[gesichtete Version] | [gesichtete Version] |
Zeile 16: | Zeile 16: | ||
</p> | </p> | ||
== sched.h == | |||
<br /> | <br /> | ||
<p> | <p> | ||
<loop_listing title="Quelltext der Datei sched.h aus dem Linux-Kernel" description="Der Prozesskontrollblock ist deklariert zwischen Zeile 1043 und 1459."> | <loop_listing title="Quelltext der Datei sched.h aus dem Linux-Kernel" description="Der Prozesskontrollblock ist deklariert zwischen Zeile 1043 und 1459."> | ||
<source lang="c | <source lang="c" line="true"> | ||
#ifndef _LINUX_SCHED_H | #ifndef _LINUX_SCHED_H | ||
#define _LINUX_SCHED_H | #define _LINUX_SCHED_H |
Weiterführende Literatur
Achilles 2006 zeigt in Kapitel 3.1 den Linux Process Control Block. Die Lektüre dieser Quelle sei ausdrücklich empfohlen.
Studierende sind oftmals berechtigt, eine PDF-Version dieses Buches ohne entstehende Kosten über ihre Hochschulen von Springerlink zu beziehen.
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <uapi/linux/sched.h>
5
6
7 struct sched_param {
8 int sched_priority;
9 };
10
11 #include <asm/param.h> /* for HZ */
12
13 #include <linux/capability.h>
14 #include <linux/threads.h>
15 #include <linux/kernel.h>
16 #include <linux/types.h>
17 #include <linux/timex.h>
18 #include <linux/jiffies.h>
19 #include <linux/rbtree.h>
20 #include <linux/thread_info.h>
21 #include <linux/cpumask.h>
22 #include <linux/errno.h>
23 #include <linux/nodemask.h>
24 #include <linux/mm_types.h>
25 #include <linux/preempt_mask.h>
26
27 #include <asm/page.h>
28 #include <asm/ptrace.h>
29 #include <asm/cputime.h>
30
31 #include <linux/smp.h>
32 #include <linux/sem.h>
33 #include <linux/signal.h>
34 #include <linux/compiler.h>
35 #include <linux/completion.h>
36 #include <linux/pid.h>
37 #include <linux/percpu.h>
38 #include <linux/topology.h>
39 #include <linux/proportions.h>
40 #include <linux/seccomp.h>
41 #include <linux/rcupdate.h>
42 #include <linux/rculist.h>
43 #include <linux/rtmutex.h>
44
45 #include <linux/time.h>
46 #include <linux/param.h>
47 #include <linux/resource.h>
48 #include <linux/timer.h>
49 #include <linux/hrtimer.h>
50 #include <linux/task_io_accounting.h>
51 #include <linux/latencytop.h>
52 #include <linux/cred.h>
53 #include <linux/llist.h>
54 #include <linux/uidgid.h>
55 #include <linux/gfp.h>
56
57 #include <asm/processor.h>
58
59 struct exec_domain;
60 struct futex_pi_state;
61 struct robust_list_head;
62 struct bio_list;
63 struct fs_struct;
64 struct perf_event_context;
65 struct blk_plug;
66
67 /*
68 * List of flags we want to share for kernel threads,
69 * if only because they are not used by them anyway.
70 */
71 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
72
73 /*
74 * These are the constant used to fake the fixed-point load-average
75 * counting. Some notes:
76 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
77 * a load-average precision of 10 bits integer + 11 bits fractional
78 * - if you want to count load-averages more often, you need more
79 * precision, or rounding will get you. With 2-second counting freq,
80 * the EXP_n values would be 1981, 2034 and 2043 if still using only
81 * 11 bit fractions.
82 */
83 extern unsigned long avenrun[]; /* Load averages */
84 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
85
86 #define FSHIFT 11 /* nr of bits of precision */
87 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
88 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
89 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
90 #define EXP_5 2014 /* 1/exp(5sec/5min) */
91 #define EXP_15 2037 /* 1/exp(5sec/15min) */
92
93 #define CALC_LOAD(load,exp,n) \
94 load *= exp; \
95 load += n*(FIXED_1-exp); \
96 load >>= FSHIFT;
97
98 extern unsigned long total_forks;
99 extern int nr_threads;
100 DECLARE_PER_CPU(unsigned long, process_counts);
101 extern int nr_processes(void);
102 extern unsigned long nr_running(void);
103 extern unsigned long nr_iowait(void);
104 extern unsigned long nr_iowait_cpu(int cpu);
105 extern unsigned long this_cpu_load(void);
106
107
108 extern void calc_global_load(unsigned long ticks);
109 extern void update_cpu_load_nohz(void);
110
111 extern unsigned long get_parent_ip(unsigned long addr);
112
113 extern void dump_cpu_task(int cpu);
114
115 struct seq_file;
116 struct cfs_rq;
117 struct task_group;
118 #ifdef CONFIG_SCHED_DEBUG
119 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
120 extern void proc_sched_set_task(struct task_struct *p);
121 extern void
122 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
123 #endif
124
125 /*
126 * Task state bitmask. NOTE! These bits are also
127 * encoded in fs/proc/array.c: get_task_state().
128 *
129 * We have two separate sets of flags: task->state
130 * is about runnability, while task->exit_state are
131 * about the task exiting. Confusing, but this way
132 * modifying one set can't modify the other one by
133 * mistake.
134 */
135 #define TASK_RUNNING 0
136 #define TASK_INTERRUPTIBLE 1
137 #define TASK_UNINTERRUPTIBLE 2
138 #define __TASK_STOPPED 4
139 #define __TASK_TRACED 8
140 /* in tsk->exit_state */
141 #define EXIT_ZOMBIE 16
142 #define EXIT_DEAD 32
143 /* in tsk->state again */
144 #define TASK_DEAD 64
145 #define TASK_WAKEKILL 128
146 #define TASK_WAKING 256
147 #define TASK_PARKED 512
148 #define TASK_STATE_MAX 1024
149
150 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKWP"
151
152 extern char ___assert_task_state[1 - 2*!!(
153 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
154
155 /* Convenience macros for the sake of set_task_state */
156 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
157 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
158 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
159
160 /* Convenience macros for the sake of wake_up */
161 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
162 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
163
164 /* get_task_state() */
165 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
166 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
167 __TASK_TRACED)
168
169 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
170 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
171 #define task_is_dead(task) ((task)->exit_state != 0)
172 #define task_is_stopped_or_traced(task) \
173 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
174 #define task_contributes_to_load(task) \
175 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
176 (task->flags & PF_FROZEN) == 0)
177
178 #define __set_task_state(tsk, state_value) \
179 do { (tsk)->state = (state_value); } while (0)
180 #define set_task_state(tsk, state_value) \
181 set_mb((tsk)->state, (state_value))
182
183 /*
184 * set_current_state() includes a barrier so that the write of current->state
185 * is correctly serialised wrt the caller's subsequent test of whether to
186 * actually sleep:
187 *
188 * set_current_state(TASK_UNINTERRUPTIBLE);
189 * if (do_i_need_to_sleep())
190 * schedule();
191 *
192 * If the caller does not need such serialisation then use __set_current_state()
193 */
194 #define __set_current_state(state_value) \
195 do { current->state = (state_value); } while (0)
196 #define set_current_state(state_value) \
197 set_mb(current->state, (state_value))
198
199 /* Task command name length */
200 #define TASK_COMM_LEN 16
201
202 #include <linux/spinlock.h>
203
204 /*
205 * This serializes "schedule()" and also protects
206 * the run-queue from deletions/modifications (but
207 * _adding_ to the beginning of the run-queue has
208 * a separate lock).
209 */
210 extern rwlock_t tasklist_lock;
211 extern spinlock_t mmlist_lock;
212
213 struct task_struct;
214
215 #ifdef CONFIG_PROVE_RCU
216 extern int lockdep_tasklist_lock_is_held(void);
217 #endif /* #ifdef CONFIG_PROVE_RCU */
218
219 extern void sched_init(void);
220 extern void sched_init_smp(void);
221 extern asmlinkage void schedule_tail(struct task_struct *prev);
222 extern void init_idle(struct task_struct *idle, int cpu);
223 extern void init_idle_bootup_task(struct task_struct *idle);
224
225 extern int runqueue_is_locked(int cpu);
226
227 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
228 extern void nohz_balance_enter_idle(int cpu);
229 extern void set_cpu_sd_state_idle(void);
230 extern int get_nohz_timer_target(void);
231 #else
232 static inline void nohz_balance_enter_idle(int cpu) { }
233 static inline void set_cpu_sd_state_idle(void) { }
234 #endif
235
236 /*
237 * Only dump TASK_* tasks. (0 for all tasks)
238 */
239 extern void show_state_filter(unsigned long state_filter);
240
241 static inline void show_state(void)
242 {
243 show_state_filter(0);
244 }
245
246 extern void show_regs(struct pt_regs *);
247
248 /*
249 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
250 * task), SP is the stack pointer of the first frame that should be shown in the back
251 * trace (or NULL if the entire call-chain of the task should be shown).
252 */
253 extern void show_stack(struct task_struct *task, unsigned long *sp);
254
255 void io_schedule(void);
256 long io_schedule_timeout(long timeout);
257
258 extern void cpu_init (void);
259 extern void trap_init(void);
260 extern void update_process_times(int user);
261 extern void scheduler_tick(void);
262
263 extern void sched_show_task(struct task_struct *p);
264
265 #ifdef CONFIG_LOCKUP_DETECTOR
266 extern void touch_softlockup_watchdog(void);
267 extern void touch_softlockup_watchdog_sync(void);
268 extern void touch_all_softlockup_watchdogs(void);
269 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
270 void __user *buffer,
271 size_t *lenp, loff_t *ppos);
272 extern unsigned int softlockup_panic;
273 void lockup_detector_init(void);
274 #else
275 static inline void touch_softlockup_watchdog(void)
276 {
277 }
278 static inline void touch_softlockup_watchdog_sync(void)
279 {
280 }
281 static inline void touch_all_softlockup_watchdogs(void)
282 {
283 }
284 static inline void lockup_detector_init(void)
285 {
286 }
287 #endif
288
289 #ifdef CONFIG_DETECT_HUNG_TASK
290 void reset_hung_task_detector(void);
291 #else
292 static inline void reset_hung_task_detector(void)
293 {
294 }
295 #endif
296
297 /* Attach to any functions which should be ignored in wchan output. */
298 #define __sched __attribute__((__section__(".sched.text")))
299
300 /* Linker adds these: start and end of __sched functions */
301 extern char __sched_text_start[], __sched_text_end[];
302
303 /* Is this address in the __sched functions? */
304 extern int in_sched_functions(unsigned long addr);
305
306 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
307 extern signed long schedule_timeout(signed long timeout);
308 extern signed long schedule_timeout_interruptible(signed long timeout);
309 extern signed long schedule_timeout_killable(signed long timeout);
310 extern signed long schedule_timeout_uninterruptible(signed long timeout);
311 asmlinkage void schedule(void);
312 extern void schedule_preempt_disabled(void);
313
314 struct nsproxy;
315 struct user_namespace;
316
317 #ifdef CONFIG_MMU
318 extern void arch_pick_mmap_layout(struct mm_struct *mm);
319 extern unsigned long
320 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
321 unsigned long, unsigned long);
322 extern unsigned long
323 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
324 unsigned long len, unsigned long pgoff,
325 unsigned long flags);
326 #else
327 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
328 #endif
329
330
331 extern void set_dumpable(struct mm_struct *mm, int value);
332 extern int get_dumpable(struct mm_struct *mm);
333
334 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
335 #define SUID_DUMP_USER 1 /* Dump as user of process */
336 #define SUID_DUMP_ROOT 2 /* Dump as root */
337
338 /* mm flags */
339 /* dumpable bits */
340 #define MMF_DUMPABLE 0 /* core dump is permitted */
341 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
342
343 #define MMF_DUMPABLE_BITS 2
344 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
345
346 /* coredump filter bits */
347 #define MMF_DUMP_ANON_PRIVATE 2
348 #define MMF_DUMP_ANON_SHARED 3
349 #define MMF_DUMP_MAPPED_PRIVATE 4
350 #define MMF_DUMP_MAPPED_SHARED 5
351 #define MMF_DUMP_ELF_HEADERS 6
352 #define MMF_DUMP_HUGETLB_PRIVATE 7
353 #define MMF_DUMP_HUGETLB_SHARED 8
354
355 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
356 #define MMF_DUMP_FILTER_BITS 7
357 #define MMF_DUMP_FILTER_MASK \
358 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
359 #define MMF_DUMP_FILTER_DEFAULT \
360 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
361 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
362
363 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
364 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
365 #else
366 # define MMF_DUMP_MASK_DEFAULT_ELF 0
367 #endif
368 /* leave room for more dump flags */
369 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
370 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
371 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
372
373 #define MMF_HAS_UPROBES 19 /* has uprobes */
374 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
375
376 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
377
378 struct sighand_struct {
379 atomic_t count;
380 struct k_sigaction action[_NSIG];
381 spinlock_t siglock;
382 wait_queue_head_t signalfd_wqh;
383 };
384
385 struct pacct_struct {
386 int ac_flag;
387 long ac_exitcode;
388 unsigned long ac_mem;
389 cputime_t ac_utime, ac_stime;
390 unsigned long ac_minflt, ac_majflt;
391 };
392
393 struct cpu_itimer {
394 cputime_t expires;
395 cputime_t incr;
396 u32 error;
397 u32 incr_error;
398 };
399
400 /**
401 * struct cputime - snaphsot of system and user cputime
402 * @utime: time spent in user mode
403 * @stime: time spent in system mode
404 *
405 * Gathers a generic snapshot of user and system time.
406 */
407 struct cputime {
408 cputime_t utime;
409 cputime_t stime;
410 };
411
412 /**
413 * struct task_cputime - collected CPU time counts
414 * @utime: time spent in user mode, in &cputime_t units
415 * @stime: time spent in kernel mode, in &cputime_t units
416 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
417 *
418 * This is an extension of struct cputime that includes the total runtime
419 * spent by the task from the scheduler point of view.
420 *
421 * As a result, this structure groups together three kinds of CPU time
422 * that are tracked for threads and thread groups. Most things considering
423 * CPU time want to group these counts together and treat all three
424 * of them in parallel.
425 */
426 struct task_cputime {
427 cputime_t utime;
428 cputime_t stime;
429 unsigned long long sum_exec_runtime;
430 };
431 /* Alternate field names when used to cache expirations. */
432 #define prof_exp stime
433 #define virt_exp utime
434 #define sched_exp sum_exec_runtime
435
436 #define INIT_CPUTIME \
437 (struct task_cputime) { \
438 .utime = 0, \
439 .stime = 0, \
440 .sum_exec_runtime = 0, \
441 }
442
443 #ifdef CONFIG_PREEMPT_COUNT
444 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
445 #else
446 #define PREEMPT_DISABLED PREEMPT_ENABLED
447 #endif
448
449 /*
450 * Disable preemption until the scheduler is running.
451 * Reset by start_kernel()->sched_init()->init_idle().
452 *
453 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
454 * before the scheduler is active -- see should_resched().
455 */
456 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
457
458 /**
459 * struct thread_group_cputimer - thread group interval timer counts
460 * @cputime: thread group interval timers.
461 * @running: non-zero when there are timers running and
462 * @cputime receives updates.
463 * @lock: lock for fields in this struct.
464 *
465 * This structure contains the version of task_cputime, above, that is
466 * used for thread group CPU timer calculations.
467 */
468 struct thread_group_cputimer {
469 struct task_cputime cputime;
470 int running;
471 raw_spinlock_t lock;
472 };
473
474 #include <linux/rwsem.h>
475 struct autogroup;
476
477 /*
478 * NOTE! "signal_struct" does not have its own
479 * locking, because a shared signal_struct always
480 * implies a shared sighand_struct, so locking
481 * sighand_struct is always a proper superset of
482 * the locking of signal_struct.
483 */
484 struct signal_struct {
485 atomic_t sigcnt;
486 atomic_t live;
487 int nr_threads;
488 struct list_head thread_head;
489
490 wait_queue_head_t wait_chldexit; /* for wait4() */
491
492 /* current thread group signal load-balancing target: */
493 struct task_struct *curr_target;
494
495 /* shared signal handling: */
496 struct sigpending shared_pending;
497
498 /* thread group exit support */
499 int group_exit_code;
500 /* overloaded:
501 * - notify group_exit_task when ->count is equal to notify_count
502 * - everyone except group_exit_task is stopped during signal delivery
503 * of fatal signals, group_exit_task processes the signal.
504 */
505 int notify_count;
506 struct task_struct *group_exit_task;
507
508 /* thread group stop support, overloads group_exit_code too */
509 int group_stop_count;
510 unsigned int flags; /* see SIGNAL_* flags below */
511
512 /*
513 * PR_SET_CHILD_SUBREAPER marks a process, like a service
514 * manager, to re-parent orphan (double-forking) child processes
515 * to this process instead of 'init'. The service manager is
516 * able to receive SIGCHLD signals and is able to investigate
517 * the process until it calls wait(). All children of this
518 * process will inherit a flag if they should look for a
519 * child_subreaper process at exit.
520 */
521 unsigned int is_child_subreaper:1;
522 unsigned int has_child_subreaper:1;
523
524 /* POSIX.1b Interval Timers */
525 int posix_timer_id;
526 struct list_head posix_timers;
527
528 /* ITIMER_REAL timer for the process */
529 struct hrtimer real_timer;
530 struct pid *leader_pid;
531 ktime_t it_real_incr;
532
533 /*
534 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
535 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
536 * values are defined to 0 and 1 respectively
537 */
538 struct cpu_itimer it[2];
539
540 /*
541 * Thread group totals for process CPU timers.
542 * See thread_group_cputimer(), et al, for details.
543 */
544 struct thread_group_cputimer cputimer;
545
546 /* Earliest-expiration cache. */
547 struct task_cputime cputime_expires;
548
549 struct list_head cpu_timers[3];
550
551 struct pid *tty_old_pgrp;
552
553 /* boolean value for session group leader */
554 int leader;
555
556 struct tty_struct *tty; /* NULL if no tty */
557
558 #ifdef CONFIG_SCHED_AUTOGROUP
559 struct autogroup *autogroup;
560 #endif
561 /*
562 * Cumulative resource counters for dead threads in the group,
563 * and for reaped dead child processes forked by this group.
564 * Live threads maintain their own counters and add to these
565 * in __exit_signal, except for the group leader.
566 */
567 cputime_t utime, stime, cutime, cstime;
568 cputime_t gtime;
569 cputime_t cgtime;
570 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
571 struct cputime prev_cputime;
572 #endif
573 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
574 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
575 unsigned long inblock, oublock, cinblock, coublock;
576 unsigned long maxrss, cmaxrss;
577 struct task_io_accounting ioac;
578
579 /*
580 * Cumulative ns of schedule CPU time fo dead threads in the
581 * group, not including a zombie group leader, (This only differs
582 * from jiffies_to_ns(utime + stime) if sched_clock uses something
583 * other than jiffies.)
584 */
585 unsigned long long sum_sched_runtime;
586
587 /*
588 * We don't bother to synchronize most readers of this at all,
589 * because there is no reader checking a limit that actually needs
590 * to get both rlim_cur and rlim_max atomically, and either one
591 * alone is a single word that can safely be read normally.
592 * getrlimit/setrlimit use task_lock(current->group_leader) to
593 * protect this instead of the siglock, because they really
594 * have no need to disable irqs.
595 */
596 struct rlimit rlim[RLIM_NLIMITS];
597
598 #ifdef CONFIG_BSD_PROCESS_ACCT
599 struct pacct_struct pacct; /* per-process accounting information */
600 #endif
601 #ifdef CONFIG_TASKSTATS
602 struct taskstats *stats;
603 #endif
604 #ifdef CONFIG_AUDIT
605 unsigned audit_tty;
606 unsigned audit_tty_log_passwd;
607 struct tty_audit_buf *tty_audit_buf;
608 #endif
609 #ifdef CONFIG_CGROUPS
610 /*
611 * group_rwsem prevents new tasks from entering the threadgroup and
612 * member tasks from exiting,a more specifically, setting of
613 * PF_EXITING. fork and exit paths are protected with this rwsem
614 * using threadgroup_change_begin/end(). Users which require
615 * threadgroup to remain stable should use threadgroup_[un]lock()
616 * which also takes care of exec path. Currently, cgroup is the
617 * only user.
618 */
619 struct rw_semaphore group_rwsem;
620 #endif
621
622 oom_flags_t oom_flags;
623 short oom_score_adj; /* OOM kill score adjustment */
624 short oom_score_adj_min; /* OOM kill score adjustment min value.
625 * Only settable by CAP_SYS_RESOURCE. */
626
627 struct mutex cred_guard_mutex; /* guard against foreign influences on
628 * credential calculations
629 * (notably. ptrace) */
630 };
631
632 /*
633 * Bits in flags field of signal_struct.
634 */
635 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
636 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
637 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
638 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
639 /*
640 * Pending notifications to parent.
641 */
642 #define SIGNAL_CLD_STOPPED 0x00000010
643 #define SIGNAL_CLD_CONTINUED 0x00000020
644 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
645
646 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
647
648 /* If true, all threads except ->group_exit_task have pending SIGKILL */
649 static inline int signal_group_exit(const struct signal_struct *sig)
650 {
651 return (sig->flags & SIGNAL_GROUP_EXIT) ||
652 (sig->group_exit_task != NULL);
653 }
654
655 /*
656 * Some day this will be a full-fledged user tracking system..
657 */
658 struct user_struct {
659 atomic_t __count; /* reference count */
660 atomic_t processes; /* How many processes does this user have? */
661 atomic_t files; /* How many open files does this user have? */
662 atomic_t sigpending; /* How many pending signals does this user have? */
663 #ifdef CONFIG_INOTIFY_USER
664 atomic_t inotify_watches; /* How many inotify watches does this user have? */
665 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
666 #endif
667 #ifdef CONFIG_FANOTIFY
668 atomic_t fanotify_listeners;
669 #endif
670 #ifdef CONFIG_EPOLL
671 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
672 #endif
673 #ifdef CONFIG_POSIX_MQUEUE
674 /* protected by mq_lock */
675 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
676 #endif
677 unsigned long locked_shm; /* How many pages of mlocked shm ? */
678
679 #ifdef CONFIG_KEYS
680 struct key *uid_keyring; /* UID specific keyring */
681 struct key *session_keyring; /* UID's default session keyring */
682 #endif
683
684 /* Hash table maintenance information */
685 struct hlist_node uidhash_node;
686 kuid_t uid;
687
688 #ifdef CONFIG_PERF_EVENTS
689 atomic_long_t locked_vm;
690 #endif
691 };
692
693 extern int uids_sysfs_init(void);
694
695 extern struct user_struct *find_user(kuid_t);
696
697 extern struct user_struct root_user;
698 #define INIT_USER (&root_user)
699
700
701 struct backing_dev_info;
702 struct reclaim_state;
703
704 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
705 struct sched_info {
706 /* cumulative counters */
707 unsigned long pcount; /* # of times run on this cpu */
708 unsigned long long run_delay; /* time spent waiting on a runqueue */
709
710 /* timestamps */
711 unsigned long long last_arrival,/* when we last ran on a cpu */
712 last_queued; /* when we were last queued to run */
713 };
714 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
715
716 #ifdef CONFIG_TASK_DELAY_ACCT
717 struct task_delay_info {
718 spinlock_t lock;
719 unsigned int flags; /* Private per-task flags */
720
721 /* For each stat XXX, add following, aligned appropriately
722 *
723 * struct timespec XXX_start, XXX_end;
724 * u64 XXX_delay;
725 * u32 XXX_count;
726 *
727 * Atomicity of updates to XXX_delay, XXX_count protected by
728 * single lock above (split into XXX_lock if contention is an issue).
729 */
730
731 /*
732 * XXX_count is incremented on every XXX operation, the delay
733 * associated with the operation is added to XXX_delay.
734 * XXX_delay contains the accumulated delay time in nanoseconds.
735 */
736 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
737 u64 blkio_delay; /* wait for sync block io completion */
738 u64 swapin_delay; /* wait for swapin block io completion */
739 u32 blkio_count; /* total count of the number of sync block */
740 /* io operations performed */
741 u32 swapin_count; /* total count of the number of swapin block */
742 /* io operations performed */
743
744 struct timespec freepages_start, freepages_end;
745 u64 freepages_delay; /* wait for memory reclaim */
746 u32 freepages_count; /* total count of memory reclaim */
747 };
748 #endif /* CONFIG_TASK_DELAY_ACCT */
749
750 static inline int sched_info_on(void)
751 {
752 #ifdef CONFIG_SCHEDSTATS
753 return 1;
754 #elif defined(CONFIG_TASK_DELAY_ACCT)
755 extern int delayacct_on;
756 return delayacct_on;
757 #else
758 return 0;
759 #endif
760 }
761
762 enum cpu_idle_type {
763 CPU_IDLE,
764 CPU_NOT_IDLE,
765 CPU_NEWLY_IDLE,
766 CPU_MAX_IDLE_TYPES
767 };
768
769 /*
770 * Increase resolution of cpu_power calculations
771 */
772 #define SCHED_POWER_SHIFT 10
773 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
774
775 /*
776 * sched-domains (multiprocessor balancing) declarations:
777 */
778 #ifdef CONFIG_SMP
779 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
780 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
781 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
782 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
783 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
784 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
785 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
786 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
787 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
788 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
789 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
790 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
791 #define SD_NUMA 0x4000 /* cross-node balancing */
792
793 extern int __weak arch_sd_sibiling_asym_packing(void);
794
795 struct sched_domain_attr {
796 int relax_domain_level;
797 };
798
799 #define SD_ATTR_INIT (struct sched_domain_attr) { \
800 .relax_domain_level = -1, \
801 }
802
803 extern int sched_domain_level_max;
804
805 struct sched_group;
806
807 struct sched_domain {
808 /* These fields must be setup */
809 struct sched_domain *parent; /* top domain must be null terminated */
810 struct sched_domain *child; /* bottom domain must be null terminated */
811 struct sched_group *groups; /* the balancing groups of the domain */
812 unsigned long min_interval; /* Minimum balance interval ms */
813 unsigned long max_interval; /* Maximum balance interval ms */
814 unsigned int busy_factor; /* less balancing by factor if busy */
815 unsigned int imbalance_pct; /* No balance until over watermark */
816 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
817 unsigned int busy_idx;
818 unsigned int idle_idx;
819 unsigned int newidle_idx;
820 unsigned int wake_idx;
821 unsigned int forkexec_idx;
822 unsigned int smt_gain;
823
824 int nohz_idle; /* NOHZ IDLE status */
825 int flags; /* See SD_* */
826 int level;
827
828 /* Runtime fields. */
829 unsigned long last_balance; /* init to jiffies. units in jiffies */
830 unsigned int balance_interval; /* initialise to 1. units in ms. */
831 unsigned int nr_balance_failed; /* initialise to 0 */
832
833 /* idle_balance() stats */
834 u64 max_newidle_lb_cost;
835 unsigned long next_decay_max_lb_cost;
836
837 #ifdef CONFIG_SCHEDSTATS
838 /* load_balance() stats */
839 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
840 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
841 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
842 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
843 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
844 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
845 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
846 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
847
848 /* Active load balancing */
849 unsigned int alb_count;
850 unsigned int alb_failed;
851 unsigned int alb_pushed;
852
853 /* SD_BALANCE_EXEC stats */
854 unsigned int sbe_count;
855 unsigned int sbe_balanced;
856 unsigned int sbe_pushed;
857
858 /* SD_BALANCE_FORK stats */
859 unsigned int sbf_count;
860 unsigned int sbf_balanced;
861 unsigned int sbf_pushed;
862
863 /* try_to_wake_up() stats */
864 unsigned int ttwu_wake_remote;
865 unsigned int ttwu_move_affine;
866 unsigned int ttwu_move_balance;
867 #endif
868 #ifdef CONFIG_SCHED_DEBUG
869 char *name;
870 #endif
871 union {
872 void *private; /* used during construction */
873 struct rcu_head rcu; /* used during destruction */
874 };
875
876 unsigned int span_weight;
877 /*
878 * Span of all CPUs in this domain.
879 *
880 * NOTE: this field is variable length. (Allocated dynamically
881 * by attaching extra space to the end of the structure,
882 * depending on how many CPUs the kernel has booted up with)
883 */
884 unsigned long span[0];
885 };
886
887 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
888 {
889 return to_cpumask(sd->span);
890 }
891
892 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
893 struct sched_domain_attr *dattr_new);
894
895 /* Allocate an array of sched domains, for partition_sched_domains(). */
896 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
897 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
898
899 bool cpus_share_cache(int this_cpu, int that_cpu);
900
901 #else /* CONFIG_SMP */
902
903 struct sched_domain_attr;
904
905 static inline void
906 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
907 struct sched_domain_attr *dattr_new)
908 {
909 }
910
911 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
912 {
913 return true;
914 }
915
916 #endif /* !CONFIG_SMP */
917
918
919 struct io_context; /* See blkdev.h */
920
921
922 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
923 extern void prefetch_stack(struct task_struct *t);
924 #else
925 static inline void prefetch_stack(struct task_struct *t) { }
926 #endif
927
928 struct audit_context; /* See audit.c */
929 struct mempolicy;
930 struct pipe_inode_info;
931 struct uts_namespace;
932
933 struct load_weight {
934 unsigned long weight;
935 u32 inv_weight;
936 };
937
938 struct sched_avg {
939 /*
940 * These sums represent an infinite geometric series and so are bound
941 * above by 1024/(1-y). Thus we only need a u32 to store them for all
942 * choices of y < 1-2^(-32)*1024.
943 */
944 u32 runnable_avg_sum, runnable_avg_period;
945 u64 last_runnable_update;
946 s64 decay_count;
947 unsigned long load_avg_contrib;
948 };
949
950 #ifdef CONFIG_SCHEDSTATS
951 struct sched_statistics {
952 u64 wait_start;
953 u64 wait_max;
954 u64 wait_count;
955 u64 wait_sum;
956 u64 iowait_count;
957 u64 iowait_sum;
958
959 u64 sleep_start;
960 u64 sleep_max;
961 s64 sum_sleep_runtime;
962
963 u64 block_start;
964 u64 block_max;
965 u64 exec_max;
966 u64 slice_max;
967
968 u64 nr_migrations_cold;
969 u64 nr_failed_migrations_affine;
970 u64 nr_failed_migrations_running;
971 u64 nr_failed_migrations_hot;
972 u64 nr_forced_migrations;
973
974 u64 nr_wakeups;
975 u64 nr_wakeups_sync;
976 u64 nr_wakeups_migrate;
977 u64 nr_wakeups_local;
978 u64 nr_wakeups_remote;
979 u64 nr_wakeups_affine;
980 u64 nr_wakeups_affine_attempts;
981 u64 nr_wakeups_passive;
982 u64 nr_wakeups_idle;
983 };
984 #endif
985
986 struct sched_entity {
987 struct load_weight load; /* for load-balancing */
988 struct rb_node run_node;
989 struct list_head group_node;
990 unsigned int on_rq;
991
992 u64 exec_start;
993 u64 sum_exec_runtime;
994 u64 vruntime;
995 u64 prev_sum_exec_runtime;
996
997 u64 nr_migrations;
998
999 #ifdef CONFIG_SCHEDSTATS
1000 struct sched_statistics statistics;
1001 #endif
1002
1003 #ifdef CONFIG_FAIR_GROUP_SCHED
1004 struct sched_entity *parent;
1005 /* rq on which this entity is (to be) queued: */
1006 struct cfs_rq *cfs_rq;
1007 /* rq "owned" by this entity/group: */
1008 struct cfs_rq *my_q;
1009 #endif
1010
1011 #ifdef CONFIG_SMP
1012 /* Per-entity load-tracking */
1013 struct sched_avg avg;
1014 #endif
1015 };
1016
1017 struct sched_rt_entity {
1018 struct list_head run_list;
1019 unsigned long timeout;
1020 unsigned long watchdog_stamp;
1021 unsigned int time_slice;
1022
1023 struct sched_rt_entity *back;
1024 #ifdef CONFIG_RT_GROUP_SCHED
1025 struct sched_rt_entity *parent;
1026 /* rq on which this entity is (to be) queued: */
1027 struct rt_rq *rt_rq;
1028 /* rq "owned" by this entity/group: */
1029 struct rt_rq *my_q;
1030 #endif
1031 };
1032
1033
1034 struct rcu_node;
1035
1036 enum perf_event_task_context {
1037 perf_invalid_context = -1,
1038 perf_hw_context = 0,
1039 perf_sw_context,
1040 perf_nr_task_contexts,
1041 };
1042
1043 struct task_struct {
1044 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1045 void *stack;
1046 atomic_t usage;
1047 unsigned int flags; /* per process flags, defined below */
1048 unsigned int ptrace;
1049
1050 #ifdef CONFIG_SMP
1051 struct llist_node wake_entry;
1052 int on_cpu;
1053 struct task_struct *last_wakee;
1054 unsigned long wakee_flips;
1055 unsigned long wakee_flip_decay_ts;
1056
1057 int wake_cpu;
1058 #endif
1059 int on_rq;
1060
1061 int prio, static_prio, normal_prio;
1062 unsigned int rt_priority;
1063 const struct sched_class *sched_class;
1064 struct sched_entity se;
1065 struct sched_rt_entity rt;
1066 #ifdef CONFIG_CGROUP_SCHED
1067 struct task_group *sched_task_group;
1068 #endif
1069
1070 #ifdef CONFIG_PREEMPT_NOTIFIERS
1071 /* list of struct preempt_notifier: */
1072 struct hlist_head preempt_notifiers;
1073 #endif
1074
1075 #ifdef CONFIG_BLK_DEV_IO_TRACE
1076 unsigned int btrace_seq;
1077 #endif
1078
1079 unsigned int policy;
1080 int nr_cpus_allowed;
1081 cpumask_t cpus_allowed;
1082
1083 #ifdef CONFIG_PREEMPT_RCU
1084 int rcu_read_lock_nesting;
1085 char rcu_read_unlock_special;
1086 struct list_head rcu_node_entry;
1087 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1088 #ifdef CONFIG_TREE_PREEMPT_RCU
1089 struct rcu_node *rcu_blocked_node;
1090 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1091 #ifdef CONFIG_RCU_BOOST
1092 struct rt_mutex *rcu_boost_mutex;
1093 #endif /* #ifdef CONFIG_RCU_BOOST */
1094
1095 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1096 struct sched_info sched_info;
1097 #endif
1098
1099 struct list_head tasks;
1100 #ifdef CONFIG_SMP
1101 struct plist_node pushable_tasks;
1102 #endif
1103
1104 struct mm_struct *mm, *active_mm;
1105 #ifdef CONFIG_COMPAT_BRK
1106 unsigned brk_randomized:1;
1107 #endif
1108 #if defined(SPLIT_RSS_COUNTING)
1109 struct task_rss_stat rss_stat;
1110 #endif
1111 /* task state */
1112 int exit_state;
1113 int exit_code, exit_signal;
1114 int pdeath_signal; /* The signal sent when the parent dies */
1115 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1116
1117 /* Used for emulating ABI behavior of previous Linux versions */
1118 unsigned int personality;
1119
1120 unsigned did_exec:1;
1121 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1122 * execve */
1123 unsigned in_iowait:1;
1124
1125 /* task may not gain privileges */
1126 unsigned no_new_privs:1;
1127
1128 /* Revert to default priority/policy when forking */
1129 unsigned sched_reset_on_fork:1;
1130 unsigned sched_contributes_to_load:1;
1131
1132 pid_t pid;
1133 pid_t tgid;
1134
1135 #ifdef CONFIG_CC_STACKPROTECTOR
1136 /* Canary value for the -fstack-protector gcc feature */
1137 unsigned long stack_canary;
1138 #endif
1139 /*
1140 * pointers to (original) parent process, youngest child, younger sibling,
1141 * older sibling, respectively. (p->father can be replaced with
1142 * p->real_parent->pid)
1143 */
1144 struct task_struct __rcu *real_parent; /* real parent process */
1145 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1146 /*
1147 * children/sibling forms the list of my natural children
1148 */
1149 struct list_head children; /* list of my children */
1150 struct list_head sibling; /* linkage in my parent's children list */
1151 struct task_struct *group_leader; /* threadgroup leader */
1152
1153 /*
1154 * ptraced is the list of tasks this task is using ptrace on.
1155 * This includes both natural children and PTRACE_ATTACH targets.
1156 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1157 */
1158 struct list_head ptraced;
1159 struct list_head ptrace_entry;
1160
1161 /* PID/PID hash table linkage. */
1162 struct pid_link pids[PIDTYPE_MAX];
1163 struct list_head thread_group;
1164 struct list_head thread_node;
1165
1166 struct completion *vfork_done; /* for vfork() */
1167 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1168 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1169
1170 cputime_t utime, stime, utimescaled, stimescaled;
1171 cputime_t gtime;
1172 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1173 struct cputime prev_cputime;
1174 #endif
1175 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1176 seqlock_t vtime_seqlock;
1177 unsigned long long vtime_snap;
1178 enum {
1179 VTIME_SLEEPING = 0,
1180 VTIME_USER,
1181 VTIME_SYS,
1182 } vtime_snap_whence;
1183 #endif
1184 unsigned long nvcsw, nivcsw; /* context switch counts */
1185 struct timespec start_time; /* monotonic time */
1186 struct timespec real_start_time; /* boot based time */
1187 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1188 unsigned long min_flt, maj_flt;
1189
1190 struct task_cputime cputime_expires;
1191 struct list_head cpu_timers[3];
1192
1193 /* process credentials */
1194 const struct cred __rcu *real_cred; /* objective and real subjective task
1195 * credentials (COW) */
1196 const struct cred __rcu *cred; /* effective (overridable) subjective task
1197 * credentials (COW) */
1198 char comm[TASK_COMM_LEN]; /* executable name excluding path
1199 - access with [gs]et_task_comm (which lock
1200 it with task_lock())
1201 - initialized normally by setup_new_exec */
1202 /* file system info */
1203 int link_count, total_link_count;
1204 #ifdef CONFIG_SYSVIPC
1205 /* ipc stuff */
1206 struct sysv_sem sysvsem;
1207 #endif
1208 #ifdef CONFIG_DETECT_HUNG_TASK
1209 /* hung task detection */
1210 unsigned long last_switch_count;
1211 #endif
1212 /* CPU-specific state of this task */
1213 struct thread_struct thread;
1214 /* filesystem information */
1215 struct fs_struct *fs;
1216 /* open file information */
1217 struct files_struct *files;
1218 /* namespaces */
1219 struct nsproxy *nsproxy;
1220 /* signal handlers */
1221 struct signal_struct *signal;
1222 struct sighand_struct *sighand;
1223
1224 sigset_t blocked, real_blocked;
1225 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1226 struct sigpending pending;
1227
1228 unsigned long sas_ss_sp;
1229 size_t sas_ss_size;
1230 int (*notifier)(void *priv);
1231 void *notifier_data;
1232 sigset_t *notifier_mask;
1233 struct callback_head *task_works;
1234
1235 struct audit_context *audit_context;
1236 #ifdef CONFIG_AUDITSYSCALL
1237 kuid_t loginuid;
1238 unsigned int sessionid;
1239 #endif
1240 struct seccomp seccomp;
1241
1242 /* Thread group tracking */
1243 u32 parent_exec_id;
1244 u32 self_exec_id;
1245 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1246 * mempolicy */
1247 spinlock_t alloc_lock;
1248
1249 /* Protection of the PI data structures: */
1250 raw_spinlock_t pi_lock;
1251
1252 #ifdef CONFIG_RT_MUTEXES
1253 /* PI waiters blocked on a rt_mutex held by this task */
1254 struct plist_head pi_waiters;
1255 /* Deadlock detection and priority inheritance handling */
1256 struct rt_mutex_waiter *pi_blocked_on;
1257 #endif
1258
1259 #ifdef CONFIG_DEBUG_MUTEXES
1260 /* mutex deadlock detection */
1261 struct mutex_waiter *blocked_on;
1262 #endif
1263 #ifdef CONFIG_TRACE_IRQFLAGS
1264 unsigned int irq_events;
1265 unsigned long hardirq_enable_ip;
1266 unsigned long hardirq_disable_ip;
1267 unsigned int hardirq_enable_event;
1268 unsigned int hardirq_disable_event;
1269 int hardirqs_enabled;
1270 int hardirq_context;
1271 unsigned long softirq_disable_ip;
1272 unsigned long softirq_enable_ip;
1273 unsigned int softirq_disable_event;
1274 unsigned int softirq_enable_event;
1275 int softirqs_enabled;
1276 int softirq_context;
1277 #endif
1278 #ifdef CONFIG_LOCKDEP
1279 # define MAX_LOCK_DEPTH 48UL
1280 u64 curr_chain_key;
1281 int lockdep_depth;
1282 unsigned int lockdep_recursion;
1283 struct held_lock held_locks[MAX_LOCK_DEPTH];
1284 gfp_t lockdep_reclaim_gfp;
1285 #endif
1286
1287 /* journalling filesystem info */
1288 void *journal_info;
1289
1290 /* stacked block device info */
1291 struct bio_list *bio_list;
1292
1293 #ifdef CONFIG_BLOCK
1294 /* stack plugging */
1295 struct blk_plug *plug;
1296 #endif
1297
1298 /* VM state */
1299 struct reclaim_state *reclaim_state;
1300
1301 struct backing_dev_info *backing_dev_info;
1302
1303 struct io_context *io_context;
1304
1305 unsigned long ptrace_message;
1306 siginfo_t *last_siginfo; /* For ptrace use. */
1307 struct task_io_accounting ioac;
1308 #if defined(CONFIG_TASK_XACCT)
1309 u64 acct_rss_mem1; /* accumulated rss usage */
1310 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1311 cputime_t acct_timexpd; /* stime + utime since last update */
1312 #endif
1313 #ifdef CONFIG_CPUSETS
1314 nodemask_t mems_allowed; /* Protected by alloc_lock */
1315 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1316 int cpuset_mem_spread_rotor;
1317 int cpuset_slab_spread_rotor;
1318 #endif
1319 #ifdef CONFIG_CGROUPS
1320 /* Control Group info protected by css_set_lock */
1321 struct css_set __rcu *cgroups;
1322 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1323 struct list_head cg_list;
1324 #endif
1325 #ifdef CONFIG_FUTEX
1326 struct robust_list_head __user *robust_list;
1327 #ifdef CONFIG_COMPAT
1328 struct compat_robust_list_head __user *compat_robust_list;
1329 #endif
1330 struct list_head pi_state_list;
1331 struct futex_pi_state *pi_state_cache;
1332 #endif
1333 #ifdef CONFIG_PERF_EVENTS
1334 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1335 struct mutex perf_event_mutex;
1336 struct list_head perf_event_list;
1337 #endif
1338 #ifdef CONFIG_NUMA
1339 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1340 short il_next;
1341 short pref_node_fork;
1342 #endif
1343 #ifdef CONFIG_NUMA_BALANCING
1344 int numa_scan_seq;
1345 unsigned int numa_scan_period;
1346 unsigned int numa_scan_period_max;
1347 int numa_preferred_nid;
1348 int numa_migrate_deferred;
1349 unsigned long numa_migrate_retry;
1350 u64 node_stamp; /* migration stamp */
1351 struct callback_head numa_work;
1352
1353 struct list_head numa_entry;
1354 struct numa_group *numa_group;
1355
1356 /*
1357 * Exponential decaying average of faults on a per-node basis.
1358 * Scheduling placement decisions are made based on the these counts.
1359 * The values remain static for the duration of a PTE scan
1360 */
1361 unsigned long *numa_faults;
1362 unsigned long total_numa_faults;
1363
1364 /*
1365 * numa_faults_buffer records faults per node during the current
1366 * scan window. When the scan completes, the counts in numa_faults
1367 * decay and these values are copied.
1368 */
1369 unsigned long *numa_faults_buffer;
1370
1371 /*
1372 * numa_faults_locality tracks if faults recorded during the last
1373 * scan window were remote/local. The task scan period is adapted
1374 * based on the locality of the faults with different weights
1375 * depending on whether they were shared or private faults
1376 */
1377 unsigned long numa_faults_locality[2];
1378
1379 unsigned long numa_pages_migrated;
1380 #endif /* CONFIG_NUMA_BALANCING */
1381
1382 struct rcu_head rcu;
1383
1384 /*
1385 * cache last used pipe for splice
1386 */
1387 struct pipe_inode_info *splice_pipe;
1388
1389 struct page_frag task_frag;
1390
1391 #ifdef CONFIG_TASK_DELAY_ACCT
1392 struct task_delay_info *delays;
1393 #endif
1394 #ifdef CONFIG_FAULT_INJECTION
1395 int make_it_fail;
1396 #endif
1397 /*
1398 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1399 * balance_dirty_pages() for some dirty throttling pause
1400 */
1401 int nr_dirtied;
1402 int nr_dirtied_pause;
1403 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1404
1405 #ifdef CONFIG_LATENCYTOP
1406 int latency_record_count;
1407 struct latency_record latency_record[LT_SAVECOUNT];
1408 #endif
1409 /*
1410 * time slack values; these are used to round up poll() and
1411 * select() etc timeout values. These are in nanoseconds.
1412 */
1413 unsigned long timer_slack_ns;
1414 unsigned long default_timer_slack_ns;
1415
1416 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1417 /* Index of current stored address in ret_stack */
1418 int curr_ret_stack;
1419 /* Stack of return addresses for return function tracing */
1420 struct ftrace_ret_stack *ret_stack;
1421 /* time stamp for last schedule */
1422 unsigned long long ftrace_timestamp;
1423 /*
1424 * Number of functions that haven't been traced
1425 * because of depth overrun.
1426 */
1427 atomic_t trace_overrun;
1428 /* Pause for the tracing */
1429 atomic_t tracing_graph_pause;
1430 #endif
1431 #ifdef CONFIG_TRACING
1432 /* state flags for use by tracers */
1433 unsigned long trace;
1434 /* bitmask and counter of trace recursion */
1435 unsigned long trace_recursion;
1436 #endif /* CONFIG_TRACING */
1437 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1438 struct memcg_batch_info {
1439 int do_batch; /* incremented when batch uncharge started */
1440 struct mem_cgroup *memcg; /* target memcg of uncharge */
1441 unsigned long nr_pages; /* uncharged usage */
1442 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1443 } memcg_batch;
1444 unsigned int memcg_kmem_skip_account;
1445 struct memcg_oom_info {
1446 struct mem_cgroup *memcg;
1447 gfp_t gfp_mask;
1448 int order;
1449 unsigned int may_oom:1;
1450 } memcg_oom;
1451 #endif
1452 #ifdef CONFIG_UPROBES
1453 struct uprobe_task *utask;
1454 #endif
1455 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1456 unsigned int sequential_io;
1457 unsigned int sequential_io_avg;
1458 #endif
1459 };
1460
1461 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1462 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1463
1464 #define TNF_MIGRATED 0x01
1465 #define TNF_NO_GROUP 0x02
1466 #define TNF_SHARED 0x04
1467 #define TNF_FAULT_LOCAL 0x08
1468
1469 #ifdef CONFIG_NUMA_BALANCING
1470 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1471 extern pid_t task_numa_group_id(struct task_struct *p);
1472 extern void set_numabalancing_state(bool enabled);
1473 extern void task_numa_free(struct task_struct *p);
1474
1475 extern unsigned int sysctl_numa_balancing_migrate_deferred;
1476 #else
1477 static inline void task_numa_fault(int last_node, int node, int pages,
1478 int flags)
1479 {
1480 }
1481 static inline pid_t task_numa_group_id(struct task_struct *p)
1482 {
1483 return 0;
1484 }
1485 static inline void set_numabalancing_state(bool enabled)
1486 {
1487 }
1488 static inline void task_numa_free(struct task_struct *p)
1489 {
1490 }
1491 #endif
1492
1493 static inline struct pid *task_pid(struct task_struct *task)
1494 {
1495 return task->pids[PIDTYPE_PID].pid;
1496 }
1497
1498 static inline struct pid *task_tgid(struct task_struct *task)
1499 {
1500 return task->group_leader->pids[PIDTYPE_PID].pid;
1501 }
1502
1503 /*
1504 * Without tasklist or rcu lock it is not safe to dereference
1505 * the result of task_pgrp/task_session even if task == current,
1506 * we can race with another thread doing sys_setsid/sys_setpgid.
1507 */
1508 static inline struct pid *task_pgrp(struct task_struct *task)
1509 {
1510 return task->group_leader->pids[PIDTYPE_PGID].pid;
1511 }
1512
1513 static inline struct pid *task_session(struct task_struct *task)
1514 {
1515 return task->group_leader->pids[PIDTYPE_SID].pid;
1516 }
1517
1518 struct pid_namespace;
1519
1520 /*
1521 * the helpers to get the task's different pids as they are seen
1522 * from various namespaces
1523 *
1524 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1525 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1526 * current.
1527 * task_xid_nr_ns() : id seen from the ns specified;
1528 *
1529 * set_task_vxid() : assigns a virtual id to a task;
1530 *
1531 * see also pid_nr() etc in include/linux/pid.h
1532 */
1533 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1534 struct pid_namespace *ns);
1535
1536 static inline pid_t task_pid_nr(struct task_struct *tsk)
1537 {
1538 return tsk->pid;
1539 }
1540
1541 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1542 struct pid_namespace *ns)
1543 {
1544 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1545 }
1546
1547 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1548 {
1549 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1550 }
1551
1552
1553 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1554 {
1555 return tsk->tgid;
1556 }
1557
1558 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1559
1560 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1561 {
1562 return pid_vnr(task_tgid(tsk));
1563 }
1564
1565
1566 static int pid_alive(const struct task_struct *p);
1567 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1568 {
1569 pid_t pid = 0;
1570
1571 rcu_read_lock();
1572 if (pid_alive(tsk))
1573 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1574 rcu_read_unlock();
1575
1576 return pid;
1577 }
1578
1579 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1580 {
1581 return task_ppid_nr_ns(tsk, &init_pid_ns);
1582 }
1583
1584 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1585 struct pid_namespace *ns)
1586 {
1587 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1588 }
1589
1590 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1591 {
1592 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1593 }
1594
1595
1596 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1597 struct pid_namespace *ns)
1598 {
1599 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1600 }
1601
1602 static inline pid_t task_session_vnr(struct task_struct *tsk)
1603 {
1604 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1605 }
1606
1607 /* obsolete, do not use */
1608 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1609 {
1610 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1611 }
1612
1613 /**
1614 * pid_alive - check that a task structure is not stale
1615 * @p: Task structure to be checked.
1616 *
1617 * Test if a process is not yet dead (at most zombie state)
1618 * If pid_alive fails, then pointers within the task structure
1619 * can be stale and must not be dereferenced.
1620 *
1621 * Return: 1 if the process is alive. 0 otherwise.
1622 */
1623 static inline int pid_alive(const struct task_struct *p)
1624 {
1625 return p->pids[PIDTYPE_PID].pid != NULL;
1626 }
1627
1628 /**
1629 * is_global_init - check if a task structure is init
1630 * @tsk: Task structure to be checked.
1631 *
1632 * Check if a task structure is the first user space task the kernel created.
1633 *
1634 * Return: 1 if the task structure is init. 0 otherwise.
1635 */
1636 static inline int is_global_init(struct task_struct *tsk)
1637 {
1638 return tsk->pid == 1;
1639 }
1640
1641 extern struct pid *cad_pid;
1642
1643 extern void free_task(struct task_struct *tsk);
1644 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1645
1646 extern void __put_task_struct(struct task_struct *t);
1647
1648 static inline void put_task_struct(struct task_struct *t)
1649 {
1650 if (atomic_dec_and_test(&t->usage))
1651 __put_task_struct(t);
1652 }
1653
1654 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1655 extern void task_cputime(struct task_struct *t,
1656 cputime_t *utime, cputime_t *stime);
1657 extern void task_cputime_scaled(struct task_struct *t,
1658 cputime_t *utimescaled, cputime_t *stimescaled);
1659 extern cputime_t task_gtime(struct task_struct *t);
1660 #else
1661 static inline void task_cputime(struct task_struct *t,
1662 cputime_t *utime, cputime_t *stime)
1663 {
1664 if (utime)
1665 *utime = t->utime;
1666 if (stime)
1667 *stime = t->stime;
1668 }
1669
1670 static inline void task_cputime_scaled(struct task_struct *t,
1671 cputime_t *utimescaled,
1672 cputime_t *stimescaled)
1673 {
1674 if (utimescaled)
1675 *utimescaled = t->utimescaled;
1676 if (stimescaled)
1677 *stimescaled = t->stimescaled;
1678 }
1679
1680 static inline cputime_t task_gtime(struct task_struct *t)
1681 {
1682 return t->gtime;
1683 }
1684 #endif
1685 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1686 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1687
1688 /*
1689 * Per process flags
1690 */
1691 #define PF_EXITING 0x00000004 /* getting shut down */
1692 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1693 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1694 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1695 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1696 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1697 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1698 #define PF_DUMPCORE 0x00000200 /* dumped core */
1699 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1700 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1701 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1702 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1703 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1704 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1705 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1706 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1707 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1708 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1709 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1710 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1711 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1712 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1713 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1714 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1715 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1716 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1717 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1718 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1719 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1720 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1721
1722 /*
1723 * Only the _current_ task can read/write to tsk->flags, but other
1724 * tasks can access tsk->flags in readonly mode for example
1725 * with tsk_used_math (like during threaded core dumping).
1726 * There is however an exception to this rule during ptrace
1727 * or during fork: the ptracer task is allowed to write to the
1728 * child->flags of its traced child (same goes for fork, the parent
1729 * can write to the child->flags), because we're guaranteed the
1730 * child is not running and in turn not changing child->flags
1731 * at the same time the parent does it.
1732 */
1733 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1734 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1735 #define clear_used_math() clear_stopped_child_used_math(current)
1736 #define set_used_math() set_stopped_child_used_math(current)
1737 #define conditional_stopped_child_used_math(condition, child) \
1738 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1739 #define conditional_used_math(condition) \
1740 conditional_stopped_child_used_math(condition, current)
1741 #define copy_to_stopped_child_used_math(child) \
1742 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1743 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1744 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1745 #define used_math() tsk_used_math(current)
1746
1747 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1748 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1749 {
1750 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1751 flags &= ~__GFP_IO;
1752 return flags;
1753 }
1754
1755 static inline unsigned int memalloc_noio_save(void)
1756 {
1757 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1758 current->flags |= PF_MEMALLOC_NOIO;
1759 return flags;
1760 }
1761
1762 static inline void memalloc_noio_restore(unsigned int flags)
1763 {
1764 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1765 }
1766
1767 /*
1768 * task->jobctl flags
1769 */
1770 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1771
1772 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1773 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1774 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1775 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1776 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1777 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1778 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1779
1780 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1781 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1782 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1783 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1784 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1785 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1786 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1787
1788 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1789 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1790
1791 extern bool task_set_jobctl_pending(struct task_struct *task,
1792 unsigned int mask);
1793 extern void task_clear_jobctl_trapping(struct task_struct *task);
1794 extern void task_clear_jobctl_pending(struct task_struct *task,
1795 unsigned int mask);
1796
1797 #ifdef CONFIG_PREEMPT_RCU
1798
1799 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1800 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1801
1802 static inline void rcu_copy_process(struct task_struct *p)
1803 {
1804 p->rcu_read_lock_nesting = 0;
1805 p->rcu_read_unlock_special = 0;
1806 #ifdef CONFIG_TREE_PREEMPT_RCU
1807 p->rcu_blocked_node = NULL;
1808 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1809 #ifdef CONFIG_RCU_BOOST
1810 p->rcu_boost_mutex = NULL;
1811 #endif /* #ifdef CONFIG_RCU_BOOST */
1812 INIT_LIST_HEAD(&p->rcu_node_entry);
1813 }
1814
1815 #else
1816
1817 static inline void rcu_copy_process(struct task_struct *p)
1818 {
1819 }
1820
1821 #endif
1822
1823 static inline void tsk_restore_flags(struct task_struct *task,
1824 unsigned long orig_flags, unsigned long flags)
1825 {
1826 task->flags &= ~flags;
1827 task->flags |= orig_flags & flags;
1828 }
1829
1830 #ifdef CONFIG_SMP
1831 extern void do_set_cpus_allowed(struct task_struct *p,
1832 const struct cpumask *new_mask);
1833
1834 extern int set_cpus_allowed_ptr(struct task_struct *p,
1835 const struct cpumask *new_mask);
1836 #else
1837 static inline void do_set_cpus_allowed(struct task_struct *p,
1838 const struct cpumask *new_mask)
1839 {
1840 }
1841 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1842 const struct cpumask *new_mask)
1843 {
1844 if (!cpumask_test_cpu(0, new_mask))
1845 return -EINVAL;
1846 return 0;
1847 }
1848 #endif
1849
1850 #ifdef CONFIG_NO_HZ_COMMON
1851 void calc_load_enter_idle(void);
1852 void calc_load_exit_idle(void);
1853 #else
1854 static inline void calc_load_enter_idle(void) { }
1855 static inline void calc_load_exit_idle(void) { }
1856 #endif /* CONFIG_NO_HZ_COMMON */
1857
1858 #ifndef CONFIG_CPUMASK_OFFSTACK
1859 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1860 {
1861 return set_cpus_allowed_ptr(p, &new_mask);
1862 }
1863 #endif
1864
1865 /*
1866 * Do not use outside of architecture code which knows its limitations.
1867 *
1868 * sched_clock() has no promise of monotonicity or bounded drift between
1869 * CPUs, use (which you should not) requires disabling IRQs.
1870 *
1871 * Please use one of the three interfaces below.
1872 */
1873 extern unsigned long long notrace sched_clock(void);
1874 /*
1875 * See the comment in kernel/sched/clock.c
1876 */
1877 extern u64 cpu_clock(int cpu);
1878 extern u64 local_clock(void);
1879 extern u64 sched_clock_cpu(int cpu);
1880
1881
1882 extern void sched_clock_init(void);
1883
1884 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1885 static inline void sched_clock_tick(void)
1886 {
1887 }
1888
1889 static inline void sched_clock_idle_sleep_event(void)
1890 {
1891 }
1892
1893 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1894 {
1895 }
1896 #else
1897 /*
1898 * Architectures can set this to 1 if they have specified
1899 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1900 * but then during bootup it turns out that sched_clock()
1901 * is reliable after all:
1902 */
1903 extern int sched_clock_stable;
1904
1905 extern void sched_clock_tick(void);
1906 extern void sched_clock_idle_sleep_event(void);
1907 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1908 #endif
1909
1910 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1911 /*
1912 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1913 * The reason for this explicit opt-in is not to have perf penalty with
1914 * slow sched_clocks.
1915 */
1916 extern void enable_sched_clock_irqtime(void);
1917 extern void disable_sched_clock_irqtime(void);
1918 #else
1919 static inline void enable_sched_clock_irqtime(void) {}
1920 static inline void disable_sched_clock_irqtime(void) {}
1921 #endif
1922
1923 extern unsigned long long
1924 task_sched_runtime(struct task_struct *task);
1925
1926 /* sched_exec is called by processes performing an exec */
1927 #ifdef CONFIG_SMP
1928 extern void sched_exec(void);
1929 #else
1930 #define sched_exec() {}
1931 #endif
1932
1933 extern void sched_clock_idle_sleep_event(void);
1934 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1935
1936 #ifdef CONFIG_HOTPLUG_CPU
1937 extern void idle_task_exit(void);
1938 #else
1939 static inline void idle_task_exit(void) {}
1940 #endif
1941
1942 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
1943 extern void wake_up_nohz_cpu(int cpu);
1944 #else
1945 static inline void wake_up_nohz_cpu(int cpu) { }
1946 #endif
1947
1948 #ifdef CONFIG_NO_HZ_FULL
1949 extern bool sched_can_stop_tick(void);
1950 extern u64 scheduler_tick_max_deferment(void);
1951 #else
1952 static inline bool sched_can_stop_tick(void) { return false; }
1953 #endif
1954
1955 #ifdef CONFIG_SCHED_AUTOGROUP
1956 extern void sched_autogroup_create_attach(struct task_struct *p);
1957 extern void sched_autogroup_detach(struct task_struct *p);
1958 extern void sched_autogroup_fork(struct signal_struct *sig);
1959 extern void sched_autogroup_exit(struct signal_struct *sig);
1960 #ifdef CONFIG_PROC_FS
1961 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1962 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
1963 #endif
1964 #else
1965 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1966 static inline void sched_autogroup_detach(struct task_struct *p) { }
1967 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1968 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1969 #endif
1970
1971 extern bool yield_to(struct task_struct *p, bool preempt);
1972 extern void set_user_nice(struct task_struct *p, long nice);
1973 extern int task_prio(const struct task_struct *p);
1974 extern int task_nice(const struct task_struct *p);
1975 extern int can_nice(const struct task_struct *p, const int nice);
1976 extern int task_curr(const struct task_struct *p);
1977 extern int idle_cpu(int cpu);
1978 extern int sched_setscheduler(struct task_struct *, int,
1979 const struct sched_param *);
1980 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1981 const struct sched_param *);
1982 extern struct task_struct *idle_task(int cpu);
1983 /**
1984 * is_idle_task - is the specified task an idle task?
1985 * @p: the task in question.
1986 *
1987 * Return: 1 if @p is an idle task. 0 otherwise.
1988 */
1989 static inline bool is_idle_task(const struct task_struct *p)
1990 {
1991 return p->pid == 0;
1992 }
1993 extern struct task_struct *curr_task(int cpu);
1994 extern void set_curr_task(int cpu, struct task_struct *p);
1995
1996 void yield(void);
1997
1998 /*
1999 * The default (Linux) execution domain.
2000 */
2001 extern struct exec_domain default_exec_domain;
2002
2003 union thread_union {
2004 struct thread_info thread_info;
2005 unsigned long stack[THREAD_SIZE/sizeof(long)];
2006 };
2007
2008 #ifndef __HAVE_ARCH_KSTACK_END
2009 static inline int kstack_end(void *addr)
2010 {
2011 /* Reliable end of stack detection:
2012 * Some APM bios versions misalign the stack
2013 */
2014 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2015 }
2016 #endif
2017
2018 extern union thread_union init_thread_union;
2019 extern struct task_struct init_task;
2020
2021 extern struct mm_struct init_mm;
2022
2023 extern struct pid_namespace init_pid_ns;
2024
2025 /*
2026 * find a task by one of its numerical ids
2027 *
2028 * find_task_by_pid_ns():
2029 * finds a task by its pid in the specified namespace
2030 * find_task_by_vpid():
2031 * finds a task by its virtual pid
2032 *
2033 * see also find_vpid() etc in include/linux/pid.h
2034 */
2035
2036 extern struct task_struct *find_task_by_vpid(pid_t nr);
2037 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2038 struct pid_namespace *ns);
2039
2040 /* per-UID process charging. */
2041 extern struct user_struct * alloc_uid(kuid_t);
2042 static inline struct user_struct *get_uid(struct user_struct *u)
2043 {
2044 atomic_inc(&u->__count);
2045 return u;
2046 }
2047 extern void free_uid(struct user_struct *);
2048
2049 #include <asm/current.h>
2050
2051 extern void xtime_update(unsigned long ticks);
2052
2053 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2054 extern int wake_up_process(struct task_struct *tsk);
2055 extern void wake_up_new_task(struct task_struct *tsk);
2056 #ifdef CONFIG_SMP
2057 extern void kick_process(struct task_struct *tsk);
2058 #else
2059 static inline void kick_process(struct task_struct *tsk) { }
2060 #endif
2061 extern void sched_fork(unsigned long clone_flags, struct task_struct *p);
2062 extern void sched_dead(struct task_struct *p);
2063
2064 extern void proc_caches_init(void);
2065 extern void flush_signals(struct task_struct *);
2066 extern void __flush_signals(struct task_struct *);
2067 extern void ignore_signals(struct task_struct *);
2068 extern void flush_signal_handlers(struct task_struct *, int force_default);
2069 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2070
2071 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2072 {
2073 unsigned long flags;
2074 int ret;
2075
2076 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2077 ret = dequeue_signal(tsk, mask, info);
2078 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2079
2080 return ret;
2081 }
2082
2083 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2084 sigset_t *mask);
2085 extern void unblock_all_signals(void);
2086 extern void release_task(struct task_struct * p);
2087 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2088 extern int force_sigsegv(int, struct task_struct *);
2089 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2090 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2091 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2092 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2093 const struct cred *, u32);
2094 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2095 extern int kill_pid(struct pid *pid, int sig, int priv);
2096 extern int kill_proc_info(int, struct siginfo *, pid_t);
2097 extern __must_check bool do_notify_parent(struct task_struct *, int);
2098 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2099 extern void force_sig(int, struct task_struct *);
2100 extern int send_sig(int, struct task_struct *, int);
2101 extern int zap_other_threads(struct task_struct *p);
2102 extern struct sigqueue *sigqueue_alloc(void);
2103 extern void sigqueue_free(struct sigqueue *);
2104 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2105 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2106
2107 static inline void restore_saved_sigmask(void)
2108 {
2109 if (test_and_clear_restore_sigmask())
2110 __set_current_blocked(¤t->saved_sigmask);
2111 }
2112
2113 static inline sigset_t *sigmask_to_save(void)
2114 {
2115 sigset_t *res = ¤t->blocked;
2116 if (unlikely(test_restore_sigmask()))
2117 res = ¤t->saved_sigmask;
2118 return res;
2119 }
2120
2121 static inline int kill_cad_pid(int sig, int priv)
2122 {
2123 return kill_pid(cad_pid, sig, priv);
2124 }
2125
2126 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2127 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2128 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2129 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2130
2131 /*
2132 * True if we are on the alternate signal stack.
2133 */
2134 static inline int on_sig_stack(unsigned long sp)
2135 {
2136 #ifdef CONFIG_STACK_GROWSUP
2137 return sp >= current->sas_ss_sp &&
2138 sp - current->sas_ss_sp < current->sas_ss_size;
2139 #else
2140 return sp > current->sas_ss_sp &&
2141 sp - current->sas_ss_sp <= current->sas_ss_size;
2142 #endif
2143 }
2144
2145 static inline int sas_ss_flags(unsigned long sp)
2146 {
2147 return (current->sas_ss_size == 0 ? SS_DISABLE
2148 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2149 }
2150
2151 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2152 {
2153 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2154 #ifdef CONFIG_STACK_GROWSUP
2155 return current->sas_ss_sp;
2156 #else
2157 return current->sas_ss_sp + current->sas_ss_size;
2158 #endif
2159 return sp;
2160 }
2161
2162 /*
2163 * Routines for handling mm_structs
2164 */
2165 extern struct mm_struct * mm_alloc(void);
2166
2167 /* mmdrop drops the mm and the page tables */
2168 extern void __mmdrop(struct mm_struct *);
2169 static inline void mmdrop(struct mm_struct * mm)
2170 {
2171 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2172 __mmdrop(mm);
2173 }
2174
2175 /* mmput gets rid of the mappings and all user-space */
2176 extern void mmput(struct mm_struct *);
2177 /* Grab a reference to a task's mm, if it is not already going away */
2178 extern struct mm_struct *get_task_mm(struct task_struct *task);
2179 /*
2180 * Grab a reference to a task's mm, if it is not already going away
2181 * and ptrace_may_access with the mode parameter passed to it
2182 * succeeds.
2183 */
2184 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2185 /* Remove the current tasks stale references to the old mm_struct */
2186 extern void mm_release(struct task_struct *, struct mm_struct *);
2187 /* Allocate a new mm structure and copy contents from tsk->mm */
2188 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2189
2190 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2191 struct task_struct *);
2192 extern void flush_thread(void);
2193 extern void exit_thread(void);
2194
2195 extern void exit_files(struct task_struct *);
2196 extern void __cleanup_sighand(struct sighand_struct *);
2197
2198 extern void exit_itimers(struct signal_struct *);
2199 extern void flush_itimer_signals(void);
2200
2201 extern void do_group_exit(int);
2202
2203 extern int allow_signal(int);
2204 extern int disallow_signal(int);
2205
2206 extern int do_execve(const char *,
2207 const char __user * const __user *,
2208 const char __user * const __user *);
2209 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2210 struct task_struct *fork_idle(int);
2211 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2212
2213 extern void set_task_comm(struct task_struct *tsk, char *from);
2214 extern char *get_task_comm(char *to, struct task_struct *tsk);
2215
2216 #ifdef CONFIG_SMP
2217 void scheduler_ipi(void);
2218 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2219 #else
2220 static inline void scheduler_ipi(void) { }
2221 static inline unsigned long wait_task_inactive(struct task_struct *p,
2222 long match_state)
2223 {
2224 return 1;
2225 }
2226 #endif
2227
2228 #define next_task(p) \
2229 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2230
2231 #define for_each_process(p) \
2232 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2233
2234 extern bool current_is_single_threaded(void);
2235
2236 /*
2237 * Careful: do_each_thread/while_each_thread is a double loop so
2238 * 'break' will not work as expected - use goto instead.
2239 */
2240 #define do_each_thread(g, t) \
2241 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2242
2243 #define while_each_thread(g, t) \
2244 while ((t = next_thread(t)) != g)
2245
2246 #define __for_each_thread(signal, t) \
2247 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2248
2249 #define for_each_thread(p, t) \
2250 __for_each_thread((p)->signal, t)
2251
2252 /* Careful: this is a double loop, 'break' won't work as expected. */
2253 #define for_each_process_thread(p, t) \
2254 for_each_process(p) for_each_thread(p, t)
2255
2256 static inline int get_nr_threads(struct task_struct *tsk)
2257 {
2258 return tsk->signal->nr_threads;
2259 }
2260
2261 static inline bool thread_group_leader(struct task_struct *p)
2262 {
2263 return p->exit_signal >= 0;
2264 }
2265
2266 /* Do to the insanities of de_thread it is possible for a process
2267 * to have the pid of the thread group leader without actually being
2268 * the thread group leader. For iteration through the pids in proc
2269 * all we care about is that we have a task with the appropriate
2270 * pid, we don't actually care if we have the right task.
2271 */
2272 static inline bool has_group_leader_pid(struct task_struct *p)
2273 {
2274 return task_pid(p) == p->signal->leader_pid;
2275 }
2276
2277 static inline
2278 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2279 {
2280 return p1->signal == p2->signal;
2281 }
2282
2283 static inline struct task_struct *next_thread(const struct task_struct *p)
2284 {
2285 return list_entry_rcu(p->thread_group.next,
2286 struct task_struct, thread_group);
2287 }
2288
2289 static inline int thread_group_empty(struct task_struct *p)
2290 {
2291 return list_empty(&p->thread_group);
2292 }
2293
2294 #define delay_group_leader(p) \
2295 (thread_group_leader(p) && !thread_group_empty(p))
2296
2297 /*
2298 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2299 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2300 * pins the final release of task.io_context. Also protects ->cpuset and
2301 * ->cgroup.subsys[]. And ->vfork_done.
2302 *
2303 * Nests both inside and outside of read_lock(&tasklist_lock).
2304 * It must not be nested with write_lock_irq(&tasklist_lock),
2305 * neither inside nor outside.
2306 */
2307 static inline void task_lock(struct task_struct *p)
2308 {
2309 spin_lock(&p->alloc_lock);
2310 }
2311
2312 static inline void task_unlock(struct task_struct *p)
2313 {
2314 spin_unlock(&p->alloc_lock);
2315 }
2316
2317 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2318 unsigned long *flags);
2319
2320 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2321 unsigned long *flags)
2322 {
2323 struct sighand_struct *ret;
2324
2325 ret = __lock_task_sighand(tsk, flags);
2326 (void)__cond_lock(&tsk->sighand->siglock, ret);
2327 return ret;
2328 }
2329
2330 static inline void unlock_task_sighand(struct task_struct *tsk,
2331 unsigned long *flags)
2332 {
2333 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2334 }
2335
2336 #ifdef CONFIG_CGROUPS
2337 static inline void threadgroup_change_begin(struct task_struct *tsk)
2338 {
2339 down_read(&tsk->signal->group_rwsem);
2340 }
2341 static inline void threadgroup_change_end(struct task_struct *tsk)
2342 {
2343 up_read(&tsk->signal->group_rwsem);
2344 }
2345
2346 /**
2347 * threadgroup_lock - lock threadgroup
2348 * @tsk: member task of the threadgroup to lock
2349 *
2350 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2351 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2352 * change ->group_leader/pid. This is useful for cases where the threadgroup
2353 * needs to stay stable across blockable operations.
2354 *
2355 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2356 * synchronization. While held, no new task will be added to threadgroup
2357 * and no existing live task will have its PF_EXITING set.
2358 *
2359 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2360 * sub-thread becomes a new leader.
2361 */
2362 static inline void threadgroup_lock(struct task_struct *tsk)
2363 {
2364 down_write(&tsk->signal->group_rwsem);
2365 }
2366
2367 /**
2368 * threadgroup_unlock - unlock threadgroup
2369 * @tsk: member task of the threadgroup to unlock
2370 *
2371 * Reverse threadgroup_lock().
2372 */
2373 static inline void threadgroup_unlock(struct task_struct *tsk)
2374 {
2375 up_write(&tsk->signal->group_rwsem);
2376 }
2377 #else
2378 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2379 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2380 static inline void threadgroup_lock(struct task_struct *tsk) {}
2381 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2382 #endif
2383
2384 #ifndef __HAVE_THREAD_FUNCTIONS
2385
2386 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2387 #define task_stack_page(task) ((task)->stack)
2388
2389 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2390 {
2391 *task_thread_info(p) = *task_thread_info(org);
2392 task_thread_info(p)->task = p;
2393 }
2394
2395 static inline unsigned long *end_of_stack(struct task_struct *p)
2396 {
2397 return (unsigned long *)(task_thread_info(p) + 1);
2398 }
2399
2400 #endif
2401
2402 static inline int object_is_on_stack(void *obj)
2403 {
2404 void *stack = task_stack_page(current);
2405
2406 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2407 }
2408
2409 extern void thread_info_cache_init(void);
2410
2411 #ifdef CONFIG_DEBUG_STACK_USAGE
2412 static inline unsigned long stack_not_used(struct task_struct *p)
2413 {
2414 unsigned long *n = end_of_stack(p);
2415
2416 do { /* Skip over canary */
2417 n++;
2418 } while (!*n);
2419
2420 return (unsigned long)n - (unsigned long)end_of_stack(p);
2421 }
2422 #endif
2423
2424 /* set thread flags in other task's structures
2425 * - see asm/thread_info.h for TIF_xxxx flags available
2426 */
2427 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2428 {
2429 set_ti_thread_flag(task_thread_info(tsk), flag);
2430 }
2431
2432 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2433 {
2434 clear_ti_thread_flag(task_thread_info(tsk), flag);
2435 }
2436
2437 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2438 {
2439 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2440 }
2441
2442 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2443 {
2444 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2445 }
2446
2447 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2448 {
2449 return test_ti_thread_flag(task_thread_info(tsk), flag);
2450 }
2451
2452 static inline void set_tsk_need_resched(struct task_struct *tsk)
2453 {
2454 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2455 }
2456
2457 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2458 {
2459 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2460 }
2461
2462 static inline int test_tsk_need_resched(struct task_struct *tsk)
2463 {
2464 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2465 }
2466
2467 static inline int restart_syscall(void)
2468 {
2469 set_tsk_thread_flag(current, TIF_SIGPENDING);
2470 return -ERESTARTNOINTR;
2471 }
2472
2473 static inline int signal_pending(struct task_struct *p)
2474 {
2475 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2476 }
2477
2478 static inline int __fatal_signal_pending(struct task_struct *p)
2479 {
2480 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2481 }
2482
2483 static inline int fatal_signal_pending(struct task_struct *p)
2484 {
2485 return signal_pending(p) && __fatal_signal_pending(p);
2486 }
2487
2488 static inline int signal_pending_state(long state, struct task_struct *p)
2489 {
2490 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2491 return 0;
2492 if (!signal_pending(p))
2493 return 0;
2494
2495 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2496 }
2497
2498 /*
2499 * cond_resched() and cond_resched_lock(): latency reduction via
2500 * explicit rescheduling in places that are safe. The return
2501 * value indicates whether a reschedule was done in fact.
2502 * cond_resched_lock() will drop the spinlock before scheduling,
2503 * cond_resched_softirq() will enable bhs before scheduling.
2504 */
2505 extern int _cond_resched(void);
2506
2507 #define cond_resched() ({ \
2508 __might_sleep(__FILE__, __LINE__, 0); \
2509 _cond_resched(); \
2510 })
2511
2512 extern int __cond_resched_lock(spinlock_t *lock);
2513
2514 #ifdef CONFIG_PREEMPT_COUNT
2515 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2516 #else
2517 #define PREEMPT_LOCK_OFFSET 0
2518 #endif
2519
2520 #define cond_resched_lock(lock) ({ \
2521 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2522 __cond_resched_lock(lock); \
2523 })
2524
2525 extern int __cond_resched_softirq(void);
2526
2527 #define cond_resched_softirq() ({ \
2528 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2529 __cond_resched_softirq(); \
2530 })
2531
2532 static inline void cond_resched_rcu(void)
2533 {
2534 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2535 rcu_read_unlock();
2536 cond_resched();
2537 rcu_read_lock();
2538 #endif
2539 }
2540
2541 /*
2542 * Does a critical section need to be broken due to another
2543 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2544 * but a general need for low latency)
2545 */
2546 static inline int spin_needbreak(spinlock_t *lock)
2547 {
2548 #ifdef CONFIG_PREEMPT
2549 return spin_is_contended(lock);
2550 #else
2551 return 0;
2552 #endif
2553 }
2554
2555 /*
2556 * Idle thread specific functions to determine the need_resched
2557 * polling state. We have two versions, one based on TS_POLLING in
2558 * thread_info.status and one based on TIF_POLLING_NRFLAG in
2559 * thread_info.flags
2560 */
2561 #ifdef TS_POLLING
2562 static inline int tsk_is_polling(struct task_struct *p)
2563 {
2564 return task_thread_info(p)->status & TS_POLLING;
2565 }
2566 static inline void __current_set_polling(void)
2567 {
2568 current_thread_info()->status |= TS_POLLING;
2569 }
2570
2571 static inline bool __must_check current_set_polling_and_test(void)
2572 {
2573 __current_set_polling();
2574
2575 /*
2576 * Polling state must be visible before we test NEED_RESCHED,
2577 * paired by resched_task()
2578 */
2579 smp_mb();
2580
2581 return unlikely(tif_need_resched());
2582 }
2583
2584 static inline void __current_clr_polling(void)
2585 {
2586 current_thread_info()->status &= ~TS_POLLING;
2587 }
2588
2589 static inline bool __must_check current_clr_polling_and_test(void)
2590 {
2591 __current_clr_polling();
2592
2593 /*
2594 * Polling state must be visible before we test NEED_RESCHED,
2595 * paired by resched_task()
2596 */
2597 smp_mb();
2598
2599 return unlikely(tif_need_resched());
2600 }
2601 #elif defined(TIF_POLLING_NRFLAG)
2602 static inline int tsk_is_polling(struct task_struct *p)
2603 {
2604 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2605 }
2606
2607 static inline void __current_set_polling(void)
2608 {
2609 set_thread_flag(TIF_POLLING_NRFLAG);
2610 }
2611
2612 static inline bool __must_check current_set_polling_and_test(void)
2613 {
2614 __current_set_polling();
2615
2616 /*
2617 * Polling state must be visible before we test NEED_RESCHED,
2618 * paired by resched_task()
2619 *
2620 * XXX: assumes set/clear bit are identical barrier wise.
2621 */
2622 smp_mb__after_clear_bit();
2623
2624 return unlikely(tif_need_resched());
2625 }
2626
2627 static inline void __current_clr_polling(void)
2628 {
2629 clear_thread_flag(TIF_POLLING_NRFLAG);
2630 }
2631
2632 static inline bool __must_check current_clr_polling_and_test(void)
2633 {
2634 __current_clr_polling();
2635
2636 /*
2637 * Polling state must be visible before we test NEED_RESCHED,
2638 * paired by resched_task()
2639 */
2640 smp_mb__after_clear_bit();
2641
2642 return unlikely(tif_need_resched());
2643 }
2644
2645 #else
2646 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2647 static inline void __current_set_polling(void) { }
2648 static inline void __current_clr_polling(void) { }
2649
2650 static inline bool __must_check current_set_polling_and_test(void)
2651 {
2652 return unlikely(tif_need_resched());
2653 }
2654 static inline bool __must_check current_clr_polling_and_test(void)
2655 {
2656 return unlikely(tif_need_resched());
2657 }
2658 #endif
2659
2660 static __always_inline bool need_resched(void)
2661 {
2662 return unlikely(tif_need_resched());
2663 }
2664
2665 /*
2666 * Thread group CPU time accounting.
2667 */
2668 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2669 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2670
2671 static inline void thread_group_cputime_init(struct signal_struct *sig)
2672 {
2673 raw_spin_lock_init(&sig->cputimer.lock);
2674 }
2675
2676 /*
2677 * Reevaluate whether the task has signals pending delivery.
2678 * Wake the task if so.
2679 * This is required every time the blocked sigset_t changes.
2680 * callers must hold sighand->siglock.
2681 */
2682 extern void recalc_sigpending_and_wake(struct task_struct *t);
2683 extern void recalc_sigpending(void);
2684
2685 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2686
2687 static inline void signal_wake_up(struct task_struct *t, bool resume)
2688 {
2689 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2690 }
2691 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2692 {
2693 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2694 }
2695
2696 /*
2697 * Wrappers for p->thread_info->cpu access. No-op on UP.
2698 */
2699 #ifdef CONFIG_SMP
2700
2701 static inline unsigned int task_cpu(const struct task_struct *p)
2702 {
2703 return task_thread_info(p)->cpu;
2704 }
2705
2706 static inline int task_node(const struct task_struct *p)
2707 {
2708 return cpu_to_node(task_cpu(p));
2709 }
2710
2711 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2712
2713 #else
2714
2715 static inline unsigned int task_cpu(const struct task_struct *p)
2716 {
2717 return 0;
2718 }
2719
2720 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2721 {
2722 }
2723
2724 #endif /* CONFIG_SMP */
2725
2726 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2727 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2728
2729 #ifdef CONFIG_CGROUP_SCHED
2730 extern struct task_group root_task_group;
2731 #endif /* CONFIG_CGROUP_SCHED */
2732
2733 extern int task_can_switch_user(struct user_struct *up,
2734 struct task_struct *tsk);
2735
2736 #ifdef CONFIG_TASK_XACCT
2737 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2738 {
2739 tsk->ioac.rchar += amt;
2740 }
2741
2742 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2743 {
2744 tsk->ioac.wchar += amt;
2745 }
2746
2747 static inline void inc_syscr(struct task_struct *tsk)
2748 {
2749 tsk->ioac.syscr++;
2750 }
2751
2752 static inline void inc_syscw(struct task_struct *tsk)
2753 {
2754 tsk->ioac.syscw++;
2755 }
2756 #else
2757 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2758 {
2759 }
2760
2761 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2762 {
2763 }
2764
2765 static inline void inc_syscr(struct task_struct *tsk)
2766 {
2767 }
2768
2769 static inline void inc_syscw(struct task_struct *tsk)
2770 {
2771 }
2772 #endif
2773
2774 #ifndef TASK_SIZE_OF
2775 #define TASK_SIZE_OF(tsk) TASK_SIZE
2776 #endif
2777
2778 #ifdef CONFIG_MM_OWNER
2779 extern void mm_update_next_owner(struct mm_struct *mm);
2780 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2781 #else
2782 static inline void mm_update_next_owner(struct mm_struct *mm)
2783 {
2784 }
2785
2786 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2787 {
2788 }
2789 #endif /* CONFIG_MM_OWNER */
2790
2791 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2792 unsigned int limit)
2793 {
2794 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2795 }
2796
2797 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2798 unsigned int limit)
2799 {
2800 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2801 }
2802
2803 static inline unsigned long rlimit(unsigned int limit)
2804 {
2805 return task_rlimit(current, limit);
2806 }
2807
2808 static inline unsigned long rlimit_max(unsigned int limit)
2809 {
2810 return task_rlimit_max(current, limit);
2811 }
2812
2813 #endif
Diese Seite steht unter der Creative Commons Namensnennung 3.0 Unported Lizenz http://i.creativecommons.org/l/by/3.0/80x15.png