int hardirq_enable_event;
1407 unsigned int hardirq_disable_event;
1408 int hardirqs_enabled;
1409 int hardirq_context;
1410 unsigned long softirq_disable_ip;
1411 unsigned long softirq_enable_ip;
1412 unsigned int softirq_disable_event;
1413 unsigned int softirq_enable_event;
1414 int softirqs_enabled;
1415 int softirq_context;
1416 #endif
1417 #ifdef CONFIG_LOCKDEP
1418 # define MAX_LOCK_DEPTH 48UL
1419 u64 curr_chain_key;
1420 int lockdep_depth;
1421 unsigned int lockdep_recursion;
1422 struct held_lock held_locks[MAX_LOCK_DEPTH];
1423 gfp_t lockdep_reclaim_gfp;
1424 #endif
1426 /* journalling filesystem info */
1427 void *journal_info;
1429 /* stacked block device info */
1430 struct bio_list *bio_list;
1432 /* VM state */
1433 struct reclaim_state *reclaim_state;
1435 struct backing_dev_info *backing_dev_info;
1437 struct io_context *io_context;
1439 unsigned long ptrace_message;
1440 siginfo_t *last_siginfo; /* For ptrace use. */
1441 struct task_io_accounting ioac;
1442 #if defined(CONFIG_TASK_XACCT)
1443 u64 acct_rss_mem1; /* accumulated rss usage */
1444 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1445 cputime_t acct_timexpd; /* stime + utime since last update */
1446 #endif
1447 #ifdef CONFIG_CPUSETS
1448 nodemask_t mems_allowed; /* Protected by alloc_lock */
1449 int mems_allowed_change_disable;
1450 int cpuset_mem_spread_rotor;
1451 int cpuset_slab_spread_rotor;
1452 #endif
1453 #ifdef CONFIG_CGROUPS
1454 /* Control Group info protected by css_set_lock */
1455 struct css_set __rcu *cgroups;
1456 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1457 struct list_head cg_list;
1458 #endif
1459 #ifdef CONFIG_FUTEX
1460 struct robust_list_head __user *robust_list;
1461 #ifdef CONFIG_COMPAT
1462 struct compat_robust_list_head __user *compat_robust_list;
1463 #endif
1464 struct list_head pi_state_list;
1465 struct futex_pi_state *pi_state_cache;
1466 #endif
1467 #ifdef CONFIG_PERF_EVENTS
1468 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1469 struct mutex perf_event_mutex;
1470 struct list_head perf_event_list;
1471 #endif
1472 #ifdef CONFIG_NUMA
1473 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1474 short il_next;
1475 #endif
1476 atomic_t fs_excl; /* holding fs exclusive resources */
1477 struct rcu_head rcu;
1480 * cache last used pipe for splice
1482 struct pipe_inode_info *splice_pipe;
1483 #ifdef CONFIG_TASK_DELAY_ACCT
1484 struct task_delay_info *delays;
1485 #endif
1486 #ifdef CONFIG_FAULT_INJECTION
1487 int make_it_fail;
1488 #endif
1489 struct prop_local_single dirties;
1490 #ifdef CONFIG_LATENCYTOP
1491 int latency_record_count;
1492 struct latency_record latency_record[LT_SAVECOUNT];
1493 #endif
1495 * time slack values; these are used to round up poll() and
1496 * select() etc timeout values. These are in nanoseconds.
1498 unsigned long timer_slack_ns;
1499 unsigned long default_timer_slack_ns;
1501 struct list_head *scm_work_list;
1502 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1503 /* Index of current stored address in ret_stack */
1504 int curr_ret_stack;
1505 /* Stack of return addresses for return function tracing */
1506 struct ftrace_ret_stack *ret_stack;
1507 /* time stamp for last schedule */
1508 unsigned long long ftrace_timestamp;
1510 * Number of functions that haven't been traced
1511 * because of depth overrun.
1513 atomic_t trace_overrun;
1514 /* Pause for the tracing */
1515 atomic_t tracing_graph_pause;
1516 #endif
1517 #ifdef CONFIG_TRACING
1518 /* state flags for use by tracers */
1519 unsigned long trace;
1520 /* bitmask of trace recursion */
1521 unsigned long trace_recursion;
1522 #endif /* CONFIG_TRACING */
1523 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1524 struct memcg_batch_info {
1525 int do_batch; /* incremented when batch uncharge started */
1526 struct mem_cgroup *memcg; /* target memcg of uncharge */
1527 unsigned long bytes; /* uncharged usage */
1528 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1529 } memcg_batch;
1530 #endif
1533 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1534 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1537 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1538 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1539 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1540 * values are inverted: lower p->prio value means higher priority.
1542 * The MAX_USER_RT_PRIO value allows the actual maximum
1543 * RT priority to be separate from the value exported to
1544 * user-space. This allows kernel threads to set their
1545 * priority to a value higher than any user task. Note:
1546 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1549 #define MAX_USER_RT_PRIO 100
1550 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1552 #define MAX_PRIO (MAX_RT_PRIO + 40)
1553 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1555 static inline int rt_prio(int prio)
1557 if (unlikely(prio < MAX_RT_PRIO))
1558 return 1;
1559 return 0;
1562 static inline int rt_task(struct task_struct *p)
1564 return rt_prio(p->prio);
1567 static inline struct pid *task_pid(struct task_struct *task)
1569 return task->pids[PIDTYPE_PID].pid;
1572 static inline struct pid *task_tgid(struct task_struct *task)
1574 return task->group_leader->pids[PIDTYPE_PID].pid;
1578 * Without tasklist or rcu lock it is not safe to dereference
1579 * the result of task_pgrp/task_session even if task == current,
1580 * we can race with another thread doing sys_setsid/sys_setpgid.
1582 static inline struct pid *task_pgrp(struct task_struct *task)
1584 return task->group_leader->pids[PIDTYPE_PGID].pid;
1587 static inline struct pid *task_session(struct task_struct *task)
1589 return task->group_leader->pids[PIDTYPE_SID].pid;
1592 struct pid_namespace;
1595 * the helpers to get the task's different pids as they are seen
1596 * from various namespaces
1598 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1599 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1600 * current.
1601 * task_xid_nr_ns() : id seen from the ns specified;
1603 * set_task_vxid() : assigns a virtual id to a task;
1605 * see also pid_nr() etc in include/linux/pid.h
1607 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1608 struct pid_namespace *ns);
1610 static inline pid_t task_pid_nr(struct task_struct *tsk)
1612 return tsk->pid;
1615 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1616 struct pid_namespace *ns)
1618 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1621 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1623 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1627 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1629 return tsk->tgid;
1632 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1634 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1636 return pid_vnr(task_tgid(tsk));
1640 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1641 struct pid_namespace *ns)
1643 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1646 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1648 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1652 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1653 struct pid_namespace *ns)
1655 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1658 static inline pid_t task_session_vnr(struct task_struct *tsk)
1660 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1663 /* obsolete, do not use */
1664 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1666 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1670 * pid_alive - check that a task structure is not stale
1671 * @p: Task structure to be checked.
1673 * Test if a process is not yet dead (at most zombie state)
1674 * If pid_alive fails, then pointers within the task structure
1675 * can be stale and must not be dereferenced.
1677 static inline int pid_alive(struct task_struct *p)
1679 return p->pids[PIDTYPE_PID].pid != NULL;
1683 * is_global_init - check if a task structure is init
1684 * @tsk: Task structure to be checked.
1686 * Check if a task structure is the first user space task the kernel created.
1688 static inline int is_global_init(struct task_struct *tsk)
1690 return tsk->pid == 1;
1694 * is_container_init:
1695 * check whether in the task is init in its own pid namespace.
1697 extern int is_container_init(struct task_struct *tsk);
1699 extern struct pid *cad_pid;
1701 extern void free_task(struct task_struct *tsk);
1702 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1704 extern void __put_task_struct(struct task_struct *t);
1706 static inline void put_task_struct(struct task_struct *t)
1708 if (atomic_dec_and_test(&t->usage))
1709 __put_task_struct(t);
1712 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1713 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1716 * Per process flags
1718 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1719 #define PF_STARTING 0x00000002 /* being created */
1720 #define PF_EXITING 0x00000004 /* getting shut down */
1721 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1722 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1723 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1724 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1725 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1726 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1727 #define PF_DUMPCORE 0x00000200 /* dumped core */
1728 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1729 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1730 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1731 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1732 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1733 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1734 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1735 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1736 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1737 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1738 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1739 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1740 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1741 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1742 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1743 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1744 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1745 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1746 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1747 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1748 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1751 * Only the _current_ task can read/write to tsk->flags, but other
1752 * tasks can access tsk->flags in readonly mode for example
1753 * with tsk_used_math (like during threaded core dumping).
1754 * There is however an exception to this rule during ptrace
1755 * or during fork: the ptracer task is allowed to write to the
1756 * child->flags of its traced child (same goes for fork, the parent
1757 * can write to the child->flags), because we're guaranteed the
1758 * child is not running and in turn not changing child->flags
1759 * at the same time the parent does it.
1761 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1762 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1763 #define clear_used_math() clear_stopped_child_used_math(current)
1764 #define set_used_math() set_stopped_child_used_math(current)
1765 #define conditional_stopped_child_used_math(condition, child) \
1766 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1767 #define conditional_used_math(condition) \
1768 conditional_stopped_child_used_math(condition, current)
1769 #define copy_to_stopped_child_used_math(child) \
1770 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1771 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1772 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1773 #define used_math() tsk_used_math(current)
1775 #ifdef CONFIG_PREEMPT_RCU
1777 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1778 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1779 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1781 static inline void rcu_copy_process(struct task_struct *p)
1783 p->rcu_read_lock_nesting = 0;
1784 p->rcu_read_unlock_special = 0;
1785 #ifdef CONFIG_TREE_PREEMPT_RCU
1786 p->rcu_blocked_node = NULL;
1787 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1788 #ifdef CONFIG_RCU_BOOST
1789 p->rcu_boost_mutex = NULL;
1790 #endif /* #ifdef CONFIG_RCU_BOOST */
1791 INIT_LIST_HEAD(&p->rcu_node_entry);
1794 #else
1796 static inline void rcu_copy_process(struct task_struct *p)
1800 #endif
1802 #ifdef CONFIG_SMP
1803 extern int set_cpus_allowed_ptr(struct task_struct *p,
1804 const struct cpumask *new_mask);
1805 #else
1806 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1807 const struct cpumask *new_mask)
1809 if (!cpumask_test_cpu(0, new_mask))
1810 return -EINVAL;
1811 return 0;
1813 #endif
1815 #ifndef CONFIG_CPUMASK_OFFSTACK
1816 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1818 return set_cpus_allowed_ptr(p, &new_mask);
1820 #endif
1823 * Do not use outside of architecture code which knows its limitations.
1825 * sched_clock() has no promise of monotonicity or bounded drift between
1826 * CPUs, use (which you should not) requires disabling IRQs.
1828 * Please use one of the three interfaces below.
1830 extern unsigned long long notrace sched_clock(void);
1832 * See the comment in kernel/sched_clock.c
1834 extern u64 cpu_clock(int cpu);
1835 extern u64 local_clock(void);
1836 extern u64 sched_clock_cpu(int cpu);
1839 extern void sched_clock_init(void);
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
1855 * Architectures can set this to 1 if they have specified
1856 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1857 * but then during bootup it turns out that sched_clock()
1858 * is reliable after all:
1860 extern int sched_clock_stable;
1862 extern void sched_clock_tick(void);
1863 extern void sched_clock_idle_sleep_event(void);
1864 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1865 #endif
1867 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1869 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1870 * The reason for this explicit opt-in is not to have perf penalty with
1871 * slow sched_clocks.
1873 extern void enable_sched_clock_irqtime(void);
1874 extern void disable_sched_clock_irqtime(void);
1875 #else
1876 static inline void enable_sched_clock_irqtime(void) {}
1877 static inline void disable_sched_clock_irqtime(void) {}
1878 #endif
1880 extern unsigned long long
1881 task_sched_runtime(struct task_struct *task);
1882 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1884 /* sched_exec is called by processes performing an exec */
1885 #ifdef CONFIG_SMP
1886 extern void sched_exec(void);
1887 #else
1888 #define sched_exec() {}
1889 #endif
1891 extern void sched_clock_idle_sleep_event(void);
1892 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1894 #ifdef CONFIG_HOTPLUG_CPU
1895 extern void idle_task_exit(void);
1896 #else
1897 static inline void idle_task_exit(void) {}
1898 #endif
1900 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1901 extern void wake_up_idle_cpu(int cpu);
1902 #else
1903 static inline void wake_up_idle_cpu(int cpu) { }
1904 #endif
1906 extern unsigned int sysctl_sched_latency;
1907 extern unsigned int sysctl_sched_min_granularity;
1908 extern unsigned int sysctl_sched_wakeup_granularity;
1909 extern unsigned int sysctl_sched_child_runs_first;
1911 enum sched_tunable_scaling {
1912 SCHED_TUNABLESCALING_NONE,
1913 SCHED_TUNABLESCALING_LOG,
1914 SCHED_TUNABLESCALING_LINEAR,
1915 SCHED_TUNABLESCALING_END,
1917 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1919 #ifdef CONFIG_SCHED_DEBUG
1920 extern unsigned int sysctl_sched_migration_cost;
1921 extern unsigned int sysctl_sched_nr_migrate;
1922 extern unsigned int sysctl_sched_time_avg;
1923 extern unsigned int sysctl_timer_migration;
1924 extern unsigned int sysctl_sched_shares_window;
1926 int sched_proc_update_handler(struct ctl_table *table, int write,
1927 void __user *buffer, size_t *length,
1928 loff_t *ppos);
1929 #endif
1930 #ifdef CONFIG_SCHED_DEBUG
1931 static inline unsigned int get_sysctl_timer_migration(void)
1933 return sysctl_timer_migration;
1935 #else
1936 static inline unsigned int get_sysctl_timer_migration(void)
1938 return 1;
1940 #endif
1941 extern unsigned int sysctl_sched_rt_period;
1942 extern int sysctl_sched_rt_runtime;
1944 int sched_rt_handler(struct ctl_table *table, int write,
1945 void __user *buffer, size_t *lenp,
1946 loff_t *ppos);
1948 extern unsigned int sysctl_sched_compat_yield;
1950 #ifdef CONFIG_SCHED_AUTOGROUP
1951 extern unsigned int sysctl_sched_autogroup_enabled;
1953 extern void sched_autogroup_create_attach(struct task_struct *p);
1954 extern void sched_autogroup_detach(struct task_struct *p);
1955 extern void sched_autogroup_fork(struct signal_struct *sig);
1956 extern void sched_autogroup_exit(struct signal_struct *sig);
1957 #ifdef CONFIG_PROC_FS
1958 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1959 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1960 #endif
1961 #else
1962 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1963 static inline void sched_autogroup_detach(struct task_struct *p) { }
1964 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1965 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1966 #endif
1968 #ifdef CONFIG_RT_MUTEXES
1969 extern int rt_mutex_getprio(struct task_struct *p);
1970 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1971 extern void rt_mutex_adjust_pi(struct task_struct *p);
1972 #else
1973 static inline int rt_mutex_getprio(struct task_struct *p)
1975 return p->normal_prio;
1977 # define rt_mutex_adjust_pi(p) do { } while (0)
1978 #endif
1980 extern void set_user_nice(struct task_struct *p, long nice);
1981 extern int task_prio(const struct task_struct *p);
1982 extern int task_nice(const struct task_struct *p);
1983 extern int can_nice(const struct task_struct *p, const int nice);
1984 extern int task_curr(const struct task_struct *p);
1985 extern int idle_cpu(int cpu);
1986 extern int sched_setscheduler(struct task_struct *, int,
1987 const struct sched_param *);
1988 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1989 const struct sched_param *);
1990 extern struct task_struct *idle_task(int cpu);
1991 extern struct task_struct *curr_task(int cpu);
1992 extern void set_curr_task(int cpu, struct task_struct *p);
1994 void yield(void);
1997 * The default (Linux) execution domain.
1999 extern struct exec_domain default_exec_domain;
2001 union thread_union {
2002 struct thread_info thread_info;
2003 unsigned long stack[THREAD_SIZE/sizeof(long)];
2006 #ifndef __HAVE_ARCH_KSTACK_END
2007 static inline int kstack_end(void *addr)
2009 /* Reliable end of stack detection:
2010 * Some APM bios versions misalign the stack
2012 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2014 #endif
2016 extern union thread_union init_thread_union;
2017 extern struct task_struct init_task;
2019 extern struct mm_struct init_mm;
2021 extern struct pid_namespace init_pid_ns;
2024 * find a task by one of its numerical ids
2026 * find_task_by_pid_ns():
2027 * finds a task by its pid in the specified namespace
2028 * find_task_by_vpid():
2029 * finds a task by its virtual pid
2031 * see also find_vpid() etc in include/linux/pid.h
2034 extern struct task_struct *find_task_by_vpid(pid_t nr);
2035 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2036 struct pid_namespace *ns);
2038 extern void __set_special_pids(struct pid *pid);
2040 /* per-UID process charging. */
2041 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2042 static inline struct user_struct *get_uid(struct user_struct *u)
2044 atomic_inc(&u->__count);
2045 return u;
2047 extern void free_uid(struct user_struct *);
2048 extern void release_uids(struct user_namespace *ns);
2050 #include <asm/current.h>
2052 extern void do_timer(unsigned long ticks);
2054 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2055 extern int wake_up_process(struct task_struct *tsk);
2056 extern void wake_up_new_task(struct task_struct *tsk,
2057 unsigned long clone_flags);
2058 #ifdef CONFIG_SMP
2059 extern void kick_process(struct task_struct *tsk);
2060 #else
2061 static inline void kick_process(struct task_struct *tsk) { }
2062 #endif
2063 extern void sched_fork(struct task_struct *p, int clone_flags);
2064 extern void sched_dead(struct task_struct *p);
2066 extern void proc_caches_init(void);
2067 extern void flush_signals(struct task_struct *);
2068 extern void __flush_signals(struct task_struct *);
2069 extern void ignore_signals(struct task_struct *);
2070 extern void flush_signal_handlers(struct task_struct *, int force_default);
2071 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2073 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2075 unsigned long flags;
2076 int ret;
2078 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2079 ret = dequeue_signal(tsk, mask, info);
2080 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2082 return ret;
2085 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2086 sigset_t *mask);
2087 extern void unblock_all_signals(void);
2088 extern void release_task(struct task_struct * p);
2089 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2090 extern int force_sigsegv(int, struct task_struct *);
2091 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2092 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2093 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2094 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2095 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2096 extern int kill_pid(struct pid *pid, int sig, int priv);
2097 extern int kill_proc_info(int, struct siginfo *, pid_t);
2098 extern int do_notify_parent(struct task_struct *, int);
2099 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2100 extern void force_sig(int, struct task_struct *);
2101 extern int send_sig(int, struct task_struct *, int);
2102 extern int zap_other_threads(struct task_struct *p);
2103 extern struct sigqueue *sigqueue_alloc(void);
2104 extern void sigqueue_free(struct sigqueue *);
2105 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2106 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2107 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2109 static inline int kill_cad_pid(int sig, int priv)
2111 return kill_pid(cad_pid, sig, priv);
2114 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2115 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2116 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2117 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2120 * True if we are on the alternate signal stack.
2122 static inline int on_sig_stack(unsigned long sp)
2124 #ifdef CONFIG_STACK_GROWSUP
2125 return sp >= current->sas_ss_sp &&
2126 sp - current->sas_ss_sp < current->sas_ss_size;
2127 #else
2128 return sp > current->sas_ss_sp &&
2129 sp - current->sas_ss_sp <= current->sas_ss_size;
2130 #endif
2133 static inline int sas_ss_flags(unsigned long sp)
2135 return (current->sas_ss_size == 0 ? SS_DISABLE
2136 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2140 * Routines for handling mm_structs
2142 extern struct mm_struct * mm_alloc(void);
2144 /* mmdrop drops the mm and the page tables */
2145 extern void __mmdrop(struct mm_struct *);
2146 static inline void mmdrop(struct mm_struct * mm)
2148 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2149 __mmdrop(mm);
2152 /* mmput gets rid of the mappings and all user-space */
2153 extern void mmput(struct mm_struct *);
2154 /* Grab a reference to a task's mm, if it is not already going away */
2155 extern struct mm_struct *get_task_mm(struct task_struct *task);
2156 /* Remove the current tasks stale references to the old mm_struct */
2157 extern void mm_release(struct task_struct *, struct mm_struct *);
2158 /* Allocate a new mm structure and copy contents from tsk->mm */
2159 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2161 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2162 struct task_struct *, struct pt_regs *);
2163 extern void flush_thread(void);
2164 extern void exit_thread(void);
2166 extern void exit_files(struct task_struct *);
2167 extern void __cleanup_sighand(struct sighand_struct *);
2169 extern void exit_itimers(struct signal_struct *);
2170 extern void flush_itimer_signals(void);
2172 extern NORET_TYPE void do_group_exit(int);
2174 extern void daemonize(const char *, ...);
2175 extern int allow_signal(int);
2176 extern int disallow_signal(int);
2178 extern int do_execve(const char *,
2179 const char __user * const __user *,
2180 const char __user * const __user *, struct pt_regs *);
2181 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2182 struct task_struct *fork_idle(int);
2184 extern void set_task_comm(struct task_struct *tsk, char *from);
2185 extern char *get_task_comm(char *to, struct task_struct *tsk);
2187 #ifdef CONFIG_SMP
2188 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2189 #else
2190 static inline unsigned long wait_task_inactive(struct task_struct *p,
2191 long match_state)
2193 return 1;
2195 #endif
2197 #define next_task(p) \
2198 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2200 #define for_each_process(p) \
2201 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2203 extern bool current_is_single_threaded(void);
2206 * Careful: do_each_thread/while_each_thread is a double loop so
2207 * 'break' will not work as expected - use goto instead.
2209 #define do_each_thread(g, t) \
2210 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2212 #define while_each_thread(g, t) \
2213 while ((t = next_thread(t)) != g)
2215 static inline int get_nr_threads(struct task_struct *tsk)
2217 return tsk->signal->nr_threads;
2220 /* de_thread depends on thread_group_leader not being a pid based check */
2221 #define thread_group_leader(p) (p == p->group_leader)
2223 /* Do to the insanities of de_thread it is possible for a process
2224 * to have the pid of the thread group leader without actually being
2225 * the thread group leader. For iteration through the pids in proc
2226 * all we care about is that we have a task with the appropriate
2227 * pid, we don't actually care if we have the right task.
2229 static inline int has_group_leader_pid(struct task_struct *p)
2231 return p->pid == p->tgid;
2234 static inline
2235 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2237 return p1->tgid == p2->tgid;
2240 static inline struct task_struct *next_thread(const struct task_struct *p)
2242 return list_entry_rcu(p->thread_group.next,
2243 struct task_struct, thread_group);
2246 static inline int thread_group_empty(struct task_struct *p)
2248 return list_empty(&p->thread_group);
2251 #define delay_group_leader(p) \
2252 (thread_group_leader(p) && !thread_group_empty(p))
2254 static inline int task_detached(struct task_struct *p)
2256 return p->exit_signal == -1;
2260 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2261 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2262 * pins the final release of task.io_context. Also protects ->cpuset and
2263 * ->cgroup.subsys[].
2265 * Nests both inside and outside of read_lock(&tasklist_lock).
2266 * It must not be nested with write_lock_irq(&tasklist_lock),
2267 * neither inside nor outside.
2269 static inline void task_lock(struct task_struct *p)
2271 spin_lock(&p->alloc_lock);
2274 static inline void task_unlock(struct task_struct *p)
2276 spin_unlock(&p->alloc_lock);
2279 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2280 unsigned long *flags);
2282 #define lock_task_sighand(tsk, flags) \
2283 ({ struct sighand_struct *__ss; \
2284 __cond_lock(&(tsk)->sighand->siglock, \
2285 (__ss = __lock_task_sighand(tsk, flags))); \
2286 __ss; \
2287 }) \
2289 static inline void unlock_task_sighand(struct task_struct *tsk,
2290 unsigned long *flags)
2292 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2295 #ifndef __HAVE_THREAD_FUNCTIONS
2297 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2298 #define task_stack_page(task) ((task)->stack)
2300 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2302 *task_thread_info(p) = *task_thread_info(org);
2303 task_thread_info(p)->task = p;
2306 static inline unsigned long *end_of_stack(struct task_struct *p)
2308 return (unsigned long *)(task_thread_info(p) + 1);
2311 #endif
2313 static inline int object_is_on_stack(void *obj)
2315 void *stack = task_stack_page(current);
2317 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2320 extern void thread_info_cache_init(void);
2322 #ifdef CONFIG_DEBUG_STACK_USAGE
2323 static inline unsigned long stack_not_used(struct task_struct *p)
2325 unsigned long *n = end_of_stack(p);
2327 do { /* Skip over canary */
2328 n++;
2329 } while (!*n);
2331 return (unsigned long)n - (unsigned long)end_of_stack(p);
2333 #endif
2335 /* set thread flags in other task's structures
2336 * - see asm/thread_info.h for TIF_xxxx flags available
2338 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2340 set_ti_thread_flag(task_thread_info(tsk), flag);
2343 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2345 clear_ti_thread_flag(task_thread_info(tsk), flag);
2348 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2350 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2353 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2355 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2358 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2360 return test_ti_thread_flag(task_thread_info(tsk), flag);
2363 static inline void set_tsk_need_resched(struct task_struct *tsk)
2365 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2368 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2370 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2373 static inline int test_tsk_need_resched(struct task_struct *tsk)
2375 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2378 static inline int restart_syscall(void)
2380 set_tsk_thread_flag(current, TIF_SIGPENDING);
2381 return -ERESTARTNOINTR;
2384 static inline int signal_pending(struct task_struct *p)
2386 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2389 static inline int __fatal_signal_pending(struct task_struct *p)
2391 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2394 static inline int fatal_signal_pending(struct task_struct *p)
2396 return signal_pending(p) && __fatal_signal_pending(p);
2399 static inline int signal_pending_state(long state, struct task_struct *p)
2401 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2402 return 0;
2403 if (!signal_pending(p))
2404 return 0;
2406 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2409 static inline int need_resched(void)
2411 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2415 * cond_resched() and cond_resched_lock(): latency reduction via
2416 * explicit rescheduling in places that are safe. The return
2417 * value indicates whether a reschedule was done in fact.
2418 * cond_resched_lock() will drop the spinlock before scheduling,
2419 * cond_resched_softirq() will enable bhs before scheduling.
2421 extern int _cond_resched(void);
2423 #define cond_resched() ({ \
2424 __might_sleep(__FILE__, __LINE__, 0); \
2425 _cond_resched(); \
2428 extern int __cond_resched_lock(spinlock_t *lock);
2430 #ifdef CONFIG_PREEMPT
2431 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2432 #else
2433 #define PREEMPT_LOCK_OFFSET 0
2434 #endif
2436 #define cond_resched_lock(lock) ({ \
2437 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2438 __cond_resched_lock(lock); \
2441 extern int __cond_resched_softirq(void);
2443 #define cond_resched_softirq() ({ \
2444 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2445 __cond_resched_softirq(); \
2449 * Does a critical section need to be broken due to another
2450 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2451 * but a general need for low latency)
2453 static inline int spin_needbreak(spinlock_t *lock)
2455 #ifdef CONFIG_PREEMPT
2456 return spin_is_contended(lock);
2457 #else
2458 return 0;
2459 #endif
2463 * Thread group CPU time accounting.
2465 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2466 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2468 static inline void thread_group_cputime_init(struct signal_struct *sig)
2470 spin_lock_init(&sig->cputimer.lock);
2474 * Reevaluate whether the task has signals pending delivery.
2475 * Wake the task if so.
2476 * This is required every time the blocked sigset_t changes.
2477 * callers must hold sighand->siglock.
2479 extern void recalc_sigpending_and_wake(struct task_struct *t);
2480 extern void recalc_sigpending(void);
2482 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2485 * Wrappers for p->thread_info->cpu access. No-op on UP.
2487 #ifdef CONFIG_SMP
2489 static inline unsigned int task_cpu(const struct task_struct *p)
2491 return task_thread_info(p)->cpu;
2494 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2496 #else
2498 static inline unsigned int task_cpu(const struct task_struct *p)
2500 return 0;
2503 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2507 #endif /* CONFIG_SMP */
2509 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2510 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2512 extern void normalize_rt_tasks(void);
2514 #ifdef CONFIG_CGROUP_SCHED
2516 extern struct task_group root_task_group;
2518 extern struct task_group *sched_create_group(struct task_group *parent);
2519 extern void sched_destroy_group(struct task_group *tg);
2520 extern void sched_move_task(struct task_struct *tsk);
2521 #ifdef CONFIG_FAIR_GROUP_SCHED
2522 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2523 extern unsigned long sched_group_shares(struct task_group *tg);
2524 #endif
2525 #ifdef CONFIG_RT_GROUP_SCHED
2526 extern int sched_group_set_rt_runtime(struct task_group *tg,
2527 long rt_runtime_us);
2528 extern long sched_group_rt_runtime(struct task_group *tg);
2529 extern int sched_group_set_rt_period(struct task_group *tg,
2530 long rt_period_us);
2531 extern long sched_group_rt_period(struct task_group *tg);
2532 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2533 #endif
2534 #endif
2536 extern int task_can_switch_user(struct user_struct *up,
2537 struct task_struct *tsk);
2539 #ifdef CONFIG_TASK_XACCT
2540 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2542 tsk->ioac.rchar += amt;
2545 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2547 tsk->ioac.wchar += amt;
2550 static inline void inc_syscr(struct task_struct *tsk)
2552 tsk->ioac.syscr++;
2555 static inline void inc_syscw(struct task_struct *tsk)
2557 tsk->ioac.syscw++;
2559 #else
2560 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2564 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2568 static inline void inc_syscr(struct task_struct *tsk)
2572 static inline void inc_syscw(struct task_struct *tsk)
2575 #endif
2577 #ifndef TASK_SIZE_OF
2578 #define TASK_SIZE_OF(tsk) TASK_SIZE
2579 #endif
2582 * Call the function if the target task is executing on a CPU right now:
2584 extern void task_oncpu_function_call(struct task_struct *p,
2585 void (*func) (void *info), void *info);
2588 #ifdef CONFIG_MM_OWNER
2589 extern void mm_update_next_owner(struct mm_struct *mm);
2590 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2591 #else
2592 static inline void mm_update_next_owner(struct mm_struct *mm)
2596 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2599 #endif /* CONFIG_MM_OWNER */
2601 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2602 unsigned int limit)
2604 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2607 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2608 unsigned int limit)
2610 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2613 static inline unsigned long rlimit(unsigned int limit)
2615 return task_rlimit(current, limit);
2618 static inline unsigned long rlimit_max(unsigned int limit)
2620 return task_rlimit_max(current, limit);
2623 #endif /* __KERNEL__ */
2625 #endif