sched: Remove unused parameters from sched_fork() and wake_up_new_task()
[linux-2.6.git] / include / linux / sched.h
blob12211e1666e2510b6389c47db75110b6ec8232ac
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;
102 struct blk_plug;
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 * 11 bit fractions.
120 extern unsigned long avenrun[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
130 #define CALC_LOAD(load,exp,n) \
131 load *= exp; \
132 load += n*(FIXED_1-exp); \
133 load >>= FSHIFT;
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
146 extern void calc_global_load(unsigned long ticks);
148 extern unsigned long get_parent_ip(unsigned long addr);
150 struct seq_file;
151 struct cfs_rq;
152 struct task_group;
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
156 extern void
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 #else
159 static inline void
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
163 static inline void proc_sched_set_task(struct task_struct *p)
166 static inline void
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
170 #endif
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
180 * mistake.
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
189 #define EXIT_DEAD 32
190 /* in tsk->state again */
191 #define TASK_DEAD 64
192 #define TASK_WAKEKILL 128
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 __TASK_TRACED)
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
232 * actually sleep:
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
236 * schedule();
238 * If the caller does not need such serialisation then use __set_current_state()
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
248 #include <linux/spinlock.h>
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
254 * a separate lock).
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
259 struct task_struct;
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
271 extern int runqueue_is_locked(int cpu);
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern void select_nohz_load_balancer(int stop_tick);
276 extern int get_nohz_timer_target(void);
277 #else
278 static inline void select_nohz_load_balancer(int stop_tick) { }
279 #endif
282 * Only dump TASK_* tasks. (0 for all tasks)
284 extern void show_state_filter(unsigned long state_filter);
286 static inline void show_state(void)
288 show_state_filter(0);
291 extern void show_regs(struct pt_regs *);
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
298 extern void show_stack(struct task_struct *task, unsigned long *sp);
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout);
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user);
306 extern void scheduler_tick(void);
308 extern void sched_show_task(struct task_struct *p);
310 #ifdef CONFIG_LOCKUP_DETECTOR
311 extern void touch_softlockup_watchdog(void);
312 extern void touch_softlockup_watchdog_sync(void);
313 extern void touch_all_softlockup_watchdogs(void);
314 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
315 void __user *buffer,
316 size_t *lenp, loff_t *ppos);
317 extern unsigned int softlockup_panic;
318 extern int softlockup_thresh;
319 void lockup_detector_init(void);
320 #else
321 static inline void touch_softlockup_watchdog(void)
324 static inline void touch_softlockup_watchdog_sync(void)
327 static inline void touch_all_softlockup_watchdogs(void)
330 static inline void lockup_detector_init(void)
333 #endif
335 #ifdef CONFIG_DETECT_HUNG_TASK
336 extern unsigned int sysctl_hung_task_panic;
337 extern unsigned long sysctl_hung_task_check_count;
338 extern unsigned long sysctl_hung_task_timeout_secs;
339 extern unsigned long sysctl_hung_task_warnings;
340 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
341 void __user *buffer,
342 size_t *lenp, loff_t *ppos);
343 #else
344 /* Avoid need for ifdefs elsewhere in the code */
345 enum { sysctl_hung_task_timeout_secs = 0 };
346 #endif
348 /* Attach to any functions which should be ignored in wchan output. */
349 #define __sched __attribute__((__section__(".sched.text")))
351 /* Linker adds these: start and end of __sched functions */
352 extern char __sched_text_start[], __sched_text_end[];
354 /* Is this address in the __sched functions? */
355 extern int in_sched_functions(unsigned long addr);
357 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
358 extern signed long schedule_timeout(signed long timeout);
359 extern signed long schedule_timeout_interruptible(signed long timeout);
360 extern signed long schedule_timeout_killable(signed long timeout);
361 extern signed long schedule_timeout_uninterruptible(signed long timeout);
362 asmlinkage void schedule(void);
363 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
365 struct nsproxy;
366 struct user_namespace;
369 * Default maximum number of active map areas, this limits the number of vmas
370 * per mm struct. Users can overwrite this number by sysctl but there is a
371 * problem.
373 * When a program's coredump is generated as ELF format, a section is created
374 * per a vma. In ELF, the number of sections is represented in unsigned short.
375 * This means the number of sections should be smaller than 65535 at coredump.
376 * Because the kernel adds some informative sections to a image of program at
377 * generating coredump, we need some margin. The number of extra sections is
378 * 1-3 now and depends on arch. We use "5" as safe margin, here.
380 #define MAPCOUNT_ELF_CORE_MARGIN (5)
381 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
383 extern int sysctl_max_map_count;
385 #include <linux/aio.h>
387 #ifdef CONFIG_MMU
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
389 extern unsigned long
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
392 extern unsigned long
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
396 extern void arch_unmap_area(struct mm_struct *, unsigned long);
397 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
398 #else
399 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
400 #endif
403 extern void set_dumpable(struct mm_struct *mm, int value);
404 extern int get_dumpable(struct mm_struct *mm);
406 /* mm flags */
407 /* dumpable bits */
408 #define MMF_DUMPABLE 0 /* core dump is permitted */
409 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
411 #define MMF_DUMPABLE_BITS 2
412 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
414 /* coredump filter bits */
415 #define MMF_DUMP_ANON_PRIVATE 2
416 #define MMF_DUMP_ANON_SHARED 3
417 #define MMF_DUMP_MAPPED_PRIVATE 4
418 #define MMF_DUMP_MAPPED_SHARED 5
419 #define MMF_DUMP_ELF_HEADERS 6
420 #define MMF_DUMP_HUGETLB_PRIVATE 7
421 #define MMF_DUMP_HUGETLB_SHARED 8
423 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
424 #define MMF_DUMP_FILTER_BITS 7
425 #define MMF_DUMP_FILTER_MASK \
426 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
427 #define MMF_DUMP_FILTER_DEFAULT \
428 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
429 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
431 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
432 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
433 #else
434 # define MMF_DUMP_MASK_DEFAULT_ELF 0
435 #endif
436 /* leave room for more dump flags */
437 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
438 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
440 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
442 struct sighand_struct {
443 atomic_t count;
444 struct k_sigaction action[_NSIG];
445 spinlock_t siglock;
446 wait_queue_head_t signalfd_wqh;
449 struct pacct_struct {
450 int ac_flag;
451 long ac_exitcode;
452 unsigned long ac_mem;
453 cputime_t ac_utime, ac_stime;
454 unsigned long ac_minflt, ac_majflt;
457 struct cpu_itimer {
458 cputime_t expires;
459 cputime_t incr;
460 u32 error;
461 u32 incr_error;
465 * struct task_cputime - collected CPU time counts
466 * @utime: time spent in user mode, in &cputime_t units
467 * @stime: time spent in kernel mode, in &cputime_t units
468 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
470 * This structure groups together three kinds of CPU time that are
471 * tracked for threads and thread groups. Most things considering
472 * CPU time want to group these counts together and treat all three
473 * of them in parallel.
475 struct task_cputime {
476 cputime_t utime;
477 cputime_t stime;
478 unsigned long long sum_exec_runtime;
480 /* Alternate field names when used to cache expirations. */
481 #define prof_exp stime
482 #define virt_exp utime
483 #define sched_exp sum_exec_runtime
485 #define INIT_CPUTIME \
486 (struct task_cputime) { \
487 .utime = cputime_zero, \
488 .stime = cputime_zero, \
489 .sum_exec_runtime = 0, \
493 * Disable preemption until the scheduler is running.
494 * Reset by start_kernel()->sched_init()->init_idle().
496 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
497 * before the scheduler is active -- see should_resched().
499 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
502 * struct thread_group_cputimer - thread group interval timer counts
503 * @cputime: thread group interval timers.
504 * @running: non-zero when there are timers running and
505 * @cputime receives updates.
506 * @lock: lock for fields in this struct.
508 * This structure contains the version of task_cputime, above, that is
509 * used for thread group CPU timer calculations.
511 struct thread_group_cputimer {
512 struct task_cputime cputime;
513 int running;
514 spinlock_t lock;
517 struct autogroup;
520 * NOTE! "signal_struct" does not have its own
521 * locking, because a shared signal_struct always
522 * implies a shared sighand_struct, so locking
523 * sighand_struct is always a proper superset of
524 * the locking of signal_struct.
526 struct signal_struct {
527 atomic_t sigcnt;
528 atomic_t live;
529 int nr_threads;
531 wait_queue_head_t wait_chldexit; /* for wait4() */
533 /* current thread group signal load-balancing target: */
534 struct task_struct *curr_target;
536 /* shared signal handling: */
537 struct sigpending shared_pending;
539 /* thread group exit support */
540 int group_exit_code;
541 /* overloaded:
542 * - notify group_exit_task when ->count is equal to notify_count
543 * - everyone except group_exit_task is stopped during signal delivery
544 * of fatal signals, group_exit_task processes the signal.
546 int notify_count;
547 struct task_struct *group_exit_task;
549 /* thread group stop support, overloads group_exit_code too */
550 int group_stop_count;
551 unsigned int flags; /* see SIGNAL_* flags below */
553 /* POSIX.1b Interval Timers */
554 struct list_head posix_timers;
556 /* ITIMER_REAL timer for the process */
557 struct hrtimer real_timer;
558 struct pid *leader_pid;
559 ktime_t it_real_incr;
562 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
563 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
564 * values are defined to 0 and 1 respectively
566 struct cpu_itimer it[2];
569 * Thread group totals for process CPU timers.
570 * See thread_group_cputimer(), et al, for details.
572 struct thread_group_cputimer cputimer;
574 /* Earliest-expiration cache. */
575 struct task_cputime cputime_expires;
577 struct list_head cpu_timers[3];
579 struct pid *tty_old_pgrp;
581 /* boolean value for session group leader */
582 int leader;
584 struct tty_struct *tty; /* NULL if no tty */
586 #ifdef CONFIG_SCHED_AUTOGROUP
587 struct autogroup *autogroup;
588 #endif
590 * Cumulative resource counters for dead threads in the group,
591 * and for reaped dead child processes forked by this group.
592 * Live threads maintain their own counters and add to these
593 * in __exit_signal, except for the group leader.
595 cputime_t utime, stime, cutime, cstime;
596 cputime_t gtime;
597 cputime_t cgtime;
598 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
599 cputime_t prev_utime, prev_stime;
600 #endif
601 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
602 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
603 unsigned long inblock, oublock, cinblock, coublock;
604 unsigned long maxrss, cmaxrss;
605 struct task_io_accounting ioac;
608 * Cumulative ns of schedule CPU time fo dead threads in the
609 * group, not including a zombie group leader, (This only differs
610 * from jiffies_to_ns(utime + stime) if sched_clock uses something
611 * other than jiffies.)
613 unsigned long long sum_sched_runtime;
616 * We don't bother to synchronize most readers of this at all,
617 * because there is no reader checking a limit that actually needs
618 * to get both rlim_cur and rlim_max atomically, and either one
619 * alone is a single word that can safely be read normally.
620 * getrlimit/setrlimit use task_lock(current->group_leader) to
621 * protect this instead of the siglock, because they really
622 * have no need to disable irqs.
624 struct rlimit rlim[RLIM_NLIMITS];
626 #ifdef CONFIG_BSD_PROCESS_ACCT
627 struct pacct_struct pacct; /* per-process accounting information */
628 #endif
629 #ifdef CONFIG_TASKSTATS
630 struct taskstats *stats;
631 #endif
632 #ifdef CONFIG_AUDIT
633 unsigned audit_tty;
634 struct tty_audit_buf *tty_audit_buf;
635 #endif
637 int oom_adj; /* OOM kill score adjustment (bit shift) */
638 int oom_score_adj; /* OOM kill score adjustment */
639 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
640 * Only settable by CAP_SYS_RESOURCE. */
642 struct mutex cred_guard_mutex; /* guard against foreign influences on
643 * credential calculations
644 * (notably. ptrace) */
647 /* Context switch must be unlocked if interrupts are to be enabled */
648 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
649 # define __ARCH_WANT_UNLOCKED_CTXSW
650 #endif
653 * Bits in flags field of signal_struct.
655 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
656 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
657 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
658 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
660 * Pending notifications to parent.
662 #define SIGNAL_CLD_STOPPED 0x00000010
663 #define SIGNAL_CLD_CONTINUED 0x00000020
664 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
666 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
668 /* If true, all threads except ->group_exit_task have pending SIGKILL */
669 static inline int signal_group_exit(const struct signal_struct *sig)
671 return (sig->flags & SIGNAL_GROUP_EXIT) ||
672 (sig->group_exit_task != NULL);
676 * Some day this will be a full-fledged user tracking system..
678 struct user_struct {
679 atomic_t __count; /* reference count */
680 atomic_t processes; /* How many processes does this user have? */
681 atomic_t files; /* How many open files does this user have? */
682 atomic_t sigpending; /* How many pending signals does this user have? */
683 #ifdef CONFIG_INOTIFY_USER
684 atomic_t inotify_watches; /* How many inotify watches does this user have? */
685 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
686 #endif
687 #ifdef CONFIG_FANOTIFY
688 atomic_t fanotify_listeners;
689 #endif
690 #ifdef CONFIG_EPOLL
691 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
692 #endif
693 #ifdef CONFIG_POSIX_MQUEUE
694 /* protected by mq_lock */
695 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
696 #endif
697 unsigned long locked_shm; /* How many pages of mlocked shm ? */
699 #ifdef CONFIG_KEYS
700 struct key *uid_keyring; /* UID specific keyring */
701 struct key *session_keyring; /* UID's default session keyring */
702 #endif
704 /* Hash table maintenance information */
705 struct hlist_node uidhash_node;
706 uid_t uid;
707 struct user_namespace *user_ns;
709 #ifdef CONFIG_PERF_EVENTS
710 atomic_long_t locked_vm;
711 #endif
714 extern int uids_sysfs_init(void);
716 extern struct user_struct *find_user(uid_t);
718 extern struct user_struct root_user;
719 #define INIT_USER (&root_user)
722 struct backing_dev_info;
723 struct reclaim_state;
725 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
726 struct sched_info {
727 /* cumulative counters */
728 unsigned long pcount; /* # of times run on this cpu */
729 unsigned long long run_delay; /* time spent waiting on a runqueue */
731 /* timestamps */
732 unsigned long long last_arrival,/* when we last ran on a cpu */
733 last_queued; /* when we were last queued to run */
735 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
737 #ifdef CONFIG_TASK_DELAY_ACCT
738 struct task_delay_info {
739 spinlock_t lock;
740 unsigned int flags; /* Private per-task flags */
742 /* For each stat XXX, add following, aligned appropriately
744 * struct timespec XXX_start, XXX_end;
745 * u64 XXX_delay;
746 * u32 XXX_count;
748 * Atomicity of updates to XXX_delay, XXX_count protected by
749 * single lock above (split into XXX_lock if contention is an issue).
753 * XXX_count is incremented on every XXX operation, the delay
754 * associated with the operation is added to XXX_delay.
755 * XXX_delay contains the accumulated delay time in nanoseconds.
757 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
758 u64 blkio_delay; /* wait for sync block io completion */
759 u64 swapin_delay; /* wait for swapin block io completion */
760 u32 blkio_count; /* total count of the number of sync block */
761 /* io operations performed */
762 u32 swapin_count; /* total count of the number of swapin block */
763 /* io operations performed */
765 struct timespec freepages_start, freepages_end;
766 u64 freepages_delay; /* wait for memory reclaim */
767 u32 freepages_count; /* total count of memory reclaim */
769 #endif /* CONFIG_TASK_DELAY_ACCT */
771 static inline int sched_info_on(void)
773 #ifdef CONFIG_SCHEDSTATS
774 return 1;
775 #elif defined(CONFIG_TASK_DELAY_ACCT)
776 extern int delayacct_on;
777 return delayacct_on;
778 #else
779 return 0;
780 #endif
783 enum cpu_idle_type {
784 CPU_IDLE,
785 CPU_NOT_IDLE,
786 CPU_NEWLY_IDLE,
787 CPU_MAX_IDLE_TYPES
791 * sched-domains (multiprocessor balancing) declarations:
795 * Increase resolution of nice-level calculations:
797 #define SCHED_LOAD_SHIFT 10
798 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
800 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
802 #ifdef CONFIG_SMP
803 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
804 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
805 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
806 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
807 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
808 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
809 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
810 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
811 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
812 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
813 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
814 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
815 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
817 enum powersavings_balance_level {
818 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
819 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
820 * first for long running threads
822 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
823 * cpu package for power savings
825 MAX_POWERSAVINGS_BALANCE_LEVELS
828 extern int sched_mc_power_savings, sched_smt_power_savings;
830 static inline int sd_balance_for_mc_power(void)
832 if (sched_smt_power_savings)
833 return SD_POWERSAVINGS_BALANCE;
835 if (!sched_mc_power_savings)
836 return SD_PREFER_SIBLING;
838 return 0;
841 static inline int sd_balance_for_package_power(void)
843 if (sched_mc_power_savings | sched_smt_power_savings)
844 return SD_POWERSAVINGS_BALANCE;
846 return SD_PREFER_SIBLING;
849 extern int __weak arch_sd_sibiling_asym_packing(void);
852 * Optimise SD flags for power savings:
853 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
854 * Keep default SD flags if sched_{smt,mc}_power_saving=0
857 static inline int sd_power_saving_flags(void)
859 if (sched_mc_power_savings | sched_smt_power_savings)
860 return SD_BALANCE_NEWIDLE;
862 return 0;
865 struct sched_group {
866 struct sched_group *next; /* Must be a circular list */
867 atomic_t ref;
870 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
871 * single CPU.
873 unsigned int cpu_power, cpu_power_orig;
874 unsigned int group_weight;
877 * The CPUs this group covers.
879 * NOTE: this field is variable length. (Allocated dynamically
880 * by attaching extra space to the end of the structure,
881 * depending on how many CPUs the kernel has booted up with)
883 unsigned long cpumask[0];
886 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
888 return to_cpumask(sg->cpumask);
891 struct sched_domain_attr {
892 int relax_domain_level;
895 #define SD_ATTR_INIT (struct sched_domain_attr) { \
896 .relax_domain_level = -1, \
899 extern int sched_domain_level_max;
901 struct sched_domain {
902 /* These fields must be setup */
903 struct sched_domain *parent; /* top domain must be null terminated */
904 struct sched_domain *child; /* bottom domain must be null terminated */
905 struct sched_group *groups; /* the balancing groups of the domain */
906 unsigned long min_interval; /* Minimum balance interval ms */
907 unsigned long max_interval; /* Maximum balance interval ms */
908 unsigned int busy_factor; /* less balancing by factor if busy */
909 unsigned int imbalance_pct; /* No balance until over watermark */
910 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
911 unsigned int busy_idx;
912 unsigned int idle_idx;
913 unsigned int newidle_idx;
914 unsigned int wake_idx;
915 unsigned int forkexec_idx;
916 unsigned int smt_gain;
917 int flags; /* See SD_* */
918 int level;
920 /* Runtime fields. */
921 unsigned long last_balance; /* init to jiffies. units in jiffies */
922 unsigned int balance_interval; /* initialise to 1. units in ms. */
923 unsigned int nr_balance_failed; /* initialise to 0 */
925 u64 last_update;
927 #ifdef CONFIG_SCHEDSTATS
928 /* load_balance() stats */
929 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
930 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
931 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
932 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
933 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
934 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
935 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
936 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
938 /* Active load balancing */
939 unsigned int alb_count;
940 unsigned int alb_failed;
941 unsigned int alb_pushed;
943 /* SD_BALANCE_EXEC stats */
944 unsigned int sbe_count;
945 unsigned int sbe_balanced;
946 unsigned int sbe_pushed;
948 /* SD_BALANCE_FORK stats */
949 unsigned int sbf_count;
950 unsigned int sbf_balanced;
951 unsigned int sbf_pushed;
953 /* try_to_wake_up() stats */
954 unsigned int ttwu_wake_remote;
955 unsigned int ttwu_move_affine;
956 unsigned int ttwu_move_balance;
957 #endif
958 #ifdef CONFIG_SCHED_DEBUG
959 char *name;
960 #endif
961 union {
962 void *private; /* used during construction */
963 struct rcu_head rcu; /* used during destruction */
966 unsigned int span_weight;
968 * Span of all CPUs in this domain.
970 * NOTE: this field is variable length. (Allocated dynamically
971 * by attaching extra space to the end of the structure,
972 * depending on how many CPUs the kernel has booted up with)
974 unsigned long span[0];
977 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
979 return to_cpumask(sd->span);
982 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
983 struct sched_domain_attr *dattr_new);
985 /* Allocate an array of sched domains, for partition_sched_domains(). */
986 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
987 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
989 /* Test a flag in parent sched domain */
990 static inline int test_sd_parent(struct sched_domain *sd, int flag)
992 if (sd->parent && (sd->parent->flags & flag))
993 return 1;
995 return 0;
998 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
999 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1001 #else /* CONFIG_SMP */
1003 struct sched_domain_attr;
1005 static inline void
1006 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1007 struct sched_domain_attr *dattr_new)
1010 #endif /* !CONFIG_SMP */
1013 struct io_context; /* See blkdev.h */
1016 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1017 extern void prefetch_stack(struct task_struct *t);
1018 #else
1019 static inline void prefetch_stack(struct task_struct *t) { }
1020 #endif
1022 struct audit_context; /* See audit.c */
1023 struct mempolicy;
1024 struct pipe_inode_info;
1025 struct uts_namespace;
1027 struct rq;
1028 struct sched_domain;
1031 * wake flags
1033 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1034 #define WF_FORK 0x02 /* child wakeup after fork */
1036 #define ENQUEUE_WAKEUP 1
1037 #define ENQUEUE_HEAD 2
1038 #ifdef CONFIG_SMP
1039 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1040 #else
1041 #define ENQUEUE_WAKING 0
1042 #endif
1044 #define DEQUEUE_SLEEP 1
1046 struct sched_class {
1047 const struct sched_class *next;
1049 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1050 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1051 void (*yield_task) (struct rq *rq);
1052 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1054 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1056 struct task_struct * (*pick_next_task) (struct rq *rq);
1057 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1059 #ifdef CONFIG_SMP
1060 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1062 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1063 void (*post_schedule) (struct rq *this_rq);
1064 void (*task_waking) (struct task_struct *task);
1065 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1067 void (*set_cpus_allowed)(struct task_struct *p,
1068 const struct cpumask *newmask);
1070 void (*rq_online)(struct rq *rq);
1071 void (*rq_offline)(struct rq *rq);
1072 #endif
1074 void (*set_curr_task) (struct rq *rq);
1075 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1076 void (*task_fork) (struct task_struct *p);
1078 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1079 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1080 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1081 int oldprio);
1083 unsigned int (*get_rr_interval) (struct rq *rq,
1084 struct task_struct *task);
1086 #ifdef CONFIG_FAIR_GROUP_SCHED
1087 void (*task_move_group) (struct task_struct *p, int on_rq);
1088 #endif
1091 struct load_weight {
1092 unsigned long weight, inv_weight;
1095 #ifdef CONFIG_SCHEDSTATS
1096 struct sched_statistics {
1097 u64 wait_start;
1098 u64 wait_max;
1099 u64 wait_count;
1100 u64 wait_sum;
1101 u64 iowait_count;
1102 u64 iowait_sum;
1104 u64 sleep_start;
1105 u64 sleep_max;
1106 s64 sum_sleep_runtime;
1108 u64 block_start;
1109 u64 block_max;
1110 u64 exec_max;
1111 u64 slice_max;
1113 u64 nr_migrations_cold;
1114 u64 nr_failed_migrations_affine;
1115 u64 nr_failed_migrations_running;
1116 u64 nr_failed_migrations_hot;
1117 u64 nr_forced_migrations;
1119 u64 nr_wakeups;
1120 u64 nr_wakeups_sync;
1121 u64 nr_wakeups_migrate;
1122 u64 nr_wakeups_local;
1123 u64 nr_wakeups_remote;
1124 u64 nr_wakeups_affine;
1125 u64 nr_wakeups_affine_attempts;
1126 u64 nr_wakeups_passive;
1127 u64 nr_wakeups_idle;
1129 #endif
1131 struct sched_entity {
1132 struct load_weight load; /* for load-balancing */
1133 struct rb_node run_node;
1134 struct list_head group_node;
1135 unsigned int on_rq;
1137 u64 exec_start;
1138 u64 sum_exec_runtime;
1139 u64 vruntime;
1140 u64 prev_sum_exec_runtime;
1142 u64 nr_migrations;
1144 #ifdef CONFIG_SCHEDSTATS
1145 struct sched_statistics statistics;
1146 #endif
1148 #ifdef CONFIG_FAIR_GROUP_SCHED
1149 struct sched_entity *parent;
1150 /* rq on which this entity is (to be) queued: */
1151 struct cfs_rq *cfs_rq;
1152 /* rq "owned" by this entity/group: */
1153 struct cfs_rq *my_q;
1154 #endif
1157 struct sched_rt_entity {
1158 struct list_head run_list;
1159 unsigned long timeout;
1160 unsigned int time_slice;
1161 int nr_cpus_allowed;
1163 struct sched_rt_entity *back;
1164 #ifdef CONFIG_RT_GROUP_SCHED
1165 struct sched_rt_entity *parent;
1166 /* rq on which this entity is (to be) queued: */
1167 struct rt_rq *rt_rq;
1168 /* rq "owned" by this entity/group: */
1169 struct rt_rq *my_q;
1170 #endif
1173 struct rcu_node;
1175 enum perf_event_task_context {
1176 perf_invalid_context = -1,
1177 perf_hw_context = 0,
1178 perf_sw_context,
1179 perf_nr_task_contexts,
1182 struct task_struct {
1183 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1184 void *stack;
1185 atomic_t usage;
1186 unsigned int flags; /* per process flags, defined below */
1187 unsigned int ptrace;
1189 #ifdef CONFIG_SMP
1190 struct task_struct *wake_entry;
1191 int on_cpu;
1192 #endif
1193 int on_rq;
1195 int prio, static_prio, normal_prio;
1196 unsigned int rt_priority;
1197 const struct sched_class *sched_class;
1198 struct sched_entity se;
1199 struct sched_rt_entity rt;
1201 #ifdef CONFIG_PREEMPT_NOTIFIERS
1202 /* list of struct preempt_notifier: */
1203 struct hlist_head preempt_notifiers;
1204 #endif
1207 * fpu_counter contains the number of consecutive context switches
1208 * that the FPU is used. If this is over a threshold, the lazy fpu
1209 * saving becomes unlazy to save the trap. This is an unsigned char
1210 * so that after 256 times the counter wraps and the behavior turns
1211 * lazy again; this to deal with bursty apps that only use FPU for
1212 * a short time
1214 unsigned char fpu_counter;
1215 #ifdef CONFIG_BLK_DEV_IO_TRACE
1216 unsigned int btrace_seq;
1217 #endif
1219 unsigned int policy;
1220 cpumask_t cpus_allowed;
1222 #ifdef CONFIG_PREEMPT_RCU
1223 int rcu_read_lock_nesting;
1224 char rcu_read_unlock_special;
1225 struct list_head rcu_node_entry;
1226 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1227 #ifdef CONFIG_TREE_PREEMPT_RCU
1228 struct rcu_node *rcu_blocked_node;
1229 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1230 #ifdef CONFIG_RCU_BOOST
1231 struct rt_mutex *rcu_boost_mutex;
1232 #endif /* #ifdef CONFIG_RCU_BOOST */
1234 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1235 struct sched_info sched_info;
1236 #endif
1238 struct list_head tasks;
1239 #ifdef CONFIG_SMP
1240 struct plist_node pushable_tasks;
1241 #endif
1243 struct mm_struct *mm, *active_mm;
1244 #ifdef CONFIG_COMPAT_BRK
1245 unsigned brk_randomized:1;
1246 #endif
1247 #if defined(SPLIT_RSS_COUNTING)
1248 struct task_rss_stat rss_stat;
1249 #endif
1250 /* task state */
1251 int exit_state;
1252 int exit_code, exit_signal;
1253 int pdeath_signal; /* The signal sent when the parent dies */
1254 /* ??? */
1255 unsigned int personality;
1256 unsigned did_exec:1;
1257 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1258 * execve */
1259 unsigned in_iowait:1;
1262 /* Revert to default priority/policy when forking */
1263 unsigned sched_reset_on_fork:1;
1264 unsigned sched_contributes_to_load:1;
1266 pid_t pid;
1267 pid_t tgid;
1269 #ifdef CONFIG_CC_STACKPROTECTOR
1270 /* Canary value for the -fstack-protector gcc feature */
1271 unsigned long stack_canary;
1272 #endif
1275 * pointers to (original) parent process, youngest child, younger sibling,
1276 * older sibling, respectively. (p->father can be replaced with
1277 * p->real_parent->pid)
1279 struct task_struct *real_parent; /* real parent process */
1280 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1282 * children/sibling forms the list of my natural children
1284 struct list_head children; /* list of my children */
1285 struct list_head sibling; /* linkage in my parent's children list */
1286 struct task_struct *group_leader; /* threadgroup leader */
1289 * ptraced is the list of tasks this task is using ptrace on.
1290 * This includes both natural children and PTRACE_ATTACH targets.
1291 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1293 struct list_head ptraced;
1294 struct list_head ptrace_entry;
1296 /* PID/PID hash table linkage. */
1297 struct pid_link pids[PIDTYPE_MAX];
1298 struct list_head thread_group;
1300 struct completion *vfork_done; /* for vfork() */
1301 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1302 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1304 cputime_t utime, stime, utimescaled, stimescaled;
1305 cputime_t gtime;
1306 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1307 cputime_t prev_utime, prev_stime;
1308 #endif
1309 unsigned long nvcsw, nivcsw; /* context switch counts */
1310 struct timespec start_time; /* monotonic time */
1311 struct timespec real_start_time; /* boot based time */
1312 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1313 unsigned long min_flt, maj_flt;
1315 struct task_cputime cputime_expires;
1316 struct list_head cpu_timers[3];
1318 /* process credentials */
1319 const struct cred __rcu *real_cred; /* objective and real subjective task
1320 * credentials (COW) */
1321 const struct cred __rcu *cred; /* effective (overridable) subjective task
1322 * credentials (COW) */
1323 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1325 char comm[TASK_COMM_LEN]; /* executable name excluding path
1326 - access with [gs]et_task_comm (which lock
1327 it with task_lock())
1328 - initialized normally by setup_new_exec */
1329 /* file system info */
1330 int link_count, total_link_count;
1331 #ifdef CONFIG_SYSVIPC
1332 /* ipc stuff */
1333 struct sysv_sem sysvsem;
1334 #endif
1335 #ifdef CONFIG_DETECT_HUNG_TASK
1336 /* hung task detection */
1337 unsigned long last_switch_count;
1338 #endif
1339 /* CPU-specific state of this task */
1340 struct thread_struct thread;
1341 /* filesystem information */
1342 struct fs_struct *fs;
1343 /* open file information */
1344 struct files_struct *files;
1345 /* namespaces */
1346 struct nsproxy *nsproxy;
1347 /* signal handlers */
1348 struct signal_struct *signal;
1349 struct sighand_struct *sighand;
1351 sigset_t blocked, real_blocked;
1352 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1353 struct sigpending pending;
1355 unsigned long sas_ss_sp;
1356 size_t sas_ss_size;
1357 int (*notifier)(void *priv);
1358 void *notifier_data;
1359 sigset_t *notifier_mask;
1360 struct audit_context *audit_context;
1361 #ifdef CONFIG_AUDITSYSCALL
1362 uid_t loginuid;
1363 unsigned int sessionid;
1364 #endif
1365 seccomp_t seccomp;
1367 /* Thread group tracking */
1368 u32 parent_exec_id;
1369 u32 self_exec_id;
1370 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1371 * mempolicy */
1372 spinlock_t alloc_lock;
1374 #ifdef CONFIG_GENERIC_HARDIRQS
1375 /* IRQ handler threads */
1376 struct irqaction *irqaction;
1377 #endif
1379 /* Protection of the PI data structures: */
1380 raw_spinlock_t pi_lock;
1382 #ifdef CONFIG_RT_MUTEXES
1383 /* PI waiters blocked on a rt_mutex held by this task */
1384 struct plist_head pi_waiters;
1385 /* Deadlock detection and priority inheritance handling */
1386 struct rt_mutex_waiter *pi_blocked_on;
1387 #endif
1389 #ifdef CONFIG_DEBUG_MUTEXES
1390 /* mutex deadlock detection */
1391 struct mutex_waiter *blocked_on;
1392 #endif
1393 #ifdef CONFIG_TRACE_IRQFLAGS
1394 unsigned int irq_events;
1395 unsigned long hardirq_enable_ip;
1396 unsigned long hardirq_disable_ip;
1397 unsigned int hardirq_enable_event;
1398 unsigned int hardirq_disable_event;
1399 int hardirqs_enabled;
1400 int hardirq_context;
1401 unsigned long softirq_disable_ip;
1402 unsigned long softirq_enable_ip;
1403 unsigned int softirq_disable_event;
1404 unsigned int softirq_enable_event;
1405 int softirqs_enabled;
1406 int softirq_context;
1407 #endif
1408 #ifdef CONFIG_LOCKDEP
1409 # define MAX_LOCK_DEPTH 48UL
1410 u64 curr_chain_key;
1411 int lockdep_depth;
1412 unsigned int lockdep_recursion;
1413 struct held_lock held_locks[MAX_LOCK_DEPTH];
1414 gfp_t lockdep_reclaim_gfp;
1415 #endif
1417 /* journalling filesystem info */
1418 void *journal_info;
1420 /* stacked block device info */
1421 struct bio_list *bio_list;
1423 #ifdef CONFIG_BLOCK
1424 /* stack plugging */
1425 struct blk_plug *plug;
1426 #endif
1428 /* VM state */
1429 struct reclaim_state *reclaim_state;
1431 struct backing_dev_info *backing_dev_info;
1433 struct io_context *io_context;
1435 unsigned long ptrace_message;
1436 siginfo_t *last_siginfo; /* For ptrace use. */
1437 struct task_io_accounting ioac;
1438 #if defined(CONFIG_TASK_XACCT)
1439 u64 acct_rss_mem1; /* accumulated rss usage */
1440 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1441 cputime_t acct_timexpd; /* stime + utime since last update */
1442 #endif
1443 #ifdef CONFIG_CPUSETS
1444 nodemask_t mems_allowed; /* Protected by alloc_lock */
1445 int mems_allowed_change_disable;
1446 int cpuset_mem_spread_rotor;
1447 int cpuset_slab_spread_rotor;
1448 #endif
1449 #ifdef CONFIG_CGROUPS
1450 /* Control Group info protected by css_set_lock */
1451 struct css_set __rcu *cgroups;
1452 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1453 struct list_head cg_list;
1454 #endif
1455 #ifdef CONFIG_FUTEX
1456 struct robust_list_head __user *robust_list;
1457 #ifdef CONFIG_COMPAT
1458 struct compat_robust_list_head __user *compat_robust_list;
1459 #endif
1460 struct list_head pi_state_list;
1461 struct futex_pi_state *pi_state_cache;
1462 #endif
1463 #ifdef CONFIG_PERF_EVENTS
1464 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1465 struct mutex perf_event_mutex;
1466 struct list_head perf_event_list;
1467 #endif
1468 #ifdef CONFIG_NUMA
1469 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1470 short il_next;
1471 short pref_node_fork;
1472 #endif
1473 atomic_t fs_excl; /* holding fs exclusive resources */
1474 struct rcu_head rcu;
1477 * cache last used pipe for splice
1479 struct pipe_inode_info *splice_pipe;
1480 #ifdef CONFIG_TASK_DELAY_ACCT
1481 struct task_delay_info *delays;
1482 #endif
1483 #ifdef CONFIG_FAULT_INJECTION
1484 int make_it_fail;
1485 #endif
1486 struct prop_local_single dirties;
1487 #ifdef CONFIG_LATENCYTOP
1488 int latency_record_count;
1489 struct latency_record latency_record[LT_SAVECOUNT];
1490 #endif
1492 * time slack values; these are used to round up poll() and
1493 * select() etc timeout values. These are in nanoseconds.
1495 unsigned long timer_slack_ns;
1496 unsigned long default_timer_slack_ns;
1498 struct list_head *scm_work_list;
1499 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1500 /* Index of current stored address in ret_stack */
1501 int curr_ret_stack;
1502 /* Stack of return addresses for return function tracing */
1503 struct ftrace_ret_stack *ret_stack;
1504 /* time stamp for last schedule */
1505 unsigned long long ftrace_timestamp;
1507 * Number of functions that haven't been traced
1508 * because of depth overrun.
1510 atomic_t trace_overrun;
1511 /* Pause for the tracing */
1512 atomic_t tracing_graph_pause;
1513 #endif
1514 #ifdef CONFIG_TRACING
1515 /* state flags for use by tracers */
1516 unsigned long trace;
1517 /* bitmask of trace recursion */
1518 unsigned long trace_recursion;
1519 #endif /* CONFIG_TRACING */
1520 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1521 struct memcg_batch_info {
1522 int do_batch; /* incremented when batch uncharge started */
1523 struct mem_cgroup *memcg; /* target memcg of uncharge */
1524 unsigned long nr_pages; /* uncharged usage */
1525 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1526 } memcg_batch;
1527 #endif
1528 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1529 atomic_t ptrace_bp_refcnt;
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_STARTING 0x00000002 /* being created */
1719 #define PF_EXITING 0x00000004 /* getting shut down */
1720 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1721 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1722 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1723 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1724 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1725 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1726 #define PF_DUMPCORE 0x00000200 /* dumped core */
1727 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1728 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1729 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1730 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1731 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1732 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1733 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1734 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1735 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1736 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1737 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1738 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1739 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1740 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1741 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1742 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1743 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1744 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1745 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1746 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1747 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1750 * Only the _current_ task can read/write to tsk->flags, but other
1751 * tasks can access tsk->flags in readonly mode for example
1752 * with tsk_used_math (like during threaded core dumping).
1753 * There is however an exception to this rule during ptrace
1754 * or during fork: the ptracer task is allowed to write to the
1755 * child->flags of its traced child (same goes for fork, the parent
1756 * can write to the child->flags), because we're guaranteed the
1757 * child is not running and in turn not changing child->flags
1758 * at the same time the parent does it.
1760 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1761 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1762 #define clear_used_math() clear_stopped_child_used_math(current)
1763 #define set_used_math() set_stopped_child_used_math(current)
1764 #define conditional_stopped_child_used_math(condition, child) \
1765 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1766 #define conditional_used_math(condition) \
1767 conditional_stopped_child_used_math(condition, current)
1768 #define copy_to_stopped_child_used_math(child) \
1769 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1770 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1771 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1772 #define used_math() tsk_used_math(current)
1774 #ifdef CONFIG_PREEMPT_RCU
1776 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1777 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1778 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1780 static inline void rcu_copy_process(struct task_struct *p)
1782 p->rcu_read_lock_nesting = 0;
1783 p->rcu_read_unlock_special = 0;
1784 #ifdef CONFIG_TREE_PREEMPT_RCU
1785 p->rcu_blocked_node = NULL;
1786 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1787 #ifdef CONFIG_RCU_BOOST
1788 p->rcu_boost_mutex = NULL;
1789 #endif /* #ifdef CONFIG_RCU_BOOST */
1790 INIT_LIST_HEAD(&p->rcu_node_entry);
1793 #else
1795 static inline void rcu_copy_process(struct task_struct *p)
1799 #endif
1801 #ifdef CONFIG_SMP
1802 extern int set_cpus_allowed_ptr(struct task_struct *p,
1803 const struct cpumask *new_mask);
1804 #else
1805 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1806 const struct cpumask *new_mask)
1808 if (!cpumask_test_cpu(0, new_mask))
1809 return -EINVAL;
1810 return 0;
1812 #endif
1814 #ifndef CONFIG_CPUMASK_OFFSTACK
1815 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1817 return set_cpus_allowed_ptr(p, &new_mask);
1819 #endif
1822 * Do not use outside of architecture code which knows its limitations.
1824 * sched_clock() has no promise of monotonicity or bounded drift between
1825 * CPUs, use (which you should not) requires disabling IRQs.
1827 * Please use one of the three interfaces below.
1829 extern unsigned long long notrace sched_clock(void);
1831 * See the comment in kernel/sched_clock.c
1833 extern u64 cpu_clock(int cpu);
1834 extern u64 local_clock(void);
1835 extern u64 sched_clock_cpu(int cpu);
1838 extern void sched_clock_init(void);
1840 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1841 static inline void sched_clock_tick(void)
1845 static inline void sched_clock_idle_sleep_event(void)
1849 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1852 #else
1854 * Architectures can set this to 1 if they have specified
1855 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1856 * but then during bootup it turns out that sched_clock()
1857 * is reliable after all:
1859 extern int sched_clock_stable;
1861 extern void sched_clock_tick(void);
1862 extern void sched_clock_idle_sleep_event(void);
1863 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1864 #endif
1866 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1868 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1869 * The reason for this explicit opt-in is not to have perf penalty with
1870 * slow sched_clocks.
1872 extern void enable_sched_clock_irqtime(void);
1873 extern void disable_sched_clock_irqtime(void);
1874 #else
1875 static inline void enable_sched_clock_irqtime(void) {}
1876 static inline void disable_sched_clock_irqtime(void) {}
1877 #endif
1879 extern unsigned long long
1880 task_sched_runtime(struct task_struct *task);
1881 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1883 /* sched_exec is called by processes performing an exec */
1884 #ifdef CONFIG_SMP
1885 extern void sched_exec(void);
1886 #else
1887 #define sched_exec() {}
1888 #endif
1890 extern void sched_clock_idle_sleep_event(void);
1891 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1893 #ifdef CONFIG_HOTPLUG_CPU
1894 extern void idle_task_exit(void);
1895 #else
1896 static inline void idle_task_exit(void) {}
1897 #endif
1899 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1900 extern void wake_up_idle_cpu(int cpu);
1901 #else
1902 static inline void wake_up_idle_cpu(int cpu) { }
1903 #endif
1905 extern unsigned int sysctl_sched_latency;
1906 extern unsigned int sysctl_sched_min_granularity;
1907 extern unsigned int sysctl_sched_wakeup_granularity;
1908 extern unsigned int sysctl_sched_child_runs_first;
1910 enum sched_tunable_scaling {
1911 SCHED_TUNABLESCALING_NONE,
1912 SCHED_TUNABLESCALING_LOG,
1913 SCHED_TUNABLESCALING_LINEAR,
1914 SCHED_TUNABLESCALING_END,
1916 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1918 #ifdef CONFIG_SCHED_DEBUG
1919 extern unsigned int sysctl_sched_migration_cost;
1920 extern unsigned int sysctl_sched_nr_migrate;
1921 extern unsigned int sysctl_sched_time_avg;
1922 extern unsigned int sysctl_timer_migration;
1923 extern unsigned int sysctl_sched_shares_window;
1925 int sched_proc_update_handler(struct ctl_table *table, int write,
1926 void __user *buffer, size_t *length,
1927 loff_t *ppos);
1928 #endif
1929 #ifdef CONFIG_SCHED_DEBUG
1930 static inline unsigned int get_sysctl_timer_migration(void)
1932 return sysctl_timer_migration;
1934 #else
1935 static inline unsigned int get_sysctl_timer_migration(void)
1937 return 1;
1939 #endif
1940 extern unsigned int sysctl_sched_rt_period;
1941 extern int sysctl_sched_rt_runtime;
1943 int sched_rt_handler(struct ctl_table *table, int write,
1944 void __user *buffer, size_t *lenp,
1945 loff_t *ppos);
1947 #ifdef CONFIG_SCHED_AUTOGROUP
1948 extern unsigned int sysctl_sched_autogroup_enabled;
1950 extern void sched_autogroup_create_attach(struct task_struct *p);
1951 extern void sched_autogroup_detach(struct task_struct *p);
1952 extern void sched_autogroup_fork(struct signal_struct *sig);
1953 extern void sched_autogroup_exit(struct signal_struct *sig);
1954 #ifdef CONFIG_PROC_FS
1955 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1956 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1957 #endif
1958 #else
1959 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1960 static inline void sched_autogroup_detach(struct task_struct *p) { }
1961 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1962 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1963 #endif
1965 #ifdef CONFIG_RT_MUTEXES
1966 extern int rt_mutex_getprio(struct task_struct *p);
1967 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1968 extern void rt_mutex_adjust_pi(struct task_struct *p);
1969 #else
1970 static inline int rt_mutex_getprio(struct task_struct *p)
1972 return p->normal_prio;
1974 # define rt_mutex_adjust_pi(p) do { } while (0)
1975 #endif
1977 extern bool yield_to(struct task_struct *p, bool preempt);
1978 extern void set_user_nice(struct task_struct *p, long nice);
1979 extern int task_prio(const struct task_struct *p);
1980 extern int task_nice(const struct task_struct *p);
1981 extern int can_nice(const struct task_struct *p, const int nice);
1982 extern int task_curr(const struct task_struct *p);
1983 extern int idle_cpu(int cpu);
1984 extern int sched_setscheduler(struct task_struct *, int,
1985 const struct sched_param *);
1986 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1987 const struct sched_param *);
1988 extern struct task_struct *idle_task(int cpu);
1989 extern struct task_struct *curr_task(int cpu);
1990 extern void set_curr_task(int cpu, struct task_struct *p);
1992 void yield(void);
1995 * The default (Linux) execution domain.
1997 extern struct exec_domain default_exec_domain;
1999 union thread_union {
2000 struct thread_info thread_info;
2001 unsigned long stack[THREAD_SIZE/sizeof(long)];
2004 #ifndef __HAVE_ARCH_KSTACK_END
2005 static inline int kstack_end(void *addr)
2007 /* Reliable end of stack detection:
2008 * Some APM bios versions misalign the stack
2010 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2012 #endif
2014 extern union thread_union init_thread_union;
2015 extern struct task_struct init_task;
2017 extern struct mm_struct init_mm;
2019 extern struct pid_namespace init_pid_ns;
2022 * find a task by one of its numerical ids
2024 * find_task_by_pid_ns():
2025 * finds a task by its pid in the specified namespace
2026 * find_task_by_vpid():
2027 * finds a task by its virtual pid
2029 * see also find_vpid() etc in include/linux/pid.h
2032 extern struct task_struct *find_task_by_vpid(pid_t nr);
2033 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2034 struct pid_namespace *ns);
2036 extern void __set_special_pids(struct pid *pid);
2038 /* per-UID process charging. */
2039 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2040 static inline struct user_struct *get_uid(struct user_struct *u)
2042 atomic_inc(&u->__count);
2043 return u;
2045 extern void free_uid(struct user_struct *);
2046 extern void release_uids(struct user_namespace *ns);
2048 #include <asm/current.h>
2050 extern void xtime_update(unsigned long ticks);
2052 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2053 extern int wake_up_process(struct task_struct *tsk);
2054 extern void wake_up_new_task(struct task_struct *tsk);
2055 #ifdef CONFIG_SMP
2056 extern void kick_process(struct task_struct *tsk);
2057 #else
2058 static inline void kick_process(struct task_struct *tsk) { }
2059 #endif
2060 extern void sched_fork(struct task_struct *p);
2061 extern void sched_dead(struct task_struct *p);
2063 extern void proc_caches_init(void);
2064 extern void flush_signals(struct task_struct *);
2065 extern void __flush_signals(struct task_struct *);
2066 extern void ignore_signals(struct task_struct *);
2067 extern void flush_signal_handlers(struct task_struct *, int force_default);
2068 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2070 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2072 unsigned long flags;
2073 int ret;
2075 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2076 ret = dequeue_signal(tsk, mask, info);
2077 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2079 return ret;
2082 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2083 sigset_t *mask);
2084 extern void unblock_all_signals(void);
2085 extern void release_task(struct task_struct * p);
2086 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2087 extern int force_sigsegv(int, struct task_struct *);
2088 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2089 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2090 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2091 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2092 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2093 extern int kill_pid(struct pid *pid, int sig, int priv);
2094 extern int kill_proc_info(int, struct siginfo *, pid_t);
2095 extern int do_notify_parent(struct task_struct *, int);
2096 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2097 extern void force_sig(int, struct task_struct *);
2098 extern int send_sig(int, struct task_struct *, int);
2099 extern int zap_other_threads(struct task_struct *p);
2100 extern struct sigqueue *sigqueue_alloc(void);
2101 extern void sigqueue_free(struct sigqueue *);
2102 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2103 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2104 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2106 static inline int kill_cad_pid(int sig, int priv)
2108 return kill_pid(cad_pid, sig, priv);
2111 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2112 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2113 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2114 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2117 * True if we are on the alternate signal stack.
2119 static inline int on_sig_stack(unsigned long sp)
2121 #ifdef CONFIG_STACK_GROWSUP
2122 return sp >= current->sas_ss_sp &&
2123 sp - current->sas_ss_sp < current->sas_ss_size;
2124 #else
2125 return sp > current->sas_ss_sp &&
2126 sp - current->sas_ss_sp <= current->sas_ss_size;
2127 #endif
2130 static inline int sas_ss_flags(unsigned long sp)
2132 return (current->sas_ss_size == 0 ? SS_DISABLE
2133 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2137 * Routines for handling mm_structs
2139 extern struct mm_struct * mm_alloc(void);
2141 /* mmdrop drops the mm and the page tables */
2142 extern void __mmdrop(struct mm_struct *);
2143 static inline void mmdrop(struct mm_struct * mm)
2145 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2146 __mmdrop(mm);
2149 /* mmput gets rid of the mappings and all user-space */
2150 extern void mmput(struct mm_struct *);
2151 /* Grab a reference to a task's mm, if it is not already going away */
2152 extern struct mm_struct *get_task_mm(struct task_struct *task);
2153 /* Remove the current tasks stale references to the old mm_struct */
2154 extern void mm_release(struct task_struct *, struct mm_struct *);
2155 /* Allocate a new mm structure and copy contents from tsk->mm */
2156 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2158 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2159 struct task_struct *, struct pt_regs *);
2160 extern void flush_thread(void);
2161 extern void exit_thread(void);
2163 extern void exit_files(struct task_struct *);
2164 extern void __cleanup_sighand(struct sighand_struct *);
2166 extern void exit_itimers(struct signal_struct *);
2167 extern void flush_itimer_signals(void);
2169 extern NORET_TYPE void do_group_exit(int);
2171 extern void daemonize(const char *, ...);
2172 extern int allow_signal(int);
2173 extern int disallow_signal(int);
2175 extern int do_execve(const char *,
2176 const char __user * const __user *,
2177 const char __user * const __user *, struct pt_regs *);
2178 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2179 struct task_struct *fork_idle(int);
2181 extern void set_task_comm(struct task_struct *tsk, char *from);
2182 extern char *get_task_comm(char *to, struct task_struct *tsk);
2184 #ifdef CONFIG_SMP
2185 void scheduler_ipi(void);
2186 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2187 #else
2188 static inline void scheduler_ipi(void) { }
2189 static inline unsigned long wait_task_inactive(struct task_struct *p,
2190 long match_state)
2192 return 1;
2194 #endif
2196 #define next_task(p) \
2197 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2199 #define for_each_process(p) \
2200 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2202 extern bool current_is_single_threaded(void);
2205 * Careful: do_each_thread/while_each_thread is a double loop so
2206 * 'break' will not work as expected - use goto instead.
2208 #define do_each_thread(g, t) \
2209 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2211 #define while_each_thread(g, t) \
2212 while ((t = next_thread(t)) != g)
2214 static inline int get_nr_threads(struct task_struct *tsk)
2216 return tsk->signal->nr_threads;
2219 /* de_thread depends on thread_group_leader not being a pid based check */
2220 #define thread_group_leader(p) (p == p->group_leader)
2222 /* Do to the insanities of de_thread it is possible for a process
2223 * to have the pid of the thread group leader without actually being
2224 * the thread group leader. For iteration through the pids in proc
2225 * all we care about is that we have a task with the appropriate
2226 * pid, we don't actually care if we have the right task.
2228 static inline int has_group_leader_pid(struct task_struct *p)
2230 return p->pid == p->tgid;
2233 static inline
2234 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2236 return p1->tgid == p2->tgid;
2239 static inline struct task_struct *next_thread(const struct task_struct *p)
2241 return list_entry_rcu(p->thread_group.next,
2242 struct task_struct, thread_group);
2245 static inline int thread_group_empty(struct task_struct *p)
2247 return list_empty(&p->thread_group);
2250 #define delay_group_leader(p) \
2251 (thread_group_leader(p) && !thread_group_empty(p))
2253 static inline int task_detached(struct task_struct *p)
2255 return p->exit_signal == -1;
2259 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2260 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2261 * pins the final release of task.io_context. Also protects ->cpuset and
2262 * ->cgroup.subsys[].
2264 * Nests both inside and outside of read_lock(&tasklist_lock).
2265 * It must not be nested with write_lock_irq(&tasklist_lock),
2266 * neither inside nor outside.
2268 static inline void task_lock(struct task_struct *p)
2270 spin_lock(&p->alloc_lock);
2273 static inline void task_unlock(struct task_struct *p)
2275 spin_unlock(&p->alloc_lock);
2278 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2279 unsigned long *flags);
2281 #define lock_task_sighand(tsk, flags) \
2282 ({ struct sighand_struct *__ss; \
2283 __cond_lock(&(tsk)->sighand->siglock, \
2284 (__ss = __lock_task_sighand(tsk, flags))); \
2285 __ss; \
2286 }) \
2288 static inline void unlock_task_sighand(struct task_struct *tsk,
2289 unsigned long *flags)
2291 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2294 #ifndef __HAVE_THREAD_FUNCTIONS
2296 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2297 #define task_stack_page(task) ((task)->stack)
2299 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2301 *task_thread_info(p) = *task_thread_info(org);
2302 task_thread_info(p)->task = p;
2305 static inline unsigned long *end_of_stack(struct task_struct *p)
2307 return (unsigned long *)(task_thread_info(p) + 1);
2310 #endif
2312 static inline int object_is_on_stack(void *obj)
2314 void *stack = task_stack_page(current);
2316 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2319 extern void thread_info_cache_init(void);
2321 #ifdef CONFIG_DEBUG_STACK_USAGE
2322 static inline unsigned long stack_not_used(struct task_struct *p)
2324 unsigned long *n = end_of_stack(p);
2326 do { /* Skip over canary */
2327 n++;
2328 } while (!*n);
2330 return (unsigned long)n - (unsigned long)end_of_stack(p);
2332 #endif
2334 /* set thread flags in other task's structures
2335 * - see asm/thread_info.h for TIF_xxxx flags available
2337 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2339 set_ti_thread_flag(task_thread_info(tsk), flag);
2342 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2344 clear_ti_thread_flag(task_thread_info(tsk), flag);
2347 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2349 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2352 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2354 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2357 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2359 return test_ti_thread_flag(task_thread_info(tsk), flag);
2362 static inline void set_tsk_need_resched(struct task_struct *tsk)
2364 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2367 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2369 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2372 static inline int test_tsk_need_resched(struct task_struct *tsk)
2374 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2377 static inline int restart_syscall(void)
2379 set_tsk_thread_flag(current, TIF_SIGPENDING);
2380 return -ERESTARTNOINTR;
2383 static inline int signal_pending(struct task_struct *p)
2385 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2388 static inline int __fatal_signal_pending(struct task_struct *p)
2390 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2393 static inline int fatal_signal_pending(struct task_struct *p)
2395 return signal_pending(p) && __fatal_signal_pending(p);
2398 static inline int signal_pending_state(long state, struct task_struct *p)
2400 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2401 return 0;
2402 if (!signal_pending(p))
2403 return 0;
2405 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2408 static inline int need_resched(void)
2410 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2414 * cond_resched() and cond_resched_lock(): latency reduction via
2415 * explicit rescheduling in places that are safe. The return
2416 * value indicates whether a reschedule was done in fact.
2417 * cond_resched_lock() will drop the spinlock before scheduling,
2418 * cond_resched_softirq() will enable bhs before scheduling.
2420 extern int _cond_resched(void);
2422 #define cond_resched() ({ \
2423 __might_sleep(__FILE__, __LINE__, 0); \
2424 _cond_resched(); \
2427 extern int __cond_resched_lock(spinlock_t *lock);
2429 #ifdef CONFIG_PREEMPT
2430 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2431 #else
2432 #define PREEMPT_LOCK_OFFSET 0
2433 #endif
2435 #define cond_resched_lock(lock) ({ \
2436 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2437 __cond_resched_lock(lock); \
2440 extern int __cond_resched_softirq(void);
2442 #define cond_resched_softirq() ({ \
2443 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2444 __cond_resched_softirq(); \
2448 * Does a critical section need to be broken due to another
2449 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2450 * but a general need for low latency)
2452 static inline int spin_needbreak(spinlock_t *lock)
2454 #ifdef CONFIG_PREEMPT
2455 return spin_is_contended(lock);
2456 #else
2457 return 0;
2458 #endif
2462 * Thread group CPU time accounting.
2464 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2465 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2467 static inline void thread_group_cputime_init(struct signal_struct *sig)
2469 spin_lock_init(&sig->cputimer.lock);
2473 * Reevaluate whether the task has signals pending delivery.
2474 * Wake the task if so.
2475 * This is required every time the blocked sigset_t changes.
2476 * callers must hold sighand->siglock.
2478 extern void recalc_sigpending_and_wake(struct task_struct *t);
2479 extern void recalc_sigpending(void);
2481 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2484 * Wrappers for p->thread_info->cpu access. No-op on UP.
2486 #ifdef CONFIG_SMP
2488 static inline unsigned int task_cpu(const struct task_struct *p)
2490 return task_thread_info(p)->cpu;
2493 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2495 #else
2497 static inline unsigned int task_cpu(const struct task_struct *p)
2499 return 0;
2502 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2506 #endif /* CONFIG_SMP */
2508 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2509 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2511 extern void normalize_rt_tasks(void);
2513 #ifdef CONFIG_CGROUP_SCHED
2515 extern struct task_group root_task_group;
2517 extern struct task_group *sched_create_group(struct task_group *parent);
2518 extern void sched_destroy_group(struct task_group *tg);
2519 extern void sched_move_task(struct task_struct *tsk);
2520 #ifdef CONFIG_FAIR_GROUP_SCHED
2521 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2522 extern unsigned long sched_group_shares(struct task_group *tg);
2523 #endif
2524 #ifdef CONFIG_RT_GROUP_SCHED
2525 extern int sched_group_set_rt_runtime(struct task_group *tg,
2526 long rt_runtime_us);
2527 extern long sched_group_rt_runtime(struct task_group *tg);
2528 extern int sched_group_set_rt_period(struct task_group *tg,
2529 long rt_period_us);
2530 extern long sched_group_rt_period(struct task_group *tg);
2531 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2532 #endif
2533 #endif
2535 extern int task_can_switch_user(struct user_struct *up,
2536 struct task_struct *tsk);
2538 #ifdef CONFIG_TASK_XACCT
2539 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2541 tsk->ioac.rchar += amt;
2544 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2546 tsk->ioac.wchar += amt;
2549 static inline void inc_syscr(struct task_struct *tsk)
2551 tsk->ioac.syscr++;
2554 static inline void inc_syscw(struct task_struct *tsk)
2556 tsk->ioac.syscw++;
2558 #else
2559 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2563 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2567 static inline void inc_syscr(struct task_struct *tsk)
2571 static inline void inc_syscw(struct task_struct *tsk)
2574 #endif
2576 #ifndef TASK_SIZE_OF
2577 #define TASK_SIZE_OF(tsk) TASK_SIZE
2578 #endif
2580 #ifdef CONFIG_MM_OWNER
2581 extern void mm_update_next_owner(struct mm_struct *mm);
2582 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2583 #else
2584 static inline void mm_update_next_owner(struct mm_struct *mm)
2588 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2591 #endif /* CONFIG_MM_OWNER */
2593 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2594 unsigned int limit)
2596 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2599 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2600 unsigned int limit)
2602 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2605 static inline unsigned long rlimit(unsigned int limit)
2607 return task_rlimit(current, limit);
2610 static inline unsigned long rlimit_max(unsigned int limit)
2612 return task_rlimit_max(current, limit);
2615 #endif /* __KERNEL__ */
2617 #endif