Merge branch 'for-next' of git://git.samba.org/sfrench/cifs-2.6
[deliverable/linux.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 #include <uapi/linux/sched.h>
5
6 #include <linux/sched/prio.h>
7
8
9 struct sched_param {
10 int sched_priority;
11 };
12
13 #include <asm/param.h> /* for HZ */
14
15 #include <linux/capability.h>
16 #include <linux/threads.h>
17 #include <linux/kernel.h>
18 #include <linux/types.h>
19 #include <linux/timex.h>
20 #include <linux/jiffies.h>
21 #include <linux/plist.h>
22 #include <linux/rbtree.h>
23 #include <linux/thread_info.h>
24 #include <linux/cpumask.h>
25 #include <linux/errno.h>
26 #include <linux/nodemask.h>
27 #include <linux/mm_types.h>
28 #include <linux/preempt_mask.h>
29
30 #include <asm/page.h>
31 #include <asm/ptrace.h>
32 #include <linux/cputime.h>
33
34 #include <linux/smp.h>
35 #include <linux/sem.h>
36 #include <linux/shm.h>
37 #include <linux/signal.h>
38 #include <linux/compiler.h>
39 #include <linux/completion.h>
40 #include <linux/pid.h>
41 #include <linux/percpu.h>
42 #include <linux/topology.h>
43 #include <linux/proportions.h>
44 #include <linux/seccomp.h>
45 #include <linux/rcupdate.h>
46 #include <linux/rculist.h>
47 #include <linux/rtmutex.h>
48
49 #include <linux/time.h>
50 #include <linux/param.h>
51 #include <linux/resource.h>
52 #include <linux/timer.h>
53 #include <linux/hrtimer.h>
54 #include <linux/task_io_accounting.h>
55 #include <linux/latencytop.h>
56 #include <linux/cred.h>
57 #include <linux/llist.h>
58 #include <linux/uidgid.h>
59 #include <linux/gfp.h>
60
61 #include <asm/processor.h>
62
63 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
64
65 /*
66 * Extended scheduling parameters data structure.
67 *
68 * This is needed because the original struct sched_param can not be
69 * altered without introducing ABI issues with legacy applications
70 * (e.g., in sched_getparam()).
71 *
72 * However, the possibility of specifying more than just a priority for
73 * the tasks may be useful for a wide variety of application fields, e.g.,
74 * multimedia, streaming, automation and control, and many others.
75 *
76 * This variant (sched_attr) is meant at describing a so-called
77 * sporadic time-constrained task. In such model a task is specified by:
78 * - the activation period or minimum instance inter-arrival time;
79 * - the maximum (or average, depending on the actual scheduling
80 * discipline) computation time of all instances, a.k.a. runtime;
81 * - the deadline (relative to the actual activation time) of each
82 * instance.
83 * Very briefly, a periodic (sporadic) task asks for the execution of
84 * some specific computation --which is typically called an instance--
85 * (at most) every period. Moreover, each instance typically lasts no more
86 * than the runtime and must be completed by time instant t equal to
87 * the instance activation time + the deadline.
88 *
89 * This is reflected by the actual fields of the sched_attr structure:
90 *
91 * @size size of the structure, for fwd/bwd compat.
92 *
93 * @sched_policy task's scheduling policy
94 * @sched_flags for customizing the scheduler behaviour
95 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
96 * @sched_priority task's static priority (SCHED_FIFO/RR)
97 * @sched_deadline representative of the task's deadline
98 * @sched_runtime representative of the task's runtime
99 * @sched_period representative of the task's period
100 *
101 * Given this task model, there are a multiplicity of scheduling algorithms
102 * and policies, that can be used to ensure all the tasks will make their
103 * timing constraints.
104 *
105 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
106 * only user of this new interface. More information about the algorithm
107 * available in the scheduling class file or in Documentation/.
108 */
109 struct sched_attr {
110 u32 size;
111
112 u32 sched_policy;
113 u64 sched_flags;
114
115 /* SCHED_NORMAL, SCHED_BATCH */
116 s32 sched_nice;
117
118 /* SCHED_FIFO, SCHED_RR */
119 u32 sched_priority;
120
121 /* SCHED_DEADLINE */
122 u64 sched_runtime;
123 u64 sched_deadline;
124 u64 sched_period;
125 };
126
127 struct exec_domain;
128 struct futex_pi_state;
129 struct robust_list_head;
130 struct bio_list;
131 struct fs_struct;
132 struct perf_event_context;
133 struct blk_plug;
134 struct filename;
135
136 #define VMACACHE_BITS 2
137 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
138 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
139
140 /*
141 * These are the constant used to fake the fixed-point load-average
142 * counting. Some notes:
143 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
144 * a load-average precision of 10 bits integer + 11 bits fractional
145 * - if you want to count load-averages more often, you need more
146 * precision, or rounding will get you. With 2-second counting freq,
147 * the EXP_n values would be 1981, 2034 and 2043 if still using only
148 * 11 bit fractions.
149 */
150 extern unsigned long avenrun[]; /* Load averages */
151 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
152
153 #define FSHIFT 11 /* nr of bits of precision */
154 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
155 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
156 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
157 #define EXP_5 2014 /* 1/exp(5sec/5min) */
158 #define EXP_15 2037 /* 1/exp(5sec/15min) */
159
160 #define CALC_LOAD(load,exp,n) \
161 load *= exp; \
162 load += n*(FIXED_1-exp); \
163 load >>= FSHIFT;
164
165 extern unsigned long total_forks;
166 extern int nr_threads;
167 DECLARE_PER_CPU(unsigned long, process_counts);
168 extern int nr_processes(void);
169 extern unsigned long nr_running(void);
170 extern unsigned long nr_iowait(void);
171 extern unsigned long nr_iowait_cpu(int cpu);
172 extern unsigned long this_cpu_load(void);
173
174
175 extern void calc_global_load(unsigned long ticks);
176 extern void update_cpu_load_nohz(void);
177
178 extern unsigned long get_parent_ip(unsigned long addr);
179
180 extern void dump_cpu_task(int cpu);
181
182 struct seq_file;
183 struct cfs_rq;
184 struct task_group;
185 #ifdef CONFIG_SCHED_DEBUG
186 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
187 extern void proc_sched_set_task(struct task_struct *p);
188 extern void
189 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
190 #endif
191
192 /*
193 * Task state bitmask. NOTE! These bits are also
194 * encoded in fs/proc/array.c: get_task_state().
195 *
196 * We have two separate sets of flags: task->state
197 * is about runnability, while task->exit_state are
198 * about the task exiting. Confusing, but this way
199 * modifying one set can't modify the other one by
200 * mistake.
201 */
202 #define TASK_RUNNING 0
203 #define TASK_INTERRUPTIBLE 1
204 #define TASK_UNINTERRUPTIBLE 2
205 #define __TASK_STOPPED 4
206 #define __TASK_TRACED 8
207 /* in tsk->exit_state */
208 #define EXIT_DEAD 16
209 #define EXIT_ZOMBIE 32
210 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
211 /* in tsk->state again */
212 #define TASK_DEAD 64
213 #define TASK_WAKEKILL 128
214 #define TASK_WAKING 256
215 #define TASK_PARKED 512
216 #define TASK_STATE_MAX 1024
217
218 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
219
220 extern char ___assert_task_state[1 - 2*!!(
221 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
222
223 /* Convenience macros for the sake of set_task_state */
224 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
225 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
226 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
227
228 /* Convenience macros for the sake of wake_up */
229 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
230 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
231
232 /* get_task_state() */
233 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
234 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
235 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
236
237 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
238 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
239 #define task_is_stopped_or_traced(task) \
240 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
241 #define task_contributes_to_load(task) \
242 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
243 (task->flags & PF_FROZEN) == 0)
244
245 #define __set_task_state(tsk, state_value) \
246 do { (tsk)->state = (state_value); } while (0)
247 #define set_task_state(tsk, state_value) \
248 set_mb((tsk)->state, (state_value))
249
250 /*
251 * set_current_state() includes a barrier so that the write of current->state
252 * is correctly serialised wrt the caller's subsequent test of whether to
253 * actually sleep:
254 *
255 * set_current_state(TASK_UNINTERRUPTIBLE);
256 * if (do_i_need_to_sleep())
257 * schedule();
258 *
259 * If the caller does not need such serialisation then use __set_current_state()
260 */
261 #define __set_current_state(state_value) \
262 do { current->state = (state_value); } while (0)
263 #define set_current_state(state_value) \
264 set_mb(current->state, (state_value))
265
266 /* Task command name length */
267 #define TASK_COMM_LEN 16
268
269 #include <linux/spinlock.h>
270
271 /*
272 * This serializes "schedule()" and also protects
273 * the run-queue from deletions/modifications (but
274 * _adding_ to the beginning of the run-queue has
275 * a separate lock).
276 */
277 extern rwlock_t tasklist_lock;
278 extern spinlock_t mmlist_lock;
279
280 struct task_struct;
281
282 #ifdef CONFIG_PROVE_RCU
283 extern int lockdep_tasklist_lock_is_held(void);
284 #endif /* #ifdef CONFIG_PROVE_RCU */
285
286 extern void sched_init(void);
287 extern void sched_init_smp(void);
288 extern asmlinkage void schedule_tail(struct task_struct *prev);
289 extern void init_idle(struct task_struct *idle, int cpu);
290 extern void init_idle_bootup_task(struct task_struct *idle);
291
292 extern int runqueue_is_locked(int cpu);
293
294 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
295 extern void nohz_balance_enter_idle(int cpu);
296 extern void set_cpu_sd_state_idle(void);
297 extern int get_nohz_timer_target(int pinned);
298 #else
299 static inline void nohz_balance_enter_idle(int cpu) { }
300 static inline void set_cpu_sd_state_idle(void) { }
301 static inline int get_nohz_timer_target(int pinned)
302 {
303 return smp_processor_id();
304 }
305 #endif
306
307 /*
308 * Only dump TASK_* tasks. (0 for all tasks)
309 */
310 extern void show_state_filter(unsigned long state_filter);
311
312 static inline void show_state(void)
313 {
314 show_state_filter(0);
315 }
316
317 extern void show_regs(struct pt_regs *);
318
319 /*
320 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
321 * task), SP is the stack pointer of the first frame that should be shown in the back
322 * trace (or NULL if the entire call-chain of the task should be shown).
323 */
324 extern void show_stack(struct task_struct *task, unsigned long *sp);
325
326 void io_schedule(void);
327 long io_schedule_timeout(long timeout);
328
329 extern void cpu_init (void);
330 extern void trap_init(void);
331 extern void update_process_times(int user);
332 extern void scheduler_tick(void);
333
334 extern void sched_show_task(struct task_struct *p);
335
336 #ifdef CONFIG_LOCKUP_DETECTOR
337 extern void touch_softlockup_watchdog(void);
338 extern void touch_softlockup_watchdog_sync(void);
339 extern void touch_all_softlockup_watchdogs(void);
340 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
341 void __user *buffer,
342 size_t *lenp, loff_t *ppos);
343 extern unsigned int softlockup_panic;
344 void lockup_detector_init(void);
345 #else
346 static inline void touch_softlockup_watchdog(void)
347 {
348 }
349 static inline void touch_softlockup_watchdog_sync(void)
350 {
351 }
352 static inline void touch_all_softlockup_watchdogs(void)
353 {
354 }
355 static inline void lockup_detector_init(void)
356 {
357 }
358 #endif
359
360 #ifdef CONFIG_DETECT_HUNG_TASK
361 void reset_hung_task_detector(void);
362 #else
363 static inline void reset_hung_task_detector(void)
364 {
365 }
366 #endif
367
368 /* Attach to any functions which should be ignored in wchan output. */
369 #define __sched __attribute__((__section__(".sched.text")))
370
371 /* Linker adds these: start and end of __sched functions */
372 extern char __sched_text_start[], __sched_text_end[];
373
374 /* Is this address in the __sched functions? */
375 extern int in_sched_functions(unsigned long addr);
376
377 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
378 extern signed long schedule_timeout(signed long timeout);
379 extern signed long schedule_timeout_interruptible(signed long timeout);
380 extern signed long schedule_timeout_killable(signed long timeout);
381 extern signed long schedule_timeout_uninterruptible(signed long timeout);
382 asmlinkage void schedule(void);
383 extern void schedule_preempt_disabled(void);
384
385 struct nsproxy;
386 struct user_namespace;
387
388 #ifdef CONFIG_MMU
389 extern void arch_pick_mmap_layout(struct mm_struct *mm);
390 extern unsigned long
391 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
392 unsigned long, unsigned long);
393 extern unsigned long
394 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
395 unsigned long len, unsigned long pgoff,
396 unsigned long flags);
397 #else
398 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #endif
400
401 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
402 #define SUID_DUMP_USER 1 /* Dump as user of process */
403 #define SUID_DUMP_ROOT 2 /* Dump as root */
404
405 /* mm flags */
406
407 /* for SUID_DUMP_* above */
408 #define MMF_DUMPABLE_BITS 2
409 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
410
411 extern void set_dumpable(struct mm_struct *mm, int value);
412 /*
413 * This returns the actual value of the suid_dumpable flag. For things
414 * that are using this for checking for privilege transitions, it must
415 * test against SUID_DUMP_USER rather than treating it as a boolean
416 * value.
417 */
418 static inline int __get_dumpable(unsigned long mm_flags)
419 {
420 return mm_flags & MMF_DUMPABLE_MASK;
421 }
422
423 static inline int get_dumpable(struct mm_struct *mm)
424 {
425 return __get_dumpable(mm->flags);
426 }
427
428 /* coredump filter bits */
429 #define MMF_DUMP_ANON_PRIVATE 2
430 #define MMF_DUMP_ANON_SHARED 3
431 #define MMF_DUMP_MAPPED_PRIVATE 4
432 #define MMF_DUMP_MAPPED_SHARED 5
433 #define MMF_DUMP_ELF_HEADERS 6
434 #define MMF_DUMP_HUGETLB_PRIVATE 7
435 #define MMF_DUMP_HUGETLB_SHARED 8
436
437 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
438 #define MMF_DUMP_FILTER_BITS 7
439 #define MMF_DUMP_FILTER_MASK \
440 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
441 #define MMF_DUMP_FILTER_DEFAULT \
442 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
443 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
444
445 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
446 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
447 #else
448 # define MMF_DUMP_MASK_DEFAULT_ELF 0
449 #endif
450 /* leave room for more dump flags */
451 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
452 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
453 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
454
455 #define MMF_HAS_UPROBES 19 /* has uprobes */
456 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
457
458 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
459
460 struct sighand_struct {
461 atomic_t count;
462 struct k_sigaction action[_NSIG];
463 spinlock_t siglock;
464 wait_queue_head_t signalfd_wqh;
465 };
466
467 struct pacct_struct {
468 int ac_flag;
469 long ac_exitcode;
470 unsigned long ac_mem;
471 cputime_t ac_utime, ac_stime;
472 unsigned long ac_minflt, ac_majflt;
473 };
474
475 struct cpu_itimer {
476 cputime_t expires;
477 cputime_t incr;
478 u32 error;
479 u32 incr_error;
480 };
481
482 /**
483 * struct cputime - snaphsot of system and user cputime
484 * @utime: time spent in user mode
485 * @stime: time spent in system mode
486 *
487 * Gathers a generic snapshot of user and system time.
488 */
489 struct cputime {
490 cputime_t utime;
491 cputime_t stime;
492 };
493
494 /**
495 * struct task_cputime - collected CPU time counts
496 * @utime: time spent in user mode, in &cputime_t units
497 * @stime: time spent in kernel mode, in &cputime_t units
498 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
499 *
500 * This is an extension of struct cputime that includes the total runtime
501 * spent by the task from the scheduler point of view.
502 *
503 * As a result, this structure groups together three kinds of CPU time
504 * that are tracked for threads and thread groups. Most things considering
505 * CPU time want to group these counts together and treat all three
506 * of them in parallel.
507 */
508 struct task_cputime {
509 cputime_t utime;
510 cputime_t stime;
511 unsigned long long sum_exec_runtime;
512 };
513 /* Alternate field names when used to cache expirations. */
514 #define prof_exp stime
515 #define virt_exp utime
516 #define sched_exp sum_exec_runtime
517
518 #define INIT_CPUTIME \
519 (struct task_cputime) { \
520 .utime = 0, \
521 .stime = 0, \
522 .sum_exec_runtime = 0, \
523 }
524
525 #ifdef CONFIG_PREEMPT_COUNT
526 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
527 #else
528 #define PREEMPT_DISABLED PREEMPT_ENABLED
529 #endif
530
531 /*
532 * Disable preemption until the scheduler is running.
533 * Reset by start_kernel()->sched_init()->init_idle().
534 *
535 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
536 * before the scheduler is active -- see should_resched().
537 */
538 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
539
540 /**
541 * struct thread_group_cputimer - thread group interval timer counts
542 * @cputime: thread group interval timers.
543 * @running: non-zero when there are timers running and
544 * @cputime receives updates.
545 * @lock: lock for fields in this struct.
546 *
547 * This structure contains the version of task_cputime, above, that is
548 * used for thread group CPU timer calculations.
549 */
550 struct thread_group_cputimer {
551 struct task_cputime cputime;
552 int running;
553 raw_spinlock_t lock;
554 };
555
556 #include <linux/rwsem.h>
557 struct autogroup;
558
559 /*
560 * NOTE! "signal_struct" does not have its own
561 * locking, because a shared signal_struct always
562 * implies a shared sighand_struct, so locking
563 * sighand_struct is always a proper superset of
564 * the locking of signal_struct.
565 */
566 struct signal_struct {
567 atomic_t sigcnt;
568 atomic_t live;
569 int nr_threads;
570 struct list_head thread_head;
571
572 wait_queue_head_t wait_chldexit; /* for wait4() */
573
574 /* current thread group signal load-balancing target: */
575 struct task_struct *curr_target;
576
577 /* shared signal handling: */
578 struct sigpending shared_pending;
579
580 /* thread group exit support */
581 int group_exit_code;
582 /* overloaded:
583 * - notify group_exit_task when ->count is equal to notify_count
584 * - everyone except group_exit_task is stopped during signal delivery
585 * of fatal signals, group_exit_task processes the signal.
586 */
587 int notify_count;
588 struct task_struct *group_exit_task;
589
590 /* thread group stop support, overloads group_exit_code too */
591 int group_stop_count;
592 unsigned int flags; /* see SIGNAL_* flags below */
593
594 /*
595 * PR_SET_CHILD_SUBREAPER marks a process, like a service
596 * manager, to re-parent orphan (double-forking) child processes
597 * to this process instead of 'init'. The service manager is
598 * able to receive SIGCHLD signals and is able to investigate
599 * the process until it calls wait(). All children of this
600 * process will inherit a flag if they should look for a
601 * child_subreaper process at exit.
602 */
603 unsigned int is_child_subreaper:1;
604 unsigned int has_child_subreaper:1;
605
606 /* POSIX.1b Interval Timers */
607 int posix_timer_id;
608 struct list_head posix_timers;
609
610 /* ITIMER_REAL timer for the process */
611 struct hrtimer real_timer;
612 struct pid *leader_pid;
613 ktime_t it_real_incr;
614
615 /*
616 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
617 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
618 * values are defined to 0 and 1 respectively
619 */
620 struct cpu_itimer it[2];
621
622 /*
623 * Thread group totals for process CPU timers.
624 * See thread_group_cputimer(), et al, for details.
625 */
626 struct thread_group_cputimer cputimer;
627
628 /* Earliest-expiration cache. */
629 struct task_cputime cputime_expires;
630
631 struct list_head cpu_timers[3];
632
633 struct pid *tty_old_pgrp;
634
635 /* boolean value for session group leader */
636 int leader;
637
638 struct tty_struct *tty; /* NULL if no tty */
639
640 #ifdef CONFIG_SCHED_AUTOGROUP
641 struct autogroup *autogroup;
642 #endif
643 /*
644 * Cumulative resource counters for dead threads in the group,
645 * and for reaped dead child processes forked by this group.
646 * Live threads maintain their own counters and add to these
647 * in __exit_signal, except for the group leader.
648 */
649 cputime_t utime, stime, cutime, cstime;
650 cputime_t gtime;
651 cputime_t cgtime;
652 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
653 struct cputime prev_cputime;
654 #endif
655 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
656 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
657 unsigned long inblock, oublock, cinblock, coublock;
658 unsigned long maxrss, cmaxrss;
659 struct task_io_accounting ioac;
660
661 /*
662 * Cumulative ns of schedule CPU time fo dead threads in the
663 * group, not including a zombie group leader, (This only differs
664 * from jiffies_to_ns(utime + stime) if sched_clock uses something
665 * other than jiffies.)
666 */
667 unsigned long long sum_sched_runtime;
668
669 /*
670 * We don't bother to synchronize most readers of this at all,
671 * because there is no reader checking a limit that actually needs
672 * to get both rlim_cur and rlim_max atomically, and either one
673 * alone is a single word that can safely be read normally.
674 * getrlimit/setrlimit use task_lock(current->group_leader) to
675 * protect this instead of the siglock, because they really
676 * have no need to disable irqs.
677 */
678 struct rlimit rlim[RLIM_NLIMITS];
679
680 #ifdef CONFIG_BSD_PROCESS_ACCT
681 struct pacct_struct pacct; /* per-process accounting information */
682 #endif
683 #ifdef CONFIG_TASKSTATS
684 struct taskstats *stats;
685 #endif
686 #ifdef CONFIG_AUDIT
687 unsigned audit_tty;
688 unsigned audit_tty_log_passwd;
689 struct tty_audit_buf *tty_audit_buf;
690 #endif
691 #ifdef CONFIG_CGROUPS
692 /*
693 * group_rwsem prevents new tasks from entering the threadgroup and
694 * member tasks from exiting,a more specifically, setting of
695 * PF_EXITING. fork and exit paths are protected with this rwsem
696 * using threadgroup_change_begin/end(). Users which require
697 * threadgroup to remain stable should use threadgroup_[un]lock()
698 * which also takes care of exec path. Currently, cgroup is the
699 * only user.
700 */
701 struct rw_semaphore group_rwsem;
702 #endif
703
704 oom_flags_t oom_flags;
705 short oom_score_adj; /* OOM kill score adjustment */
706 short oom_score_adj_min; /* OOM kill score adjustment min value.
707 * Only settable by CAP_SYS_RESOURCE. */
708
709 struct mutex cred_guard_mutex; /* guard against foreign influences on
710 * credential calculations
711 * (notably. ptrace) */
712 };
713
714 /*
715 * Bits in flags field of signal_struct.
716 */
717 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
718 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
719 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
720 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
721 /*
722 * Pending notifications to parent.
723 */
724 #define SIGNAL_CLD_STOPPED 0x00000010
725 #define SIGNAL_CLD_CONTINUED 0x00000020
726 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
727
728 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
729
730 /* If true, all threads except ->group_exit_task have pending SIGKILL */
731 static inline int signal_group_exit(const struct signal_struct *sig)
732 {
733 return (sig->flags & SIGNAL_GROUP_EXIT) ||
734 (sig->group_exit_task != NULL);
735 }
736
737 /*
738 * Some day this will be a full-fledged user tracking system..
739 */
740 struct user_struct {
741 atomic_t __count; /* reference count */
742 atomic_t processes; /* How many processes does this user have? */
743 atomic_t sigpending; /* How many pending signals does this user have? */
744 #ifdef CONFIG_INOTIFY_USER
745 atomic_t inotify_watches; /* How many inotify watches does this user have? */
746 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
747 #endif
748 #ifdef CONFIG_FANOTIFY
749 atomic_t fanotify_listeners;
750 #endif
751 #ifdef CONFIG_EPOLL
752 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
753 #endif
754 #ifdef CONFIG_POSIX_MQUEUE
755 /* protected by mq_lock */
756 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
757 #endif
758 unsigned long locked_shm; /* How many pages of mlocked shm ? */
759
760 #ifdef CONFIG_KEYS
761 struct key *uid_keyring; /* UID specific keyring */
762 struct key *session_keyring; /* UID's default session keyring */
763 #endif
764
765 /* Hash table maintenance information */
766 struct hlist_node uidhash_node;
767 kuid_t uid;
768
769 #ifdef CONFIG_PERF_EVENTS
770 atomic_long_t locked_vm;
771 #endif
772 };
773
774 extern int uids_sysfs_init(void);
775
776 extern struct user_struct *find_user(kuid_t);
777
778 extern struct user_struct root_user;
779 #define INIT_USER (&root_user)
780
781
782 struct backing_dev_info;
783 struct reclaim_state;
784
785 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
786 struct sched_info {
787 /* cumulative counters */
788 unsigned long pcount; /* # of times run on this cpu */
789 unsigned long long run_delay; /* time spent waiting on a runqueue */
790
791 /* timestamps */
792 unsigned long long last_arrival,/* when we last ran on a cpu */
793 last_queued; /* when we were last queued to run */
794 };
795 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
796
797 #ifdef CONFIG_TASK_DELAY_ACCT
798 struct task_delay_info {
799 spinlock_t lock;
800 unsigned int flags; /* Private per-task flags */
801
802 /* For each stat XXX, add following, aligned appropriately
803 *
804 * struct timespec XXX_start, XXX_end;
805 * u64 XXX_delay;
806 * u32 XXX_count;
807 *
808 * Atomicity of updates to XXX_delay, XXX_count protected by
809 * single lock above (split into XXX_lock if contention is an issue).
810 */
811
812 /*
813 * XXX_count is incremented on every XXX operation, the delay
814 * associated with the operation is added to XXX_delay.
815 * XXX_delay contains the accumulated delay time in nanoseconds.
816 */
817 u64 blkio_start; /* Shared by blkio, swapin */
818 u64 blkio_delay; /* wait for sync block io completion */
819 u64 swapin_delay; /* wait for swapin block io completion */
820 u32 blkio_count; /* total count of the number of sync block */
821 /* io operations performed */
822 u32 swapin_count; /* total count of the number of swapin block */
823 /* io operations performed */
824
825 u64 freepages_start;
826 u64 freepages_delay; /* wait for memory reclaim */
827 u32 freepages_count; /* total count of memory reclaim */
828 };
829 #endif /* CONFIG_TASK_DELAY_ACCT */
830
831 static inline int sched_info_on(void)
832 {
833 #ifdef CONFIG_SCHEDSTATS
834 return 1;
835 #elif defined(CONFIG_TASK_DELAY_ACCT)
836 extern int delayacct_on;
837 return delayacct_on;
838 #else
839 return 0;
840 #endif
841 }
842
843 enum cpu_idle_type {
844 CPU_IDLE,
845 CPU_NOT_IDLE,
846 CPU_NEWLY_IDLE,
847 CPU_MAX_IDLE_TYPES
848 };
849
850 /*
851 * Increase resolution of cpu_capacity calculations
852 */
853 #define SCHED_CAPACITY_SHIFT 10
854 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
855
856 /*
857 * sched-domains (multiprocessor balancing) declarations:
858 */
859 #ifdef CONFIG_SMP
860 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
861 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
862 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
863 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
864 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
865 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
866 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
867 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
868 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
869 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
870 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
871 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
872 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
873 #define SD_NUMA 0x4000 /* cross-node balancing */
874
875 #ifdef CONFIG_SCHED_SMT
876 static inline int cpu_smt_flags(void)
877 {
878 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
879 }
880 #endif
881
882 #ifdef CONFIG_SCHED_MC
883 static inline int cpu_core_flags(void)
884 {
885 return SD_SHARE_PKG_RESOURCES;
886 }
887 #endif
888
889 #ifdef CONFIG_NUMA
890 static inline int cpu_numa_flags(void)
891 {
892 return SD_NUMA;
893 }
894 #endif
895
896 struct sched_domain_attr {
897 int relax_domain_level;
898 };
899
900 #define SD_ATTR_INIT (struct sched_domain_attr) { \
901 .relax_domain_level = -1, \
902 }
903
904 extern int sched_domain_level_max;
905
906 struct sched_group;
907
908 struct sched_domain {
909 /* These fields must be setup */
910 struct sched_domain *parent; /* top domain must be null terminated */
911 struct sched_domain *child; /* bottom domain must be null terminated */
912 struct sched_group *groups; /* the balancing groups of the domain */
913 unsigned long min_interval; /* Minimum balance interval ms */
914 unsigned long max_interval; /* Maximum balance interval ms */
915 unsigned int busy_factor; /* less balancing by factor if busy */
916 unsigned int imbalance_pct; /* No balance until over watermark */
917 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
918 unsigned int busy_idx;
919 unsigned int idle_idx;
920 unsigned int newidle_idx;
921 unsigned int wake_idx;
922 unsigned int forkexec_idx;
923 unsigned int smt_gain;
924
925 int nohz_idle; /* NOHZ IDLE status */
926 int flags; /* See SD_* */
927 int level;
928
929 /* Runtime fields. */
930 unsigned long last_balance; /* init to jiffies. units in jiffies */
931 unsigned int balance_interval; /* initialise to 1. units in ms. */
932 unsigned int nr_balance_failed; /* initialise to 0 */
933
934 /* idle_balance() stats */
935 u64 max_newidle_lb_cost;
936 unsigned long next_decay_max_lb_cost;
937
938 #ifdef CONFIG_SCHEDSTATS
939 /* load_balance() stats */
940 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
948
949 /* Active load balancing */
950 unsigned int alb_count;
951 unsigned int alb_failed;
952 unsigned int alb_pushed;
953
954 /* SD_BALANCE_EXEC stats */
955 unsigned int sbe_count;
956 unsigned int sbe_balanced;
957 unsigned int sbe_pushed;
958
959 /* SD_BALANCE_FORK stats */
960 unsigned int sbf_count;
961 unsigned int sbf_balanced;
962 unsigned int sbf_pushed;
963
964 /* try_to_wake_up() stats */
965 unsigned int ttwu_wake_remote;
966 unsigned int ttwu_move_affine;
967 unsigned int ttwu_move_balance;
968 #endif
969 #ifdef CONFIG_SCHED_DEBUG
970 char *name;
971 #endif
972 union {
973 void *private; /* used during construction */
974 struct rcu_head rcu; /* used during destruction */
975 };
976
977 unsigned int span_weight;
978 /*
979 * Span of all CPUs in this domain.
980 *
981 * NOTE: this field is variable length. (Allocated dynamically
982 * by attaching extra space to the end of the structure,
983 * depending on how many CPUs the kernel has booted up with)
984 */
985 unsigned long span[0];
986 };
987
988 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
989 {
990 return to_cpumask(sd->span);
991 }
992
993 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
994 struct sched_domain_attr *dattr_new);
995
996 /* Allocate an array of sched domains, for partition_sched_domains(). */
997 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
998 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
999
1000 bool cpus_share_cache(int this_cpu, int that_cpu);
1001
1002 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1003 typedef int (*sched_domain_flags_f)(void);
1004
1005 #define SDTL_OVERLAP 0x01
1006
1007 struct sd_data {
1008 struct sched_domain **__percpu sd;
1009 struct sched_group **__percpu sg;
1010 struct sched_group_capacity **__percpu sgc;
1011 };
1012
1013 struct sched_domain_topology_level {
1014 sched_domain_mask_f mask;
1015 sched_domain_flags_f sd_flags;
1016 int flags;
1017 int numa_level;
1018 struct sd_data data;
1019 #ifdef CONFIG_SCHED_DEBUG
1020 char *name;
1021 #endif
1022 };
1023
1024 extern struct sched_domain_topology_level *sched_domain_topology;
1025
1026 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1027
1028 #ifdef CONFIG_SCHED_DEBUG
1029 # define SD_INIT_NAME(type) .name = #type
1030 #else
1031 # define SD_INIT_NAME(type)
1032 #endif
1033
1034 #else /* CONFIG_SMP */
1035
1036 struct sched_domain_attr;
1037
1038 static inline void
1039 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1040 struct sched_domain_attr *dattr_new)
1041 {
1042 }
1043
1044 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1045 {
1046 return true;
1047 }
1048
1049 #endif /* !CONFIG_SMP */
1050
1051
1052 struct io_context; /* See blkdev.h */
1053
1054
1055 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1056 extern void prefetch_stack(struct task_struct *t);
1057 #else
1058 static inline void prefetch_stack(struct task_struct *t) { }
1059 #endif
1060
1061 struct audit_context; /* See audit.c */
1062 struct mempolicy;
1063 struct pipe_inode_info;
1064 struct uts_namespace;
1065
1066 struct load_weight {
1067 unsigned long weight;
1068 u32 inv_weight;
1069 };
1070
1071 struct sched_avg {
1072 /*
1073 * These sums represent an infinite geometric series and so are bound
1074 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1075 * choices of y < 1-2^(-32)*1024.
1076 */
1077 u32 runnable_avg_sum, runnable_avg_period;
1078 u64 last_runnable_update;
1079 s64 decay_count;
1080 unsigned long load_avg_contrib;
1081 };
1082
1083 #ifdef CONFIG_SCHEDSTATS
1084 struct sched_statistics {
1085 u64 wait_start;
1086 u64 wait_max;
1087 u64 wait_count;
1088 u64 wait_sum;
1089 u64 iowait_count;
1090 u64 iowait_sum;
1091
1092 u64 sleep_start;
1093 u64 sleep_max;
1094 s64 sum_sleep_runtime;
1095
1096 u64 block_start;
1097 u64 block_max;
1098 u64 exec_max;
1099 u64 slice_max;
1100
1101 u64 nr_migrations_cold;
1102 u64 nr_failed_migrations_affine;
1103 u64 nr_failed_migrations_running;
1104 u64 nr_failed_migrations_hot;
1105 u64 nr_forced_migrations;
1106
1107 u64 nr_wakeups;
1108 u64 nr_wakeups_sync;
1109 u64 nr_wakeups_migrate;
1110 u64 nr_wakeups_local;
1111 u64 nr_wakeups_remote;
1112 u64 nr_wakeups_affine;
1113 u64 nr_wakeups_affine_attempts;
1114 u64 nr_wakeups_passive;
1115 u64 nr_wakeups_idle;
1116 };
1117 #endif
1118
1119 struct sched_entity {
1120 struct load_weight load; /* for load-balancing */
1121 struct rb_node run_node;
1122 struct list_head group_node;
1123 unsigned int on_rq;
1124
1125 u64 exec_start;
1126 u64 sum_exec_runtime;
1127 u64 vruntime;
1128 u64 prev_sum_exec_runtime;
1129
1130 u64 nr_migrations;
1131
1132 #ifdef CONFIG_SCHEDSTATS
1133 struct sched_statistics statistics;
1134 #endif
1135
1136 #ifdef CONFIG_FAIR_GROUP_SCHED
1137 int depth;
1138 struct sched_entity *parent;
1139 /* rq on which this entity is (to be) queued: */
1140 struct cfs_rq *cfs_rq;
1141 /* rq "owned" by this entity/group: */
1142 struct cfs_rq *my_q;
1143 #endif
1144
1145 #ifdef CONFIG_SMP
1146 /* Per-entity load-tracking */
1147 struct sched_avg avg;
1148 #endif
1149 };
1150
1151 struct sched_rt_entity {
1152 struct list_head run_list;
1153 unsigned long timeout;
1154 unsigned long watchdog_stamp;
1155 unsigned int time_slice;
1156
1157 struct sched_rt_entity *back;
1158 #ifdef CONFIG_RT_GROUP_SCHED
1159 struct sched_rt_entity *parent;
1160 /* rq on which this entity is (to be) queued: */
1161 struct rt_rq *rt_rq;
1162 /* rq "owned" by this entity/group: */
1163 struct rt_rq *my_q;
1164 #endif
1165 };
1166
1167 struct sched_dl_entity {
1168 struct rb_node rb_node;
1169
1170 /*
1171 * Original scheduling parameters. Copied here from sched_attr
1172 * during sched_setattr(), they will remain the same until
1173 * the next sched_setattr().
1174 */
1175 u64 dl_runtime; /* maximum runtime for each instance */
1176 u64 dl_deadline; /* relative deadline of each instance */
1177 u64 dl_period; /* separation of two instances (period) */
1178 u64 dl_bw; /* dl_runtime / dl_deadline */
1179
1180 /*
1181 * Actual scheduling parameters. Initialized with the values above,
1182 * they are continously updated during task execution. Note that
1183 * the remaining runtime could be < 0 in case we are in overrun.
1184 */
1185 s64 runtime; /* remaining runtime for this instance */
1186 u64 deadline; /* absolute deadline for this instance */
1187 unsigned int flags; /* specifying the scheduler behaviour */
1188
1189 /*
1190 * Some bool flags:
1191 *
1192 * @dl_throttled tells if we exhausted the runtime. If so, the
1193 * task has to wait for a replenishment to be performed at the
1194 * next firing of dl_timer.
1195 *
1196 * @dl_new tells if a new instance arrived. If so we must
1197 * start executing it with full runtime and reset its absolute
1198 * deadline;
1199 *
1200 * @dl_boosted tells if we are boosted due to DI. If so we are
1201 * outside bandwidth enforcement mechanism (but only until we
1202 * exit the critical section);
1203 *
1204 * @dl_yielded tells if task gave up the cpu before consuming
1205 * all its available runtime during the last job.
1206 */
1207 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1208
1209 /*
1210 * Bandwidth enforcement timer. Each -deadline task has its
1211 * own bandwidth to be enforced, thus we need one timer per task.
1212 */
1213 struct hrtimer dl_timer;
1214 };
1215
1216 struct rcu_node;
1217
1218 enum perf_event_task_context {
1219 perf_invalid_context = -1,
1220 perf_hw_context = 0,
1221 perf_sw_context,
1222 perf_nr_task_contexts,
1223 };
1224
1225 struct task_struct {
1226 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1227 void *stack;
1228 atomic_t usage;
1229 unsigned int flags; /* per process flags, defined below */
1230 unsigned int ptrace;
1231
1232 #ifdef CONFIG_SMP
1233 struct llist_node wake_entry;
1234 int on_cpu;
1235 struct task_struct *last_wakee;
1236 unsigned long wakee_flips;
1237 unsigned long wakee_flip_decay_ts;
1238
1239 int wake_cpu;
1240 #endif
1241 int on_rq;
1242
1243 int prio, static_prio, normal_prio;
1244 unsigned int rt_priority;
1245 const struct sched_class *sched_class;
1246 struct sched_entity se;
1247 struct sched_rt_entity rt;
1248 #ifdef CONFIG_CGROUP_SCHED
1249 struct task_group *sched_task_group;
1250 #endif
1251 struct sched_dl_entity dl;
1252
1253 #ifdef CONFIG_PREEMPT_NOTIFIERS
1254 /* list of struct preempt_notifier: */
1255 struct hlist_head preempt_notifiers;
1256 #endif
1257
1258 #ifdef CONFIG_BLK_DEV_IO_TRACE
1259 unsigned int btrace_seq;
1260 #endif
1261
1262 unsigned int policy;
1263 int nr_cpus_allowed;
1264 cpumask_t cpus_allowed;
1265
1266 #ifdef CONFIG_PREEMPT_RCU
1267 int rcu_read_lock_nesting;
1268 char rcu_read_unlock_special;
1269 struct list_head rcu_node_entry;
1270 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1271 #ifdef CONFIG_TREE_PREEMPT_RCU
1272 struct rcu_node *rcu_blocked_node;
1273 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1274
1275 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1276 struct sched_info sched_info;
1277 #endif
1278
1279 struct list_head tasks;
1280 #ifdef CONFIG_SMP
1281 struct plist_node pushable_tasks;
1282 struct rb_node pushable_dl_tasks;
1283 #endif
1284
1285 struct mm_struct *mm, *active_mm;
1286 #ifdef CONFIG_COMPAT_BRK
1287 unsigned brk_randomized:1;
1288 #endif
1289 /* per-thread vma caching */
1290 u32 vmacache_seqnum;
1291 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1292 #if defined(SPLIT_RSS_COUNTING)
1293 struct task_rss_stat rss_stat;
1294 #endif
1295 /* task state */
1296 int exit_state;
1297 int exit_code, exit_signal;
1298 int pdeath_signal; /* The signal sent when the parent dies */
1299 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1300
1301 /* Used for emulating ABI behavior of previous Linux versions */
1302 unsigned int personality;
1303
1304 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1305 * execve */
1306 unsigned in_iowait:1;
1307
1308 /* Revert to default priority/policy when forking */
1309 unsigned sched_reset_on_fork:1;
1310 unsigned sched_contributes_to_load:1;
1311
1312 unsigned long atomic_flags; /* Flags needing atomic access. */
1313
1314 pid_t pid;
1315 pid_t tgid;
1316
1317 #ifdef CONFIG_CC_STACKPROTECTOR
1318 /* Canary value for the -fstack-protector gcc feature */
1319 unsigned long stack_canary;
1320 #endif
1321 /*
1322 * pointers to (original) parent process, youngest child, younger sibling,
1323 * older sibling, respectively. (p->father can be replaced with
1324 * p->real_parent->pid)
1325 */
1326 struct task_struct __rcu *real_parent; /* real parent process */
1327 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1328 /*
1329 * children/sibling forms the list of my natural children
1330 */
1331 struct list_head children; /* list of my children */
1332 struct list_head sibling; /* linkage in my parent's children list */
1333 struct task_struct *group_leader; /* threadgroup leader */
1334
1335 /*
1336 * ptraced is the list of tasks this task is using ptrace on.
1337 * This includes both natural children and PTRACE_ATTACH targets.
1338 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1339 */
1340 struct list_head ptraced;
1341 struct list_head ptrace_entry;
1342
1343 /* PID/PID hash table linkage. */
1344 struct pid_link pids[PIDTYPE_MAX];
1345 struct list_head thread_group;
1346 struct list_head thread_node;
1347
1348 struct completion *vfork_done; /* for vfork() */
1349 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1350 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1351
1352 cputime_t utime, stime, utimescaled, stimescaled;
1353 cputime_t gtime;
1354 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1355 struct cputime prev_cputime;
1356 #endif
1357 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1358 seqlock_t vtime_seqlock;
1359 unsigned long long vtime_snap;
1360 enum {
1361 VTIME_SLEEPING = 0,
1362 VTIME_USER,
1363 VTIME_SYS,
1364 } vtime_snap_whence;
1365 #endif
1366 unsigned long nvcsw, nivcsw; /* context switch counts */
1367 u64 start_time; /* monotonic time in nsec */
1368 u64 real_start_time; /* boot based time in nsec */
1369 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1370 unsigned long min_flt, maj_flt;
1371
1372 struct task_cputime cputime_expires;
1373 struct list_head cpu_timers[3];
1374
1375 /* process credentials */
1376 const struct cred __rcu *real_cred; /* objective and real subjective task
1377 * credentials (COW) */
1378 const struct cred __rcu *cred; /* effective (overridable) subjective task
1379 * credentials (COW) */
1380 char comm[TASK_COMM_LEN]; /* executable name excluding path
1381 - access with [gs]et_task_comm (which lock
1382 it with task_lock())
1383 - initialized normally by setup_new_exec */
1384 /* file system info */
1385 int link_count, total_link_count;
1386 #ifdef CONFIG_SYSVIPC
1387 /* ipc stuff */
1388 struct sysv_sem sysvsem;
1389 struct sysv_shm sysvshm;
1390 #endif
1391 #ifdef CONFIG_DETECT_HUNG_TASK
1392 /* hung task detection */
1393 unsigned long last_switch_count;
1394 #endif
1395 /* CPU-specific state of this task */
1396 struct thread_struct thread;
1397 /* filesystem information */
1398 struct fs_struct *fs;
1399 /* open file information */
1400 struct files_struct *files;
1401 /* namespaces */
1402 struct nsproxy *nsproxy;
1403 /* signal handlers */
1404 struct signal_struct *signal;
1405 struct sighand_struct *sighand;
1406
1407 sigset_t blocked, real_blocked;
1408 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1409 struct sigpending pending;
1410
1411 unsigned long sas_ss_sp;
1412 size_t sas_ss_size;
1413 int (*notifier)(void *priv);
1414 void *notifier_data;
1415 sigset_t *notifier_mask;
1416 struct callback_head *task_works;
1417
1418 struct audit_context *audit_context;
1419 #ifdef CONFIG_AUDITSYSCALL
1420 kuid_t loginuid;
1421 unsigned int sessionid;
1422 #endif
1423 struct seccomp seccomp;
1424
1425 /* Thread group tracking */
1426 u32 parent_exec_id;
1427 u32 self_exec_id;
1428 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1429 * mempolicy */
1430 spinlock_t alloc_lock;
1431
1432 /* Protection of the PI data structures: */
1433 raw_spinlock_t pi_lock;
1434
1435 #ifdef CONFIG_RT_MUTEXES
1436 /* PI waiters blocked on a rt_mutex held by this task */
1437 struct rb_root pi_waiters;
1438 struct rb_node *pi_waiters_leftmost;
1439 /* Deadlock detection and priority inheritance handling */
1440 struct rt_mutex_waiter *pi_blocked_on;
1441 #endif
1442
1443 #ifdef CONFIG_DEBUG_MUTEXES
1444 /* mutex deadlock detection */
1445 struct mutex_waiter *blocked_on;
1446 #endif
1447 #ifdef CONFIG_TRACE_IRQFLAGS
1448 unsigned int irq_events;
1449 unsigned long hardirq_enable_ip;
1450 unsigned long hardirq_disable_ip;
1451 unsigned int hardirq_enable_event;
1452 unsigned int hardirq_disable_event;
1453 int hardirqs_enabled;
1454 int hardirq_context;
1455 unsigned long softirq_disable_ip;
1456 unsigned long softirq_enable_ip;
1457 unsigned int softirq_disable_event;
1458 unsigned int softirq_enable_event;
1459 int softirqs_enabled;
1460 int softirq_context;
1461 #endif
1462 #ifdef CONFIG_LOCKDEP
1463 # define MAX_LOCK_DEPTH 48UL
1464 u64 curr_chain_key;
1465 int lockdep_depth;
1466 unsigned int lockdep_recursion;
1467 struct held_lock held_locks[MAX_LOCK_DEPTH];
1468 gfp_t lockdep_reclaim_gfp;
1469 #endif
1470
1471 /* journalling filesystem info */
1472 void *journal_info;
1473
1474 /* stacked block device info */
1475 struct bio_list *bio_list;
1476
1477 #ifdef CONFIG_BLOCK
1478 /* stack plugging */
1479 struct blk_plug *plug;
1480 #endif
1481
1482 /* VM state */
1483 struct reclaim_state *reclaim_state;
1484
1485 struct backing_dev_info *backing_dev_info;
1486
1487 struct io_context *io_context;
1488
1489 unsigned long ptrace_message;
1490 siginfo_t *last_siginfo; /* For ptrace use. */
1491 struct task_io_accounting ioac;
1492 #if defined(CONFIG_TASK_XACCT)
1493 u64 acct_rss_mem1; /* accumulated rss usage */
1494 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1495 cputime_t acct_timexpd; /* stime + utime since last update */
1496 #endif
1497 #ifdef CONFIG_CPUSETS
1498 nodemask_t mems_allowed; /* Protected by alloc_lock */
1499 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1500 int cpuset_mem_spread_rotor;
1501 int cpuset_slab_spread_rotor;
1502 #endif
1503 #ifdef CONFIG_CGROUPS
1504 /* Control Group info protected by css_set_lock */
1505 struct css_set __rcu *cgroups;
1506 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1507 struct list_head cg_list;
1508 #endif
1509 #ifdef CONFIG_FUTEX
1510 struct robust_list_head __user *robust_list;
1511 #ifdef CONFIG_COMPAT
1512 struct compat_robust_list_head __user *compat_robust_list;
1513 #endif
1514 struct list_head pi_state_list;
1515 struct futex_pi_state *pi_state_cache;
1516 #endif
1517 #ifdef CONFIG_PERF_EVENTS
1518 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1519 struct mutex perf_event_mutex;
1520 struct list_head perf_event_list;
1521 #endif
1522 #ifdef CONFIG_DEBUG_PREEMPT
1523 unsigned long preempt_disable_ip;
1524 #endif
1525 #ifdef CONFIG_NUMA
1526 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1527 short il_next;
1528 short pref_node_fork;
1529 #endif
1530 #ifdef CONFIG_NUMA_BALANCING
1531 int numa_scan_seq;
1532 unsigned int numa_scan_period;
1533 unsigned int numa_scan_period_max;
1534 int numa_preferred_nid;
1535 unsigned long numa_migrate_retry;
1536 u64 node_stamp; /* migration stamp */
1537 u64 last_task_numa_placement;
1538 u64 last_sum_exec_runtime;
1539 struct callback_head numa_work;
1540
1541 struct list_head numa_entry;
1542 struct numa_group *numa_group;
1543
1544 /*
1545 * Exponential decaying average of faults on a per-node basis.
1546 * Scheduling placement decisions are made based on the these counts.
1547 * The values remain static for the duration of a PTE scan
1548 */
1549 unsigned long *numa_faults_memory;
1550 unsigned long total_numa_faults;
1551
1552 /*
1553 * numa_faults_buffer records faults per node during the current
1554 * scan window. When the scan completes, the counts in
1555 * numa_faults_memory decay and these values are copied.
1556 */
1557 unsigned long *numa_faults_buffer_memory;
1558
1559 /*
1560 * Track the nodes the process was running on when a NUMA hinting
1561 * fault was incurred.
1562 */
1563 unsigned long *numa_faults_cpu;
1564 unsigned long *numa_faults_buffer_cpu;
1565
1566 /*
1567 * numa_faults_locality tracks if faults recorded during the last
1568 * scan window were remote/local. The task scan period is adapted
1569 * based on the locality of the faults with different weights
1570 * depending on whether they were shared or private faults
1571 */
1572 unsigned long numa_faults_locality[2];
1573
1574 unsigned long numa_pages_migrated;
1575 #endif /* CONFIG_NUMA_BALANCING */
1576
1577 struct rcu_head rcu;
1578
1579 /*
1580 * cache last used pipe for splice
1581 */
1582 struct pipe_inode_info *splice_pipe;
1583
1584 struct page_frag task_frag;
1585
1586 #ifdef CONFIG_TASK_DELAY_ACCT
1587 struct task_delay_info *delays;
1588 #endif
1589 #ifdef CONFIG_FAULT_INJECTION
1590 int make_it_fail;
1591 #endif
1592 /*
1593 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1594 * balance_dirty_pages() for some dirty throttling pause
1595 */
1596 int nr_dirtied;
1597 int nr_dirtied_pause;
1598 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1599
1600 #ifdef CONFIG_LATENCYTOP
1601 int latency_record_count;
1602 struct latency_record latency_record[LT_SAVECOUNT];
1603 #endif
1604 /*
1605 * time slack values; these are used to round up poll() and
1606 * select() etc timeout values. These are in nanoseconds.
1607 */
1608 unsigned long timer_slack_ns;
1609 unsigned long default_timer_slack_ns;
1610
1611 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1612 /* Index of current stored address in ret_stack */
1613 int curr_ret_stack;
1614 /* Stack of return addresses for return function tracing */
1615 struct ftrace_ret_stack *ret_stack;
1616 /* time stamp for last schedule */
1617 unsigned long long ftrace_timestamp;
1618 /*
1619 * Number of functions that haven't been traced
1620 * because of depth overrun.
1621 */
1622 atomic_t trace_overrun;
1623 /* Pause for the tracing */
1624 atomic_t tracing_graph_pause;
1625 #endif
1626 #ifdef CONFIG_TRACING
1627 /* state flags for use by tracers */
1628 unsigned long trace;
1629 /* bitmask and counter of trace recursion */
1630 unsigned long trace_recursion;
1631 #endif /* CONFIG_TRACING */
1632 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1633 unsigned int memcg_kmem_skip_account;
1634 struct memcg_oom_info {
1635 struct mem_cgroup *memcg;
1636 gfp_t gfp_mask;
1637 int order;
1638 unsigned int may_oom:1;
1639 } memcg_oom;
1640 #endif
1641 #ifdef CONFIG_UPROBES
1642 struct uprobe_task *utask;
1643 #endif
1644 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1645 unsigned int sequential_io;
1646 unsigned int sequential_io_avg;
1647 #endif
1648 };
1649
1650 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1651 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1652
1653 #define TNF_MIGRATED 0x01
1654 #define TNF_NO_GROUP 0x02
1655 #define TNF_SHARED 0x04
1656 #define TNF_FAULT_LOCAL 0x08
1657
1658 #ifdef CONFIG_NUMA_BALANCING
1659 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1660 extern pid_t task_numa_group_id(struct task_struct *p);
1661 extern void set_numabalancing_state(bool enabled);
1662 extern void task_numa_free(struct task_struct *p);
1663 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1664 int src_nid, int dst_cpu);
1665 #else
1666 static inline void task_numa_fault(int last_node, int node, int pages,
1667 int flags)
1668 {
1669 }
1670 static inline pid_t task_numa_group_id(struct task_struct *p)
1671 {
1672 return 0;
1673 }
1674 static inline void set_numabalancing_state(bool enabled)
1675 {
1676 }
1677 static inline void task_numa_free(struct task_struct *p)
1678 {
1679 }
1680 static inline bool should_numa_migrate_memory(struct task_struct *p,
1681 struct page *page, int src_nid, int dst_cpu)
1682 {
1683 return true;
1684 }
1685 #endif
1686
1687 static inline struct pid *task_pid(struct task_struct *task)
1688 {
1689 return task->pids[PIDTYPE_PID].pid;
1690 }
1691
1692 static inline struct pid *task_tgid(struct task_struct *task)
1693 {
1694 return task->group_leader->pids[PIDTYPE_PID].pid;
1695 }
1696
1697 /*
1698 * Without tasklist or rcu lock it is not safe to dereference
1699 * the result of task_pgrp/task_session even if task == current,
1700 * we can race with another thread doing sys_setsid/sys_setpgid.
1701 */
1702 static inline struct pid *task_pgrp(struct task_struct *task)
1703 {
1704 return task->group_leader->pids[PIDTYPE_PGID].pid;
1705 }
1706
1707 static inline struct pid *task_session(struct task_struct *task)
1708 {
1709 return task->group_leader->pids[PIDTYPE_SID].pid;
1710 }
1711
1712 struct pid_namespace;
1713
1714 /*
1715 * the helpers to get the task's different pids as they are seen
1716 * from various namespaces
1717 *
1718 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1719 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1720 * current.
1721 * task_xid_nr_ns() : id seen from the ns specified;
1722 *
1723 * set_task_vxid() : assigns a virtual id to a task;
1724 *
1725 * see also pid_nr() etc in include/linux/pid.h
1726 */
1727 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1728 struct pid_namespace *ns);
1729
1730 static inline pid_t task_pid_nr(struct task_struct *tsk)
1731 {
1732 return tsk->pid;
1733 }
1734
1735 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1736 struct pid_namespace *ns)
1737 {
1738 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1739 }
1740
1741 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1742 {
1743 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1744 }
1745
1746
1747 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1748 {
1749 return tsk->tgid;
1750 }
1751
1752 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1753
1754 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1755 {
1756 return pid_vnr(task_tgid(tsk));
1757 }
1758
1759
1760 static inline int pid_alive(const struct task_struct *p);
1761 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1762 {
1763 pid_t pid = 0;
1764
1765 rcu_read_lock();
1766 if (pid_alive(tsk))
1767 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1768 rcu_read_unlock();
1769
1770 return pid;
1771 }
1772
1773 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1774 {
1775 return task_ppid_nr_ns(tsk, &init_pid_ns);
1776 }
1777
1778 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1779 struct pid_namespace *ns)
1780 {
1781 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1782 }
1783
1784 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1785 {
1786 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1787 }
1788
1789
1790 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1791 struct pid_namespace *ns)
1792 {
1793 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1794 }
1795
1796 static inline pid_t task_session_vnr(struct task_struct *tsk)
1797 {
1798 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1799 }
1800
1801 /* obsolete, do not use */
1802 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1803 {
1804 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1805 }
1806
1807 /**
1808 * pid_alive - check that a task structure is not stale
1809 * @p: Task structure to be checked.
1810 *
1811 * Test if a process is not yet dead (at most zombie state)
1812 * If pid_alive fails, then pointers within the task structure
1813 * can be stale and must not be dereferenced.
1814 *
1815 * Return: 1 if the process is alive. 0 otherwise.
1816 */
1817 static inline int pid_alive(const struct task_struct *p)
1818 {
1819 return p->pids[PIDTYPE_PID].pid != NULL;
1820 }
1821
1822 /**
1823 * is_global_init - check if a task structure is init
1824 * @tsk: Task structure to be checked.
1825 *
1826 * Check if a task structure is the first user space task the kernel created.
1827 *
1828 * Return: 1 if the task structure is init. 0 otherwise.
1829 */
1830 static inline int is_global_init(struct task_struct *tsk)
1831 {
1832 return tsk->pid == 1;
1833 }
1834
1835 extern struct pid *cad_pid;
1836
1837 extern void free_task(struct task_struct *tsk);
1838 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1839
1840 extern void __put_task_struct(struct task_struct *t);
1841
1842 static inline void put_task_struct(struct task_struct *t)
1843 {
1844 if (atomic_dec_and_test(&t->usage))
1845 __put_task_struct(t);
1846 }
1847
1848 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1849 extern void task_cputime(struct task_struct *t,
1850 cputime_t *utime, cputime_t *stime);
1851 extern void task_cputime_scaled(struct task_struct *t,
1852 cputime_t *utimescaled, cputime_t *stimescaled);
1853 extern cputime_t task_gtime(struct task_struct *t);
1854 #else
1855 static inline void task_cputime(struct task_struct *t,
1856 cputime_t *utime, cputime_t *stime)
1857 {
1858 if (utime)
1859 *utime = t->utime;
1860 if (stime)
1861 *stime = t->stime;
1862 }
1863
1864 static inline void task_cputime_scaled(struct task_struct *t,
1865 cputime_t *utimescaled,
1866 cputime_t *stimescaled)
1867 {
1868 if (utimescaled)
1869 *utimescaled = t->utimescaled;
1870 if (stimescaled)
1871 *stimescaled = t->stimescaled;
1872 }
1873
1874 static inline cputime_t task_gtime(struct task_struct *t)
1875 {
1876 return t->gtime;
1877 }
1878 #endif
1879 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1880 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1881
1882 /*
1883 * Per process flags
1884 */
1885 #define PF_EXITING 0x00000004 /* getting shut down */
1886 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1887 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1888 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1889 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1890 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1891 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1892 #define PF_DUMPCORE 0x00000200 /* dumped core */
1893 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1894 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1895 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1896 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1897 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1898 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1899 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1900 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1901 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1902 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1903 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1904 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1905 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1906 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1907 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1908 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1909 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1910 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1911 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1912 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1913 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1914
1915 /*
1916 * Only the _current_ task can read/write to tsk->flags, but other
1917 * tasks can access tsk->flags in readonly mode for example
1918 * with tsk_used_math (like during threaded core dumping).
1919 * There is however an exception to this rule during ptrace
1920 * or during fork: the ptracer task is allowed to write to the
1921 * child->flags of its traced child (same goes for fork, the parent
1922 * can write to the child->flags), because we're guaranteed the
1923 * child is not running and in turn not changing child->flags
1924 * at the same time the parent does it.
1925 */
1926 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1927 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1928 #define clear_used_math() clear_stopped_child_used_math(current)
1929 #define set_used_math() set_stopped_child_used_math(current)
1930 #define conditional_stopped_child_used_math(condition, child) \
1931 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1932 #define conditional_used_math(condition) \
1933 conditional_stopped_child_used_math(condition, current)
1934 #define copy_to_stopped_child_used_math(child) \
1935 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1936 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1937 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1938 #define used_math() tsk_used_math(current)
1939
1940 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1941 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1942 {
1943 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1944 flags &= ~__GFP_IO;
1945 return flags;
1946 }
1947
1948 static inline unsigned int memalloc_noio_save(void)
1949 {
1950 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1951 current->flags |= PF_MEMALLOC_NOIO;
1952 return flags;
1953 }
1954
1955 static inline void memalloc_noio_restore(unsigned int flags)
1956 {
1957 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1958 }
1959
1960 /* Per-process atomic flags. */
1961 #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1962
1963 static inline bool task_no_new_privs(struct task_struct *p)
1964 {
1965 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1966 }
1967
1968 static inline void task_set_no_new_privs(struct task_struct *p)
1969 {
1970 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1971 }
1972
1973 /*
1974 * task->jobctl flags
1975 */
1976 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1977
1978 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1979 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1980 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1981 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1982 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1983 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1984 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1985
1986 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1987 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1988 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1989 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1990 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
1991 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
1992 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
1993
1994 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
1995 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
1996
1997 extern bool task_set_jobctl_pending(struct task_struct *task,
1998 unsigned int mask);
1999 extern void task_clear_jobctl_trapping(struct task_struct *task);
2000 extern void task_clear_jobctl_pending(struct task_struct *task,
2001 unsigned int mask);
2002
2003 #ifdef CONFIG_PREEMPT_RCU
2004
2005 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2006 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2007
2008 static inline void rcu_copy_process(struct task_struct *p)
2009 {
2010 p->rcu_read_lock_nesting = 0;
2011 p->rcu_read_unlock_special = 0;
2012 #ifdef CONFIG_TREE_PREEMPT_RCU
2013 p->rcu_blocked_node = NULL;
2014 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2015 INIT_LIST_HEAD(&p->rcu_node_entry);
2016 }
2017
2018 #else
2019
2020 static inline void rcu_copy_process(struct task_struct *p)
2021 {
2022 }
2023
2024 #endif
2025
2026 static inline void tsk_restore_flags(struct task_struct *task,
2027 unsigned long orig_flags, unsigned long flags)
2028 {
2029 task->flags &= ~flags;
2030 task->flags |= orig_flags & flags;
2031 }
2032
2033 #ifdef CONFIG_SMP
2034 extern void do_set_cpus_allowed(struct task_struct *p,
2035 const struct cpumask *new_mask);
2036
2037 extern int set_cpus_allowed_ptr(struct task_struct *p,
2038 const struct cpumask *new_mask);
2039 #else
2040 static inline void do_set_cpus_allowed(struct task_struct *p,
2041 const struct cpumask *new_mask)
2042 {
2043 }
2044 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2045 const struct cpumask *new_mask)
2046 {
2047 if (!cpumask_test_cpu(0, new_mask))
2048 return -EINVAL;
2049 return 0;
2050 }
2051 #endif
2052
2053 #ifdef CONFIG_NO_HZ_COMMON
2054 void calc_load_enter_idle(void);
2055 void calc_load_exit_idle(void);
2056 #else
2057 static inline void calc_load_enter_idle(void) { }
2058 static inline void calc_load_exit_idle(void) { }
2059 #endif /* CONFIG_NO_HZ_COMMON */
2060
2061 #ifndef CONFIG_CPUMASK_OFFSTACK
2062 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2063 {
2064 return set_cpus_allowed_ptr(p, &new_mask);
2065 }
2066 #endif
2067
2068 /*
2069 * Do not use outside of architecture code which knows its limitations.
2070 *
2071 * sched_clock() has no promise of monotonicity or bounded drift between
2072 * CPUs, use (which you should not) requires disabling IRQs.
2073 *
2074 * Please use one of the three interfaces below.
2075 */
2076 extern unsigned long long notrace sched_clock(void);
2077 /*
2078 * See the comment in kernel/sched/clock.c
2079 */
2080 extern u64 cpu_clock(int cpu);
2081 extern u64 local_clock(void);
2082 extern u64 sched_clock_cpu(int cpu);
2083
2084
2085 extern void sched_clock_init(void);
2086
2087 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2088 static inline void sched_clock_tick(void)
2089 {
2090 }
2091
2092 static inline void sched_clock_idle_sleep_event(void)
2093 {
2094 }
2095
2096 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2097 {
2098 }
2099 #else
2100 /*
2101 * Architectures can set this to 1 if they have specified
2102 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2103 * but then during bootup it turns out that sched_clock()
2104 * is reliable after all:
2105 */
2106 extern int sched_clock_stable(void);
2107 extern void set_sched_clock_stable(void);
2108 extern void clear_sched_clock_stable(void);
2109
2110 extern void sched_clock_tick(void);
2111 extern void sched_clock_idle_sleep_event(void);
2112 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2113 #endif
2114
2115 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2116 /*
2117 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2118 * The reason for this explicit opt-in is not to have perf penalty with
2119 * slow sched_clocks.
2120 */
2121 extern void enable_sched_clock_irqtime(void);
2122 extern void disable_sched_clock_irqtime(void);
2123 #else
2124 static inline void enable_sched_clock_irqtime(void) {}
2125 static inline void disable_sched_clock_irqtime(void) {}
2126 #endif
2127
2128 extern unsigned long long
2129 task_sched_runtime(struct task_struct *task);
2130
2131 /* sched_exec is called by processes performing an exec */
2132 #ifdef CONFIG_SMP
2133 extern void sched_exec(void);
2134 #else
2135 #define sched_exec() {}
2136 #endif
2137
2138 extern void sched_clock_idle_sleep_event(void);
2139 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2140
2141 #ifdef CONFIG_HOTPLUG_CPU
2142 extern void idle_task_exit(void);
2143 #else
2144 static inline void idle_task_exit(void) {}
2145 #endif
2146
2147 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2148 extern void wake_up_nohz_cpu(int cpu);
2149 #else
2150 static inline void wake_up_nohz_cpu(int cpu) { }
2151 #endif
2152
2153 #ifdef CONFIG_NO_HZ_FULL
2154 extern bool sched_can_stop_tick(void);
2155 extern u64 scheduler_tick_max_deferment(void);
2156 #else
2157 static inline bool sched_can_stop_tick(void) { return false; }
2158 #endif
2159
2160 #ifdef CONFIG_SCHED_AUTOGROUP
2161 extern void sched_autogroup_create_attach(struct task_struct *p);
2162 extern void sched_autogroup_detach(struct task_struct *p);
2163 extern void sched_autogroup_fork(struct signal_struct *sig);
2164 extern void sched_autogroup_exit(struct signal_struct *sig);
2165 #ifdef CONFIG_PROC_FS
2166 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2167 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2168 #endif
2169 #else
2170 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2171 static inline void sched_autogroup_detach(struct task_struct *p) { }
2172 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2173 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2174 #endif
2175
2176 extern int yield_to(struct task_struct *p, bool preempt);
2177 extern void set_user_nice(struct task_struct *p, long nice);
2178 extern int task_prio(const struct task_struct *p);
2179 /**
2180 * task_nice - return the nice value of a given task.
2181 * @p: the task in question.
2182 *
2183 * Return: The nice value [ -20 ... 0 ... 19 ].
2184 */
2185 static inline int task_nice(const struct task_struct *p)
2186 {
2187 return PRIO_TO_NICE((p)->static_prio);
2188 }
2189 extern int can_nice(const struct task_struct *p, const int nice);
2190 extern int task_curr(const struct task_struct *p);
2191 extern int idle_cpu(int cpu);
2192 extern int sched_setscheduler(struct task_struct *, int,
2193 const struct sched_param *);
2194 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2195 const struct sched_param *);
2196 extern int sched_setattr(struct task_struct *,
2197 const struct sched_attr *);
2198 extern struct task_struct *idle_task(int cpu);
2199 /**
2200 * is_idle_task - is the specified task an idle task?
2201 * @p: the task in question.
2202 *
2203 * Return: 1 if @p is an idle task. 0 otherwise.
2204 */
2205 static inline bool is_idle_task(const struct task_struct *p)
2206 {
2207 return p->pid == 0;
2208 }
2209 extern struct task_struct *curr_task(int cpu);
2210 extern void set_curr_task(int cpu, struct task_struct *p);
2211
2212 void yield(void);
2213
2214 /*
2215 * The default (Linux) execution domain.
2216 */
2217 extern struct exec_domain default_exec_domain;
2218
2219 union thread_union {
2220 struct thread_info thread_info;
2221 unsigned long stack[THREAD_SIZE/sizeof(long)];
2222 };
2223
2224 #ifndef __HAVE_ARCH_KSTACK_END
2225 static inline int kstack_end(void *addr)
2226 {
2227 /* Reliable end of stack detection:
2228 * Some APM bios versions misalign the stack
2229 */
2230 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2231 }
2232 #endif
2233
2234 extern union thread_union init_thread_union;
2235 extern struct task_struct init_task;
2236
2237 extern struct mm_struct init_mm;
2238
2239 extern struct pid_namespace init_pid_ns;
2240
2241 /*
2242 * find a task by one of its numerical ids
2243 *
2244 * find_task_by_pid_ns():
2245 * finds a task by its pid in the specified namespace
2246 * find_task_by_vpid():
2247 * finds a task by its virtual pid
2248 *
2249 * see also find_vpid() etc in include/linux/pid.h
2250 */
2251
2252 extern struct task_struct *find_task_by_vpid(pid_t nr);
2253 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2254 struct pid_namespace *ns);
2255
2256 /* per-UID process charging. */
2257 extern struct user_struct * alloc_uid(kuid_t);
2258 static inline struct user_struct *get_uid(struct user_struct *u)
2259 {
2260 atomic_inc(&u->__count);
2261 return u;
2262 }
2263 extern void free_uid(struct user_struct *);
2264
2265 #include <asm/current.h>
2266
2267 extern void xtime_update(unsigned long ticks);
2268
2269 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2270 extern int wake_up_process(struct task_struct *tsk);
2271 extern void wake_up_new_task(struct task_struct *tsk);
2272 #ifdef CONFIG_SMP
2273 extern void kick_process(struct task_struct *tsk);
2274 #else
2275 static inline void kick_process(struct task_struct *tsk) { }
2276 #endif
2277 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2278 extern void sched_dead(struct task_struct *p);
2279
2280 extern void proc_caches_init(void);
2281 extern void flush_signals(struct task_struct *);
2282 extern void __flush_signals(struct task_struct *);
2283 extern void ignore_signals(struct task_struct *);
2284 extern void flush_signal_handlers(struct task_struct *, int force_default);
2285 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2286
2287 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2288 {
2289 unsigned long flags;
2290 int ret;
2291
2292 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2293 ret = dequeue_signal(tsk, mask, info);
2294 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2295
2296 return ret;
2297 }
2298
2299 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2300 sigset_t *mask);
2301 extern void unblock_all_signals(void);
2302 extern void release_task(struct task_struct * p);
2303 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2304 extern int force_sigsegv(int, struct task_struct *);
2305 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2306 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2307 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2308 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2309 const struct cred *, u32);
2310 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2311 extern int kill_pid(struct pid *pid, int sig, int priv);
2312 extern int kill_proc_info(int, struct siginfo *, pid_t);
2313 extern __must_check bool do_notify_parent(struct task_struct *, int);
2314 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2315 extern void force_sig(int, struct task_struct *);
2316 extern int send_sig(int, struct task_struct *, int);
2317 extern int zap_other_threads(struct task_struct *p);
2318 extern struct sigqueue *sigqueue_alloc(void);
2319 extern void sigqueue_free(struct sigqueue *);
2320 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2321 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2322
2323 static inline void restore_saved_sigmask(void)
2324 {
2325 if (test_and_clear_restore_sigmask())
2326 __set_current_blocked(&current->saved_sigmask);
2327 }
2328
2329 static inline sigset_t *sigmask_to_save(void)
2330 {
2331 sigset_t *res = &current->blocked;
2332 if (unlikely(test_restore_sigmask()))
2333 res = &current->saved_sigmask;
2334 return res;
2335 }
2336
2337 static inline int kill_cad_pid(int sig, int priv)
2338 {
2339 return kill_pid(cad_pid, sig, priv);
2340 }
2341
2342 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2343 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2344 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2345 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2346
2347 /*
2348 * True if we are on the alternate signal stack.
2349 */
2350 static inline int on_sig_stack(unsigned long sp)
2351 {
2352 #ifdef CONFIG_STACK_GROWSUP
2353 return sp >= current->sas_ss_sp &&
2354 sp - current->sas_ss_sp < current->sas_ss_size;
2355 #else
2356 return sp > current->sas_ss_sp &&
2357 sp - current->sas_ss_sp <= current->sas_ss_size;
2358 #endif
2359 }
2360
2361 static inline int sas_ss_flags(unsigned long sp)
2362 {
2363 if (!current->sas_ss_size)
2364 return SS_DISABLE;
2365
2366 return on_sig_stack(sp) ? SS_ONSTACK : 0;
2367 }
2368
2369 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2370 {
2371 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2372 #ifdef CONFIG_STACK_GROWSUP
2373 return current->sas_ss_sp;
2374 #else
2375 return current->sas_ss_sp + current->sas_ss_size;
2376 #endif
2377 return sp;
2378 }
2379
2380 /*
2381 * Routines for handling mm_structs
2382 */
2383 extern struct mm_struct * mm_alloc(void);
2384
2385 /* mmdrop drops the mm and the page tables */
2386 extern void __mmdrop(struct mm_struct *);
2387 static inline void mmdrop(struct mm_struct * mm)
2388 {
2389 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2390 __mmdrop(mm);
2391 }
2392
2393 /* mmput gets rid of the mappings and all user-space */
2394 extern void mmput(struct mm_struct *);
2395 /* Grab a reference to a task's mm, if it is not already going away */
2396 extern struct mm_struct *get_task_mm(struct task_struct *task);
2397 /*
2398 * Grab a reference to a task's mm, if it is not already going away
2399 * and ptrace_may_access with the mode parameter passed to it
2400 * succeeds.
2401 */
2402 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2403 /* Remove the current tasks stale references to the old mm_struct */
2404 extern void mm_release(struct task_struct *, struct mm_struct *);
2405
2406 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2407 struct task_struct *);
2408 extern void flush_thread(void);
2409 extern void exit_thread(void);
2410
2411 extern void exit_files(struct task_struct *);
2412 extern void __cleanup_sighand(struct sighand_struct *);
2413
2414 extern void exit_itimers(struct signal_struct *);
2415 extern void flush_itimer_signals(void);
2416
2417 extern void do_group_exit(int);
2418
2419 extern int do_execve(struct filename *,
2420 const char __user * const __user *,
2421 const char __user * const __user *);
2422 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2423 struct task_struct *fork_idle(int);
2424 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2425
2426 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2427 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2428 {
2429 __set_task_comm(tsk, from, false);
2430 }
2431 extern char *get_task_comm(char *to, struct task_struct *tsk);
2432
2433 #ifdef CONFIG_SMP
2434 void scheduler_ipi(void);
2435 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2436 #else
2437 static inline void scheduler_ipi(void) { }
2438 static inline unsigned long wait_task_inactive(struct task_struct *p,
2439 long match_state)
2440 {
2441 return 1;
2442 }
2443 #endif
2444
2445 #define next_task(p) \
2446 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2447
2448 #define for_each_process(p) \
2449 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2450
2451 extern bool current_is_single_threaded(void);
2452
2453 /*
2454 * Careful: do_each_thread/while_each_thread is a double loop so
2455 * 'break' will not work as expected - use goto instead.
2456 */
2457 #define do_each_thread(g, t) \
2458 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2459
2460 #define while_each_thread(g, t) \
2461 while ((t = next_thread(t)) != g)
2462
2463 #define __for_each_thread(signal, t) \
2464 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2465
2466 #define for_each_thread(p, t) \
2467 __for_each_thread((p)->signal, t)
2468
2469 /* Careful: this is a double loop, 'break' won't work as expected. */
2470 #define for_each_process_thread(p, t) \
2471 for_each_process(p) for_each_thread(p, t)
2472
2473 static inline int get_nr_threads(struct task_struct *tsk)
2474 {
2475 return tsk->signal->nr_threads;
2476 }
2477
2478 static inline bool thread_group_leader(struct task_struct *p)
2479 {
2480 return p->exit_signal >= 0;
2481 }
2482
2483 /* Do to the insanities of de_thread it is possible for a process
2484 * to have the pid of the thread group leader without actually being
2485 * the thread group leader. For iteration through the pids in proc
2486 * all we care about is that we have a task with the appropriate
2487 * pid, we don't actually care if we have the right task.
2488 */
2489 static inline bool has_group_leader_pid(struct task_struct *p)
2490 {
2491 return task_pid(p) == p->signal->leader_pid;
2492 }
2493
2494 static inline
2495 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2496 {
2497 return p1->signal == p2->signal;
2498 }
2499
2500 static inline struct task_struct *next_thread(const struct task_struct *p)
2501 {
2502 return list_entry_rcu(p->thread_group.next,
2503 struct task_struct, thread_group);
2504 }
2505
2506 static inline int thread_group_empty(struct task_struct *p)
2507 {
2508 return list_empty(&p->thread_group);
2509 }
2510
2511 #define delay_group_leader(p) \
2512 (thread_group_leader(p) && !thread_group_empty(p))
2513
2514 /*
2515 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2516 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2517 * pins the final release of task.io_context. Also protects ->cpuset and
2518 * ->cgroup.subsys[]. And ->vfork_done.
2519 *
2520 * Nests both inside and outside of read_lock(&tasklist_lock).
2521 * It must not be nested with write_lock_irq(&tasklist_lock),
2522 * neither inside nor outside.
2523 */
2524 static inline void task_lock(struct task_struct *p)
2525 {
2526 spin_lock(&p->alloc_lock);
2527 }
2528
2529 static inline void task_unlock(struct task_struct *p)
2530 {
2531 spin_unlock(&p->alloc_lock);
2532 }
2533
2534 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2535 unsigned long *flags);
2536
2537 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2538 unsigned long *flags)
2539 {
2540 struct sighand_struct *ret;
2541
2542 ret = __lock_task_sighand(tsk, flags);
2543 (void)__cond_lock(&tsk->sighand->siglock, ret);
2544 return ret;
2545 }
2546
2547 static inline void unlock_task_sighand(struct task_struct *tsk,
2548 unsigned long *flags)
2549 {
2550 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2551 }
2552
2553 #ifdef CONFIG_CGROUPS
2554 static inline void threadgroup_change_begin(struct task_struct *tsk)
2555 {
2556 down_read(&tsk->signal->group_rwsem);
2557 }
2558 static inline void threadgroup_change_end(struct task_struct *tsk)
2559 {
2560 up_read(&tsk->signal->group_rwsem);
2561 }
2562
2563 /**
2564 * threadgroup_lock - lock threadgroup
2565 * @tsk: member task of the threadgroup to lock
2566 *
2567 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2568 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2569 * change ->group_leader/pid. This is useful for cases where the threadgroup
2570 * needs to stay stable across blockable operations.
2571 *
2572 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2573 * synchronization. While held, no new task will be added to threadgroup
2574 * and no existing live task will have its PF_EXITING set.
2575 *
2576 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2577 * sub-thread becomes a new leader.
2578 */
2579 static inline void threadgroup_lock(struct task_struct *tsk)
2580 {
2581 down_write(&tsk->signal->group_rwsem);
2582 }
2583
2584 /**
2585 * threadgroup_unlock - unlock threadgroup
2586 * @tsk: member task of the threadgroup to unlock
2587 *
2588 * Reverse threadgroup_lock().
2589 */
2590 static inline void threadgroup_unlock(struct task_struct *tsk)
2591 {
2592 up_write(&tsk->signal->group_rwsem);
2593 }
2594 #else
2595 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2596 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2597 static inline void threadgroup_lock(struct task_struct *tsk) {}
2598 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2599 #endif
2600
2601 #ifndef __HAVE_THREAD_FUNCTIONS
2602
2603 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2604 #define task_stack_page(task) ((task)->stack)
2605
2606 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2607 {
2608 *task_thread_info(p) = *task_thread_info(org);
2609 task_thread_info(p)->task = p;
2610 }
2611
2612 static inline unsigned long *end_of_stack(struct task_struct *p)
2613 {
2614 return (unsigned long *)(task_thread_info(p) + 1);
2615 }
2616
2617 #endif
2618
2619 static inline int object_is_on_stack(void *obj)
2620 {
2621 void *stack = task_stack_page(current);
2622
2623 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2624 }
2625
2626 extern void thread_info_cache_init(void);
2627
2628 #ifdef CONFIG_DEBUG_STACK_USAGE
2629 static inline unsigned long stack_not_used(struct task_struct *p)
2630 {
2631 unsigned long *n = end_of_stack(p);
2632
2633 do { /* Skip over canary */
2634 n++;
2635 } while (!*n);
2636
2637 return (unsigned long)n - (unsigned long)end_of_stack(p);
2638 }
2639 #endif
2640
2641 /* set thread flags in other task's structures
2642 * - see asm/thread_info.h for TIF_xxxx flags available
2643 */
2644 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2645 {
2646 set_ti_thread_flag(task_thread_info(tsk), flag);
2647 }
2648
2649 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2650 {
2651 clear_ti_thread_flag(task_thread_info(tsk), flag);
2652 }
2653
2654 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2655 {
2656 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2657 }
2658
2659 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2660 {
2661 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2662 }
2663
2664 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2665 {
2666 return test_ti_thread_flag(task_thread_info(tsk), flag);
2667 }
2668
2669 static inline void set_tsk_need_resched(struct task_struct *tsk)
2670 {
2671 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2672 }
2673
2674 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2675 {
2676 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2677 }
2678
2679 static inline int test_tsk_need_resched(struct task_struct *tsk)
2680 {
2681 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2682 }
2683
2684 static inline int restart_syscall(void)
2685 {
2686 set_tsk_thread_flag(current, TIF_SIGPENDING);
2687 return -ERESTARTNOINTR;
2688 }
2689
2690 static inline int signal_pending(struct task_struct *p)
2691 {
2692 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2693 }
2694
2695 static inline int __fatal_signal_pending(struct task_struct *p)
2696 {
2697 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2698 }
2699
2700 static inline int fatal_signal_pending(struct task_struct *p)
2701 {
2702 return signal_pending(p) && __fatal_signal_pending(p);
2703 }
2704
2705 static inline int signal_pending_state(long state, struct task_struct *p)
2706 {
2707 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2708 return 0;
2709 if (!signal_pending(p))
2710 return 0;
2711
2712 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2713 }
2714
2715 /*
2716 * cond_resched() and cond_resched_lock(): latency reduction via
2717 * explicit rescheduling in places that are safe. The return
2718 * value indicates whether a reschedule was done in fact.
2719 * cond_resched_lock() will drop the spinlock before scheduling,
2720 * cond_resched_softirq() will enable bhs before scheduling.
2721 */
2722 extern int _cond_resched(void);
2723
2724 #define cond_resched() ({ \
2725 __might_sleep(__FILE__, __LINE__, 0); \
2726 _cond_resched(); \
2727 })
2728
2729 extern int __cond_resched_lock(spinlock_t *lock);
2730
2731 #ifdef CONFIG_PREEMPT_COUNT
2732 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2733 #else
2734 #define PREEMPT_LOCK_OFFSET 0
2735 #endif
2736
2737 #define cond_resched_lock(lock) ({ \
2738 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2739 __cond_resched_lock(lock); \
2740 })
2741
2742 extern int __cond_resched_softirq(void);
2743
2744 #define cond_resched_softirq() ({ \
2745 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2746 __cond_resched_softirq(); \
2747 })
2748
2749 static inline void cond_resched_rcu(void)
2750 {
2751 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2752 rcu_read_unlock();
2753 cond_resched();
2754 rcu_read_lock();
2755 #endif
2756 }
2757
2758 /*
2759 * Does a critical section need to be broken due to another
2760 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2761 * but a general need for low latency)
2762 */
2763 static inline int spin_needbreak(spinlock_t *lock)
2764 {
2765 #ifdef CONFIG_PREEMPT
2766 return spin_is_contended(lock);
2767 #else
2768 return 0;
2769 #endif
2770 }
2771
2772 /*
2773 * Idle thread specific functions to determine the need_resched
2774 * polling state.
2775 */
2776 #ifdef TIF_POLLING_NRFLAG
2777 static inline int tsk_is_polling(struct task_struct *p)
2778 {
2779 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2780 }
2781
2782 static inline void __current_set_polling(void)
2783 {
2784 set_thread_flag(TIF_POLLING_NRFLAG);
2785 }
2786
2787 static inline bool __must_check current_set_polling_and_test(void)
2788 {
2789 __current_set_polling();
2790
2791 /*
2792 * Polling state must be visible before we test NEED_RESCHED,
2793 * paired by resched_curr()
2794 */
2795 smp_mb__after_atomic();
2796
2797 return unlikely(tif_need_resched());
2798 }
2799
2800 static inline void __current_clr_polling(void)
2801 {
2802 clear_thread_flag(TIF_POLLING_NRFLAG);
2803 }
2804
2805 static inline bool __must_check current_clr_polling_and_test(void)
2806 {
2807 __current_clr_polling();
2808
2809 /*
2810 * Polling state must be visible before we test NEED_RESCHED,
2811 * paired by resched_curr()
2812 */
2813 smp_mb__after_atomic();
2814
2815 return unlikely(tif_need_resched());
2816 }
2817
2818 #else
2819 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2820 static inline void __current_set_polling(void) { }
2821 static inline void __current_clr_polling(void) { }
2822
2823 static inline bool __must_check current_set_polling_and_test(void)
2824 {
2825 return unlikely(tif_need_resched());
2826 }
2827 static inline bool __must_check current_clr_polling_and_test(void)
2828 {
2829 return unlikely(tif_need_resched());
2830 }
2831 #endif
2832
2833 static inline void current_clr_polling(void)
2834 {
2835 __current_clr_polling();
2836
2837 /*
2838 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2839 * Once the bit is cleared, we'll get IPIs with every new
2840 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2841 * fold.
2842 */
2843 smp_mb(); /* paired with resched_curr() */
2844
2845 preempt_fold_need_resched();
2846 }
2847
2848 static __always_inline bool need_resched(void)
2849 {
2850 return unlikely(tif_need_resched());
2851 }
2852
2853 /*
2854 * Thread group CPU time accounting.
2855 */
2856 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2857 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2858
2859 static inline void thread_group_cputime_init(struct signal_struct *sig)
2860 {
2861 raw_spin_lock_init(&sig->cputimer.lock);
2862 }
2863
2864 /*
2865 * Reevaluate whether the task has signals pending delivery.
2866 * Wake the task if so.
2867 * This is required every time the blocked sigset_t changes.
2868 * callers must hold sighand->siglock.
2869 */
2870 extern void recalc_sigpending_and_wake(struct task_struct *t);
2871 extern void recalc_sigpending(void);
2872
2873 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2874
2875 static inline void signal_wake_up(struct task_struct *t, bool resume)
2876 {
2877 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2878 }
2879 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2880 {
2881 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2882 }
2883
2884 /*
2885 * Wrappers for p->thread_info->cpu access. No-op on UP.
2886 */
2887 #ifdef CONFIG_SMP
2888
2889 static inline unsigned int task_cpu(const struct task_struct *p)
2890 {
2891 return task_thread_info(p)->cpu;
2892 }
2893
2894 static inline int task_node(const struct task_struct *p)
2895 {
2896 return cpu_to_node(task_cpu(p));
2897 }
2898
2899 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2900
2901 #else
2902
2903 static inline unsigned int task_cpu(const struct task_struct *p)
2904 {
2905 return 0;
2906 }
2907
2908 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2909 {
2910 }
2911
2912 #endif /* CONFIG_SMP */
2913
2914 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2915 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2916
2917 #ifdef CONFIG_CGROUP_SCHED
2918 extern struct task_group root_task_group;
2919 #endif /* CONFIG_CGROUP_SCHED */
2920
2921 extern int task_can_switch_user(struct user_struct *up,
2922 struct task_struct *tsk);
2923
2924 #ifdef CONFIG_TASK_XACCT
2925 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2926 {
2927 tsk->ioac.rchar += amt;
2928 }
2929
2930 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2931 {
2932 tsk->ioac.wchar += amt;
2933 }
2934
2935 static inline void inc_syscr(struct task_struct *tsk)
2936 {
2937 tsk->ioac.syscr++;
2938 }
2939
2940 static inline void inc_syscw(struct task_struct *tsk)
2941 {
2942 tsk->ioac.syscw++;
2943 }
2944 #else
2945 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2946 {
2947 }
2948
2949 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2950 {
2951 }
2952
2953 static inline void inc_syscr(struct task_struct *tsk)
2954 {
2955 }
2956
2957 static inline void inc_syscw(struct task_struct *tsk)
2958 {
2959 }
2960 #endif
2961
2962 #ifndef TASK_SIZE_OF
2963 #define TASK_SIZE_OF(tsk) TASK_SIZE
2964 #endif
2965
2966 #ifdef CONFIG_MEMCG
2967 extern void mm_update_next_owner(struct mm_struct *mm);
2968 #else
2969 static inline void mm_update_next_owner(struct mm_struct *mm)
2970 {
2971 }
2972 #endif /* CONFIG_MEMCG */
2973
2974 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2975 unsigned int limit)
2976 {
2977 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2978 }
2979
2980 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2981 unsigned int limit)
2982 {
2983 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2984 }
2985
2986 static inline unsigned long rlimit(unsigned int limit)
2987 {
2988 return task_rlimit(current, limit);
2989 }
2990
2991 static inline unsigned long rlimit_max(unsigned int limit)
2992 {
2993 return task_rlimit_max(current, limit);
2994 }
2995
2996 #endif
This page took 0.740949 seconds and 6 git commands to generate.