]> git.proxmox.com Git - mirror_ubuntu-eoan-kernel.git/blob - include/linux/sched.h
sched: move no_new_privs into new atomic flags
[mirror_ubuntu-eoan-kernel.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/signal.h>
37 #include <linux/compiler.h>
38 #include <linux/completion.h>
39 #include <linux/pid.h>
40 #include <linux/percpu.h>
41 #include <linux/topology.h>
42 #include <linux/proportions.h>
43 #include <linux/seccomp.h>
44 #include <linux/rcupdate.h>
45 #include <linux/rculist.h>
46 #include <linux/rtmutex.h>
47
48 #include <linux/time.h>
49 #include <linux/param.h>
50 #include <linux/resource.h>
51 #include <linux/timer.h>
52 #include <linux/hrtimer.h>
53 #include <linux/task_io_accounting.h>
54 #include <linux/latencytop.h>
55 #include <linux/cred.h>
56 #include <linux/llist.h>
57 #include <linux/uidgid.h>
58 #include <linux/gfp.h>
59
60 #include <asm/processor.h>
61
62 #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */
63
64 /*
65 * Extended scheduling parameters data structure.
66 *
67 * This is needed because the original struct sched_param can not be
68 * altered without introducing ABI issues with legacy applications
69 * (e.g., in sched_getparam()).
70 *
71 * However, the possibility of specifying more than just a priority for
72 * the tasks may be useful for a wide variety of application fields, e.g.,
73 * multimedia, streaming, automation and control, and many others.
74 *
75 * This variant (sched_attr) is meant at describing a so-called
76 * sporadic time-constrained task. In such model a task is specified by:
77 * - the activation period or minimum instance inter-arrival time;
78 * - the maximum (or average, depending on the actual scheduling
79 * discipline) computation time of all instances, a.k.a. runtime;
80 * - the deadline (relative to the actual activation time) of each
81 * instance.
82 * Very briefly, a periodic (sporadic) task asks for the execution of
83 * some specific computation --which is typically called an instance--
84 * (at most) every period. Moreover, each instance typically lasts no more
85 * than the runtime and must be completed by time instant t equal to
86 * the instance activation time + the deadline.
87 *
88 * This is reflected by the actual fields of the sched_attr structure:
89 *
90 * @size size of the structure, for fwd/bwd compat.
91 *
92 * @sched_policy task's scheduling policy
93 * @sched_flags for customizing the scheduler behaviour
94 * @sched_nice task's nice value (SCHED_NORMAL/BATCH)
95 * @sched_priority task's static priority (SCHED_FIFO/RR)
96 * @sched_deadline representative of the task's deadline
97 * @sched_runtime representative of the task's runtime
98 * @sched_period representative of the task's period
99 *
100 * Given this task model, there are a multiplicity of scheduling algorithms
101 * and policies, that can be used to ensure all the tasks will make their
102 * timing constraints.
103 *
104 * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
105 * only user of this new interface. More information about the algorithm
106 * available in the scheduling class file or in Documentation/.
107 */
108 struct sched_attr {
109 u32 size;
110
111 u32 sched_policy;
112 u64 sched_flags;
113
114 /* SCHED_NORMAL, SCHED_BATCH */
115 s32 sched_nice;
116
117 /* SCHED_FIFO, SCHED_RR */
118 u32 sched_priority;
119
120 /* SCHED_DEADLINE */
121 u64 sched_runtime;
122 u64 sched_deadline;
123 u64 sched_period;
124 };
125
126 struct exec_domain;
127 struct futex_pi_state;
128 struct robust_list_head;
129 struct bio_list;
130 struct fs_struct;
131 struct perf_event_context;
132 struct blk_plug;
133 struct filename;
134
135 #define VMACACHE_BITS 2
136 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
137 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
138
139 /*
140 * These are the constant used to fake the fixed-point load-average
141 * counting. Some notes:
142 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
143 * a load-average precision of 10 bits integer + 11 bits fractional
144 * - if you want to count load-averages more often, you need more
145 * precision, or rounding will get you. With 2-second counting freq,
146 * the EXP_n values would be 1981, 2034 and 2043 if still using only
147 * 11 bit fractions.
148 */
149 extern unsigned long avenrun[]; /* Load averages */
150 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
151
152 #define FSHIFT 11 /* nr of bits of precision */
153 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
154 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
155 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
156 #define EXP_5 2014 /* 1/exp(5sec/5min) */
157 #define EXP_15 2037 /* 1/exp(5sec/15min) */
158
159 #define CALC_LOAD(load,exp,n) \
160 load *= exp; \
161 load += n*(FIXED_1-exp); \
162 load >>= FSHIFT;
163
164 extern unsigned long total_forks;
165 extern int nr_threads;
166 DECLARE_PER_CPU(unsigned long, process_counts);
167 extern int nr_processes(void);
168 extern unsigned long nr_running(void);
169 extern unsigned long nr_iowait(void);
170 extern unsigned long nr_iowait_cpu(int cpu);
171 extern unsigned long this_cpu_load(void);
172
173
174 extern void calc_global_load(unsigned long ticks);
175 extern void update_cpu_load_nohz(void);
176
177 extern unsigned long get_parent_ip(unsigned long addr);
178
179 extern void dump_cpu_task(int cpu);
180
181 struct seq_file;
182 struct cfs_rq;
183 struct task_group;
184 #ifdef CONFIG_SCHED_DEBUG
185 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
186 extern void proc_sched_set_task(struct task_struct *p);
187 extern void
188 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
189 #endif
190
191 /*
192 * Task state bitmask. NOTE! These bits are also
193 * encoded in fs/proc/array.c: get_task_state().
194 *
195 * We have two separate sets of flags: task->state
196 * is about runnability, while task->exit_state are
197 * about the task exiting. Confusing, but this way
198 * modifying one set can't modify the other one by
199 * mistake.
200 */
201 #define TASK_RUNNING 0
202 #define TASK_INTERRUPTIBLE 1
203 #define TASK_UNINTERRUPTIBLE 2
204 #define __TASK_STOPPED 4
205 #define __TASK_TRACED 8
206 /* in tsk->exit_state */
207 #define EXIT_DEAD 16
208 #define EXIT_ZOMBIE 32
209 #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD)
210 /* in tsk->state again */
211 #define TASK_DEAD 64
212 #define TASK_WAKEKILL 128
213 #define TASK_WAKING 256
214 #define TASK_PARKED 512
215 #define TASK_STATE_MAX 1024
216
217 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWP"
218
219 extern char ___assert_task_state[1 - 2*!!(
220 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
221
222 /* Convenience macros for the sake of set_task_state */
223 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
224 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
225 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
226
227 /* Convenience macros for the sake of wake_up */
228 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
229 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
230
231 /* get_task_state() */
232 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
233 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
234 __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
235
236 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
237 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
238 #define task_is_stopped_or_traced(task) \
239 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
240 #define task_contributes_to_load(task) \
241 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
242 (task->flags & PF_FROZEN) == 0)
243
244 #define __set_task_state(tsk, state_value) \
245 do { (tsk)->state = (state_value); } while (0)
246 #define set_task_state(tsk, state_value) \
247 set_mb((tsk)->state, (state_value))
248
249 /*
250 * set_current_state() includes a barrier so that the write of current->state
251 * is correctly serialised wrt the caller's subsequent test of whether to
252 * actually sleep:
253 *
254 * set_current_state(TASK_UNINTERRUPTIBLE);
255 * if (do_i_need_to_sleep())
256 * schedule();
257 *
258 * If the caller does not need such serialisation then use __set_current_state()
259 */
260 #define __set_current_state(state_value) \
261 do { current->state = (state_value); } while (0)
262 #define set_current_state(state_value) \
263 set_mb(current->state, (state_value))
264
265 /* Task command name length */
266 #define TASK_COMM_LEN 16
267
268 #include <linux/spinlock.h>
269
270 /*
271 * This serializes "schedule()" and also protects
272 * the run-queue from deletions/modifications (but
273 * _adding_ to the beginning of the run-queue has
274 * a separate lock).
275 */
276 extern rwlock_t tasklist_lock;
277 extern spinlock_t mmlist_lock;
278
279 struct task_struct;
280
281 #ifdef CONFIG_PROVE_RCU
282 extern int lockdep_tasklist_lock_is_held(void);
283 #endif /* #ifdef CONFIG_PROVE_RCU */
284
285 extern void sched_init(void);
286 extern void sched_init_smp(void);
287 extern asmlinkage void schedule_tail(struct task_struct *prev);
288 extern void init_idle(struct task_struct *idle, int cpu);
289 extern void init_idle_bootup_task(struct task_struct *idle);
290
291 extern int runqueue_is_locked(int cpu);
292
293 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
294 extern void nohz_balance_enter_idle(int cpu);
295 extern void set_cpu_sd_state_idle(void);
296 extern int get_nohz_timer_target(int pinned);
297 #else
298 static inline void nohz_balance_enter_idle(int cpu) { }
299 static inline void set_cpu_sd_state_idle(void) { }
300 static inline int get_nohz_timer_target(int pinned)
301 {
302 return smp_processor_id();
303 }
304 #endif
305
306 /*
307 * Only dump TASK_* tasks. (0 for all tasks)
308 */
309 extern void show_state_filter(unsigned long state_filter);
310
311 static inline void show_state(void)
312 {
313 show_state_filter(0);
314 }
315
316 extern void show_regs(struct pt_regs *);
317
318 /*
319 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
320 * task), SP is the stack pointer of the first frame that should be shown in the back
321 * trace (or NULL if the entire call-chain of the task should be shown).
322 */
323 extern void show_stack(struct task_struct *task, unsigned long *sp);
324
325 void io_schedule(void);
326 long io_schedule_timeout(long timeout);
327
328 extern void cpu_init (void);
329 extern void trap_init(void);
330 extern void update_process_times(int user);
331 extern void scheduler_tick(void);
332
333 extern void sched_show_task(struct task_struct *p);
334
335 #ifdef CONFIG_LOCKUP_DETECTOR
336 extern void touch_softlockup_watchdog(void);
337 extern void touch_softlockup_watchdog_sync(void);
338 extern void touch_all_softlockup_watchdogs(void);
339 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
342 extern unsigned int softlockup_panic;
343 void lockup_detector_init(void);
344 #else
345 static inline void touch_softlockup_watchdog(void)
346 {
347 }
348 static inline void touch_softlockup_watchdog_sync(void)
349 {
350 }
351 static inline void touch_all_softlockup_watchdogs(void)
352 {
353 }
354 static inline void lockup_detector_init(void)
355 {
356 }
357 #endif
358
359 #ifdef CONFIG_DETECT_HUNG_TASK
360 void reset_hung_task_detector(void);
361 #else
362 static inline void reset_hung_task_detector(void)
363 {
364 }
365 #endif
366
367 /* Attach to any functions which should be ignored in wchan output. */
368 #define __sched __attribute__((__section__(".sched.text")))
369
370 /* Linker adds these: start and end of __sched functions */
371 extern char __sched_text_start[], __sched_text_end[];
372
373 /* Is this address in the __sched functions? */
374 extern int in_sched_functions(unsigned long addr);
375
376 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
377 extern signed long schedule_timeout(signed long timeout);
378 extern signed long schedule_timeout_interruptible(signed long timeout);
379 extern signed long schedule_timeout_killable(signed long timeout);
380 extern signed long schedule_timeout_uninterruptible(signed long timeout);
381 asmlinkage void schedule(void);
382 extern void schedule_preempt_disabled(void);
383
384 struct nsproxy;
385 struct user_namespace;
386
387 #ifdef CONFIG_MMU
388 extern void arch_pick_mmap_layout(struct mm_struct *mm);
389 extern unsigned long
390 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
391 unsigned long, unsigned long);
392 extern unsigned long
393 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
394 unsigned long len, unsigned long pgoff,
395 unsigned long flags);
396 #else
397 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
398 #endif
399
400 #define SUID_DUMP_DISABLE 0 /* No setuid dumping */
401 #define SUID_DUMP_USER 1 /* Dump as user of process */
402 #define SUID_DUMP_ROOT 2 /* Dump as root */
403
404 /* mm flags */
405
406 /* for SUID_DUMP_* above */
407 #define MMF_DUMPABLE_BITS 2
408 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
409
410 extern void set_dumpable(struct mm_struct *mm, int value);
411 /*
412 * This returns the actual value of the suid_dumpable flag. For things
413 * that are using this for checking for privilege transitions, it must
414 * test against SUID_DUMP_USER rather than treating it as a boolean
415 * value.
416 */
417 static inline int __get_dumpable(unsigned long mm_flags)
418 {
419 return mm_flags & MMF_DUMPABLE_MASK;
420 }
421
422 static inline int get_dumpable(struct mm_struct *mm)
423 {
424 return __get_dumpable(mm->flags);
425 }
426
427 /* coredump filter bits */
428 #define MMF_DUMP_ANON_PRIVATE 2
429 #define MMF_DUMP_ANON_SHARED 3
430 #define MMF_DUMP_MAPPED_PRIVATE 4
431 #define MMF_DUMP_MAPPED_SHARED 5
432 #define MMF_DUMP_ELF_HEADERS 6
433 #define MMF_DUMP_HUGETLB_PRIVATE 7
434 #define MMF_DUMP_HUGETLB_SHARED 8
435
436 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
437 #define MMF_DUMP_FILTER_BITS 7
438 #define MMF_DUMP_FILTER_MASK \
439 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
440 #define MMF_DUMP_FILTER_DEFAULT \
441 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
442 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
443
444 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
445 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
446 #else
447 # define MMF_DUMP_MASK_DEFAULT_ELF 0
448 #endif
449 /* leave room for more dump flags */
450 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
451 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
452 #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */
453
454 #define MMF_HAS_UPROBES 19 /* has uprobes */
455 #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */
456
457 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
458
459 struct sighand_struct {
460 atomic_t count;
461 struct k_sigaction action[_NSIG];
462 spinlock_t siglock;
463 wait_queue_head_t signalfd_wqh;
464 };
465
466 struct pacct_struct {
467 int ac_flag;
468 long ac_exitcode;
469 unsigned long ac_mem;
470 cputime_t ac_utime, ac_stime;
471 unsigned long ac_minflt, ac_majflt;
472 };
473
474 struct cpu_itimer {
475 cputime_t expires;
476 cputime_t incr;
477 u32 error;
478 u32 incr_error;
479 };
480
481 /**
482 * struct cputime - snaphsot of system and user cputime
483 * @utime: time spent in user mode
484 * @stime: time spent in system mode
485 *
486 * Gathers a generic snapshot of user and system time.
487 */
488 struct cputime {
489 cputime_t utime;
490 cputime_t stime;
491 };
492
493 /**
494 * struct task_cputime - collected CPU time counts
495 * @utime: time spent in user mode, in &cputime_t units
496 * @stime: time spent in kernel mode, in &cputime_t units
497 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
498 *
499 * This is an extension of struct cputime that includes the total runtime
500 * spent by the task from the scheduler point of view.
501 *
502 * As a result, this structure groups together three kinds of CPU time
503 * that are tracked for threads and thread groups. Most things considering
504 * CPU time want to group these counts together and treat all three
505 * of them in parallel.
506 */
507 struct task_cputime {
508 cputime_t utime;
509 cputime_t stime;
510 unsigned long long sum_exec_runtime;
511 };
512 /* Alternate field names when used to cache expirations. */
513 #define prof_exp stime
514 #define virt_exp utime
515 #define sched_exp sum_exec_runtime
516
517 #define INIT_CPUTIME \
518 (struct task_cputime) { \
519 .utime = 0, \
520 .stime = 0, \
521 .sum_exec_runtime = 0, \
522 }
523
524 #ifdef CONFIG_PREEMPT_COUNT
525 #define PREEMPT_DISABLED (1 + PREEMPT_ENABLED)
526 #else
527 #define PREEMPT_DISABLED PREEMPT_ENABLED
528 #endif
529
530 /*
531 * Disable preemption until the scheduler is running.
532 * Reset by start_kernel()->sched_init()->init_idle().
533 *
534 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
535 * before the scheduler is active -- see should_resched().
536 */
537 #define INIT_PREEMPT_COUNT (PREEMPT_DISABLED + PREEMPT_ACTIVE)
538
539 /**
540 * struct thread_group_cputimer - thread group interval timer counts
541 * @cputime: thread group interval timers.
542 * @running: non-zero when there are timers running and
543 * @cputime receives updates.
544 * @lock: lock for fields in this struct.
545 *
546 * This structure contains the version of task_cputime, above, that is
547 * used for thread group CPU timer calculations.
548 */
549 struct thread_group_cputimer {
550 struct task_cputime cputime;
551 int running;
552 raw_spinlock_t lock;
553 };
554
555 #include <linux/rwsem.h>
556 struct autogroup;
557
558 /*
559 * NOTE! "signal_struct" does not have its own
560 * locking, because a shared signal_struct always
561 * implies a shared sighand_struct, so locking
562 * sighand_struct is always a proper superset of
563 * the locking of signal_struct.
564 */
565 struct signal_struct {
566 atomic_t sigcnt;
567 atomic_t live;
568 int nr_threads;
569 struct list_head thread_head;
570
571 wait_queue_head_t wait_chldexit; /* for wait4() */
572
573 /* current thread group signal load-balancing target: */
574 struct task_struct *curr_target;
575
576 /* shared signal handling: */
577 struct sigpending shared_pending;
578
579 /* thread group exit support */
580 int group_exit_code;
581 /* overloaded:
582 * - notify group_exit_task when ->count is equal to notify_count
583 * - everyone except group_exit_task is stopped during signal delivery
584 * of fatal signals, group_exit_task processes the signal.
585 */
586 int notify_count;
587 struct task_struct *group_exit_task;
588
589 /* thread group stop support, overloads group_exit_code too */
590 int group_stop_count;
591 unsigned int flags; /* see SIGNAL_* flags below */
592
593 /*
594 * PR_SET_CHILD_SUBREAPER marks a process, like a service
595 * manager, to re-parent orphan (double-forking) child processes
596 * to this process instead of 'init'. The service manager is
597 * able to receive SIGCHLD signals and is able to investigate
598 * the process until it calls wait(). All children of this
599 * process will inherit a flag if they should look for a
600 * child_subreaper process at exit.
601 */
602 unsigned int is_child_subreaper:1;
603 unsigned int has_child_subreaper:1;
604
605 /* POSIX.1b Interval Timers */
606 int posix_timer_id;
607 struct list_head posix_timers;
608
609 /* ITIMER_REAL timer for the process */
610 struct hrtimer real_timer;
611 struct pid *leader_pid;
612 ktime_t it_real_incr;
613
614 /*
615 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
616 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
617 * values are defined to 0 and 1 respectively
618 */
619 struct cpu_itimer it[2];
620
621 /*
622 * Thread group totals for process CPU timers.
623 * See thread_group_cputimer(), et al, for details.
624 */
625 struct thread_group_cputimer cputimer;
626
627 /* Earliest-expiration cache. */
628 struct task_cputime cputime_expires;
629
630 struct list_head cpu_timers[3];
631
632 struct pid *tty_old_pgrp;
633
634 /* boolean value for session group leader */
635 int leader;
636
637 struct tty_struct *tty; /* NULL if no tty */
638
639 #ifdef CONFIG_SCHED_AUTOGROUP
640 struct autogroup *autogroup;
641 #endif
642 /*
643 * Cumulative resource counters for dead threads in the group,
644 * and for reaped dead child processes forked by this group.
645 * Live threads maintain their own counters and add to these
646 * in __exit_signal, except for the group leader.
647 */
648 cputime_t utime, stime, cutime, cstime;
649 cputime_t gtime;
650 cputime_t cgtime;
651 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
652 struct cputime prev_cputime;
653 #endif
654 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
655 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
656 unsigned long inblock, oublock, cinblock, coublock;
657 unsigned long maxrss, cmaxrss;
658 struct task_io_accounting ioac;
659
660 /*
661 * Cumulative ns of schedule CPU time fo dead threads in the
662 * group, not including a zombie group leader, (This only differs
663 * from jiffies_to_ns(utime + stime) if sched_clock uses something
664 * other than jiffies.)
665 */
666 unsigned long long sum_sched_runtime;
667
668 /*
669 * We don't bother to synchronize most readers of this at all,
670 * because there is no reader checking a limit that actually needs
671 * to get both rlim_cur and rlim_max atomically, and either one
672 * alone is a single word that can safely be read normally.
673 * getrlimit/setrlimit use task_lock(current->group_leader) to
674 * protect this instead of the siglock, because they really
675 * have no need to disable irqs.
676 */
677 struct rlimit rlim[RLIM_NLIMITS];
678
679 #ifdef CONFIG_BSD_PROCESS_ACCT
680 struct pacct_struct pacct; /* per-process accounting information */
681 #endif
682 #ifdef CONFIG_TASKSTATS
683 struct taskstats *stats;
684 #endif
685 #ifdef CONFIG_AUDIT
686 unsigned audit_tty;
687 unsigned audit_tty_log_passwd;
688 struct tty_audit_buf *tty_audit_buf;
689 #endif
690 #ifdef CONFIG_CGROUPS
691 /*
692 * group_rwsem prevents new tasks from entering the threadgroup and
693 * member tasks from exiting,a more specifically, setting of
694 * PF_EXITING. fork and exit paths are protected with this rwsem
695 * using threadgroup_change_begin/end(). Users which require
696 * threadgroup to remain stable should use threadgroup_[un]lock()
697 * which also takes care of exec path. Currently, cgroup is the
698 * only user.
699 */
700 struct rw_semaphore group_rwsem;
701 #endif
702
703 oom_flags_t oom_flags;
704 short oom_score_adj; /* OOM kill score adjustment */
705 short oom_score_adj_min; /* OOM kill score adjustment min value.
706 * Only settable by CAP_SYS_RESOURCE. */
707
708 struct mutex cred_guard_mutex; /* guard against foreign influences on
709 * credential calculations
710 * (notably. ptrace) */
711 };
712
713 /*
714 * Bits in flags field of signal_struct.
715 */
716 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
717 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
718 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
719 #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */
720 /*
721 * Pending notifications to parent.
722 */
723 #define SIGNAL_CLD_STOPPED 0x00000010
724 #define SIGNAL_CLD_CONTINUED 0x00000020
725 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
726
727 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
728
729 /* If true, all threads except ->group_exit_task have pending SIGKILL */
730 static inline int signal_group_exit(const struct signal_struct *sig)
731 {
732 return (sig->flags & SIGNAL_GROUP_EXIT) ||
733 (sig->group_exit_task != NULL);
734 }
735
736 /*
737 * Some day this will be a full-fledged user tracking system..
738 */
739 struct user_struct {
740 atomic_t __count; /* reference count */
741 atomic_t processes; /* How many processes does this user have? */
742 atomic_t sigpending; /* How many pending signals does this user have? */
743 #ifdef CONFIG_INOTIFY_USER
744 atomic_t inotify_watches; /* How many inotify watches does this user have? */
745 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
746 #endif
747 #ifdef CONFIG_FANOTIFY
748 atomic_t fanotify_listeners;
749 #endif
750 #ifdef CONFIG_EPOLL
751 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
752 #endif
753 #ifdef CONFIG_POSIX_MQUEUE
754 /* protected by mq_lock */
755 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
756 #endif
757 unsigned long locked_shm; /* How many pages of mlocked shm ? */
758
759 #ifdef CONFIG_KEYS
760 struct key *uid_keyring; /* UID specific keyring */
761 struct key *session_keyring; /* UID's default session keyring */
762 #endif
763
764 /* Hash table maintenance information */
765 struct hlist_node uidhash_node;
766 kuid_t uid;
767
768 #ifdef CONFIG_PERF_EVENTS
769 atomic_long_t locked_vm;
770 #endif
771 };
772
773 extern int uids_sysfs_init(void);
774
775 extern struct user_struct *find_user(kuid_t);
776
777 extern struct user_struct root_user;
778 #define INIT_USER (&root_user)
779
780
781 struct backing_dev_info;
782 struct reclaim_state;
783
784 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
785 struct sched_info {
786 /* cumulative counters */
787 unsigned long pcount; /* # of times run on this cpu */
788 unsigned long long run_delay; /* time spent waiting on a runqueue */
789
790 /* timestamps */
791 unsigned long long last_arrival,/* when we last ran on a cpu */
792 last_queued; /* when we were last queued to run */
793 };
794 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
795
796 #ifdef CONFIG_TASK_DELAY_ACCT
797 struct task_delay_info {
798 spinlock_t lock;
799 unsigned int flags; /* Private per-task flags */
800
801 /* For each stat XXX, add following, aligned appropriately
802 *
803 * struct timespec XXX_start, XXX_end;
804 * u64 XXX_delay;
805 * u32 XXX_count;
806 *
807 * Atomicity of updates to XXX_delay, XXX_count protected by
808 * single lock above (split into XXX_lock if contention is an issue).
809 */
810
811 /*
812 * XXX_count is incremented on every XXX operation, the delay
813 * associated with the operation is added to XXX_delay.
814 * XXX_delay contains the accumulated delay time in nanoseconds.
815 */
816 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
817 u64 blkio_delay; /* wait for sync block io completion */
818 u64 swapin_delay; /* wait for swapin block io completion */
819 u32 blkio_count; /* total count of the number of sync block */
820 /* io operations performed */
821 u32 swapin_count; /* total count of the number of swapin block */
822 /* io operations performed */
823
824 struct timespec freepages_start, freepages_end;
825 u64 freepages_delay; /* wait for memory reclaim */
826 u32 freepages_count; /* total count of memory reclaim */
827 };
828 #endif /* CONFIG_TASK_DELAY_ACCT */
829
830 static inline int sched_info_on(void)
831 {
832 #ifdef CONFIG_SCHEDSTATS
833 return 1;
834 #elif defined(CONFIG_TASK_DELAY_ACCT)
835 extern int delayacct_on;
836 return delayacct_on;
837 #else
838 return 0;
839 #endif
840 }
841
842 enum cpu_idle_type {
843 CPU_IDLE,
844 CPU_NOT_IDLE,
845 CPU_NEWLY_IDLE,
846 CPU_MAX_IDLE_TYPES
847 };
848
849 /*
850 * Increase resolution of cpu_capacity calculations
851 */
852 #define SCHED_CAPACITY_SHIFT 10
853 #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT)
854
855 /*
856 * sched-domains (multiprocessor balancing) declarations:
857 */
858 #ifdef CONFIG_SMP
859 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
860 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
861 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
862 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
863 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
864 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
865 #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */
866 #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */
867 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
868 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
869 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
870 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
871 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */
872 #define SD_NUMA 0x4000 /* cross-node balancing */
873
874 #ifdef CONFIG_SCHED_SMT
875 static inline const int cpu_smt_flags(void)
876 {
877 return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
878 }
879 #endif
880
881 #ifdef CONFIG_SCHED_MC
882 static inline const int cpu_core_flags(void)
883 {
884 return SD_SHARE_PKG_RESOURCES;
885 }
886 #endif
887
888 #ifdef CONFIG_NUMA
889 static inline const int cpu_numa_flags(void)
890 {
891 return SD_NUMA;
892 }
893 #endif
894
895 struct sched_domain_attr {
896 int relax_domain_level;
897 };
898
899 #define SD_ATTR_INIT (struct sched_domain_attr) { \
900 .relax_domain_level = -1, \
901 }
902
903 extern int sched_domain_level_max;
904
905 struct sched_group;
906
907 struct sched_domain {
908 /* These fields must be setup */
909 struct sched_domain *parent; /* top domain must be null terminated */
910 struct sched_domain *child; /* bottom domain must be null terminated */
911 struct sched_group *groups; /* the balancing groups of the domain */
912 unsigned long min_interval; /* Minimum balance interval ms */
913 unsigned long max_interval; /* Maximum balance interval ms */
914 unsigned int busy_factor; /* less balancing by factor if busy */
915 unsigned int imbalance_pct; /* No balance until over watermark */
916 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
917 unsigned int busy_idx;
918 unsigned int idle_idx;
919 unsigned int newidle_idx;
920 unsigned int wake_idx;
921 unsigned int forkexec_idx;
922 unsigned int smt_gain;
923
924 int nohz_idle; /* NOHZ IDLE status */
925 int flags; /* See SD_* */
926 int level;
927
928 /* Runtime fields. */
929 unsigned long last_balance; /* init to jiffies. units in jiffies */
930 unsigned int balance_interval; /* initialise to 1. units in ms. */
931 unsigned int nr_balance_failed; /* initialise to 0 */
932
933 /* idle_balance() stats */
934 u64 max_newidle_lb_cost;
935 unsigned long next_decay_max_lb_cost;
936
937 #ifdef CONFIG_SCHEDSTATS
938 /* load_balance() stats */
939 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
940 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
941 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
942 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
943 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
944 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
947
948 /* Active load balancing */
949 unsigned int alb_count;
950 unsigned int alb_failed;
951 unsigned int alb_pushed;
952
953 /* SD_BALANCE_EXEC stats */
954 unsigned int sbe_count;
955 unsigned int sbe_balanced;
956 unsigned int sbe_pushed;
957
958 /* SD_BALANCE_FORK stats */
959 unsigned int sbf_count;
960 unsigned int sbf_balanced;
961 unsigned int sbf_pushed;
962
963 /* try_to_wake_up() stats */
964 unsigned int ttwu_wake_remote;
965 unsigned int ttwu_move_affine;
966 unsigned int ttwu_move_balance;
967 #endif
968 #ifdef CONFIG_SCHED_DEBUG
969 char *name;
970 #endif
971 union {
972 void *private; /* used during construction */
973 struct rcu_head rcu; /* used during destruction */
974 };
975
976 unsigned int span_weight;
977 /*
978 * Span of all CPUs in this domain.
979 *
980 * NOTE: this field is variable length. (Allocated dynamically
981 * by attaching extra space to the end of the structure,
982 * depending on how many CPUs the kernel has booted up with)
983 */
984 unsigned long span[0];
985 };
986
987 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
988 {
989 return to_cpumask(sd->span);
990 }
991
992 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
993 struct sched_domain_attr *dattr_new);
994
995 /* Allocate an array of sched domains, for partition_sched_domains(). */
996 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
997 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
998
999 bool cpus_share_cache(int this_cpu, int that_cpu);
1000
1001 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1002 typedef const int (*sched_domain_flags_f)(void);
1003
1004 #define SDTL_OVERLAP 0x01
1005
1006 struct sd_data {
1007 struct sched_domain **__percpu sd;
1008 struct sched_group **__percpu sg;
1009 struct sched_group_capacity **__percpu sgc;
1010 };
1011
1012 struct sched_domain_topology_level {
1013 sched_domain_mask_f mask;
1014 sched_domain_flags_f sd_flags;
1015 int flags;
1016 int numa_level;
1017 struct sd_data data;
1018 #ifdef CONFIG_SCHED_DEBUG
1019 char *name;
1020 #endif
1021 };
1022
1023 extern struct sched_domain_topology_level *sched_domain_topology;
1024
1025 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1026
1027 #ifdef CONFIG_SCHED_DEBUG
1028 # define SD_INIT_NAME(type) .name = #type
1029 #else
1030 # define SD_INIT_NAME(type)
1031 #endif
1032
1033 #else /* CONFIG_SMP */
1034
1035 struct sched_domain_attr;
1036
1037 static inline void
1038 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1039 struct sched_domain_attr *dattr_new)
1040 {
1041 }
1042
1043 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1044 {
1045 return true;
1046 }
1047
1048 #endif /* !CONFIG_SMP */
1049
1050
1051 struct io_context; /* See blkdev.h */
1052
1053
1054 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1055 extern void prefetch_stack(struct task_struct *t);
1056 #else
1057 static inline void prefetch_stack(struct task_struct *t) { }
1058 #endif
1059
1060 struct audit_context; /* See audit.c */
1061 struct mempolicy;
1062 struct pipe_inode_info;
1063 struct uts_namespace;
1064
1065 struct load_weight {
1066 unsigned long weight;
1067 u32 inv_weight;
1068 };
1069
1070 struct sched_avg {
1071 /*
1072 * These sums represent an infinite geometric series and so are bound
1073 * above by 1024/(1-y). Thus we only need a u32 to store them for all
1074 * choices of y < 1-2^(-32)*1024.
1075 */
1076 u32 runnable_avg_sum, runnable_avg_period;
1077 u64 last_runnable_update;
1078 s64 decay_count;
1079 unsigned long load_avg_contrib;
1080 };
1081
1082 #ifdef CONFIG_SCHEDSTATS
1083 struct sched_statistics {
1084 u64 wait_start;
1085 u64 wait_max;
1086 u64 wait_count;
1087 u64 wait_sum;
1088 u64 iowait_count;
1089 u64 iowait_sum;
1090
1091 u64 sleep_start;
1092 u64 sleep_max;
1093 s64 sum_sleep_runtime;
1094
1095 u64 block_start;
1096 u64 block_max;
1097 u64 exec_max;
1098 u64 slice_max;
1099
1100 u64 nr_migrations_cold;
1101 u64 nr_failed_migrations_affine;
1102 u64 nr_failed_migrations_running;
1103 u64 nr_failed_migrations_hot;
1104 u64 nr_forced_migrations;
1105
1106 u64 nr_wakeups;
1107 u64 nr_wakeups_sync;
1108 u64 nr_wakeups_migrate;
1109 u64 nr_wakeups_local;
1110 u64 nr_wakeups_remote;
1111 u64 nr_wakeups_affine;
1112 u64 nr_wakeups_affine_attempts;
1113 u64 nr_wakeups_passive;
1114 u64 nr_wakeups_idle;
1115 };
1116 #endif
1117
1118 struct sched_entity {
1119 struct load_weight load; /* for load-balancing */
1120 struct rb_node run_node;
1121 struct list_head group_node;
1122 unsigned int on_rq;
1123
1124 u64 exec_start;
1125 u64 sum_exec_runtime;
1126 u64 vruntime;
1127 u64 prev_sum_exec_runtime;
1128
1129 u64 nr_migrations;
1130
1131 #ifdef CONFIG_SCHEDSTATS
1132 struct sched_statistics statistics;
1133 #endif
1134
1135 #ifdef CONFIG_FAIR_GROUP_SCHED
1136 int depth;
1137 struct sched_entity *parent;
1138 /* rq on which this entity is (to be) queued: */
1139 struct cfs_rq *cfs_rq;
1140 /* rq "owned" by this entity/group: */
1141 struct cfs_rq *my_q;
1142 #endif
1143
1144 #ifdef CONFIG_SMP
1145 /* Per-entity load-tracking */
1146 struct sched_avg avg;
1147 #endif
1148 };
1149
1150 struct sched_rt_entity {
1151 struct list_head run_list;
1152 unsigned long timeout;
1153 unsigned long watchdog_stamp;
1154 unsigned int time_slice;
1155
1156 struct sched_rt_entity *back;
1157 #ifdef CONFIG_RT_GROUP_SCHED
1158 struct sched_rt_entity *parent;
1159 /* rq on which this entity is (to be) queued: */
1160 struct rt_rq *rt_rq;
1161 /* rq "owned" by this entity/group: */
1162 struct rt_rq *my_q;
1163 #endif
1164 };
1165
1166 struct sched_dl_entity {
1167 struct rb_node rb_node;
1168
1169 /*
1170 * Original scheduling parameters. Copied here from sched_attr
1171 * during sched_setattr(), they will remain the same until
1172 * the next sched_setattr().
1173 */
1174 u64 dl_runtime; /* maximum runtime for each instance */
1175 u64 dl_deadline; /* relative deadline of each instance */
1176 u64 dl_period; /* separation of two instances (period) */
1177 u64 dl_bw; /* dl_runtime / dl_deadline */
1178
1179 /*
1180 * Actual scheduling parameters. Initialized with the values above,
1181 * they are continously updated during task execution. Note that
1182 * the remaining runtime could be < 0 in case we are in overrun.
1183 */
1184 s64 runtime; /* remaining runtime for this instance */
1185 u64 deadline; /* absolute deadline for this instance */
1186 unsigned int flags; /* specifying the scheduler behaviour */
1187
1188 /*
1189 * Some bool flags:
1190 *
1191 * @dl_throttled tells if we exhausted the runtime. If so, the
1192 * task has to wait for a replenishment to be performed at the
1193 * next firing of dl_timer.
1194 *
1195 * @dl_new tells if a new instance arrived. If so we must
1196 * start executing it with full runtime and reset its absolute
1197 * deadline;
1198 *
1199 * @dl_boosted tells if we are boosted due to DI. If so we are
1200 * outside bandwidth enforcement mechanism (but only until we
1201 * exit the critical section);
1202 *
1203 * @dl_yielded tells if task gave up the cpu before consuming
1204 * all its available runtime during the last job.
1205 */
1206 int dl_throttled, dl_new, dl_boosted, dl_yielded;
1207
1208 /*
1209 * Bandwidth enforcement timer. Each -deadline task has its
1210 * own bandwidth to be enforced, thus we need one timer per task.
1211 */
1212 struct hrtimer dl_timer;
1213 };
1214
1215 struct rcu_node;
1216
1217 enum perf_event_task_context {
1218 perf_invalid_context = -1,
1219 perf_hw_context = 0,
1220 perf_sw_context,
1221 perf_nr_task_contexts,
1222 };
1223
1224 struct task_struct {
1225 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1226 void *stack;
1227 atomic_t usage;
1228 unsigned int flags; /* per process flags, defined below */
1229 unsigned int ptrace;
1230
1231 #ifdef CONFIG_SMP
1232 struct llist_node wake_entry;
1233 int on_cpu;
1234 struct task_struct *last_wakee;
1235 unsigned long wakee_flips;
1236 unsigned long wakee_flip_decay_ts;
1237
1238 int wake_cpu;
1239 #endif
1240 int on_rq;
1241
1242 int prio, static_prio, normal_prio;
1243 unsigned int rt_priority;
1244 const struct sched_class *sched_class;
1245 struct sched_entity se;
1246 struct sched_rt_entity rt;
1247 #ifdef CONFIG_CGROUP_SCHED
1248 struct task_group *sched_task_group;
1249 #endif
1250 struct sched_dl_entity dl;
1251
1252 #ifdef CONFIG_PREEMPT_NOTIFIERS
1253 /* list of struct preempt_notifier: */
1254 struct hlist_head preempt_notifiers;
1255 #endif
1256
1257 #ifdef CONFIG_BLK_DEV_IO_TRACE
1258 unsigned int btrace_seq;
1259 #endif
1260
1261 unsigned int policy;
1262 int nr_cpus_allowed;
1263 cpumask_t cpus_allowed;
1264
1265 #ifdef CONFIG_PREEMPT_RCU
1266 int rcu_read_lock_nesting;
1267 char rcu_read_unlock_special;
1268 struct list_head rcu_node_entry;
1269 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1270 #ifdef CONFIG_TREE_PREEMPT_RCU
1271 struct rcu_node *rcu_blocked_node;
1272 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1273 #ifdef CONFIG_RCU_BOOST
1274 struct rt_mutex *rcu_boost_mutex;
1275 #endif /* #ifdef CONFIG_RCU_BOOST */
1276
1277 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1278 struct sched_info sched_info;
1279 #endif
1280
1281 struct list_head tasks;
1282 #ifdef CONFIG_SMP
1283 struct plist_node pushable_tasks;
1284 struct rb_node pushable_dl_tasks;
1285 #endif
1286
1287 struct mm_struct *mm, *active_mm;
1288 #ifdef CONFIG_COMPAT_BRK
1289 unsigned brk_randomized:1;
1290 #endif
1291 /* per-thread vma caching */
1292 u32 vmacache_seqnum;
1293 struct vm_area_struct *vmacache[VMACACHE_SIZE];
1294 #if defined(SPLIT_RSS_COUNTING)
1295 struct task_rss_stat rss_stat;
1296 #endif
1297 /* task state */
1298 int exit_state;
1299 int exit_code, exit_signal;
1300 int pdeath_signal; /* The signal sent when the parent dies */
1301 unsigned int jobctl; /* JOBCTL_*, siglock protected */
1302
1303 /* Used for emulating ABI behavior of previous Linux versions */
1304 unsigned int personality;
1305
1306 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1307 * execve */
1308 unsigned in_iowait:1;
1309
1310 /* Revert to default priority/policy when forking */
1311 unsigned sched_reset_on_fork:1;
1312 unsigned sched_contributes_to_load:1;
1313
1314 unsigned long atomic_flags; /* Flags needing atomic access. */
1315
1316 pid_t pid;
1317 pid_t tgid;
1318
1319 #ifdef CONFIG_CC_STACKPROTECTOR
1320 /* Canary value for the -fstack-protector gcc feature */
1321 unsigned long stack_canary;
1322 #endif
1323 /*
1324 * pointers to (original) parent process, youngest child, younger sibling,
1325 * older sibling, respectively. (p->father can be replaced with
1326 * p->real_parent->pid)
1327 */
1328 struct task_struct __rcu *real_parent; /* real parent process */
1329 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1330 /*
1331 * children/sibling forms the list of my natural children
1332 */
1333 struct list_head children; /* list of my children */
1334 struct list_head sibling; /* linkage in my parent's children list */
1335 struct task_struct *group_leader; /* threadgroup leader */
1336
1337 /*
1338 * ptraced is the list of tasks this task is using ptrace on.
1339 * This includes both natural children and PTRACE_ATTACH targets.
1340 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1341 */
1342 struct list_head ptraced;
1343 struct list_head ptrace_entry;
1344
1345 /* PID/PID hash table linkage. */
1346 struct pid_link pids[PIDTYPE_MAX];
1347 struct list_head thread_group;
1348 struct list_head thread_node;
1349
1350 struct completion *vfork_done; /* for vfork() */
1351 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1352 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1353
1354 cputime_t utime, stime, utimescaled, stimescaled;
1355 cputime_t gtime;
1356 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
1357 struct cputime prev_cputime;
1358 #endif
1359 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1360 seqlock_t vtime_seqlock;
1361 unsigned long long vtime_snap;
1362 enum {
1363 VTIME_SLEEPING = 0,
1364 VTIME_USER,
1365 VTIME_SYS,
1366 } vtime_snap_whence;
1367 #endif
1368 unsigned long nvcsw, nivcsw; /* context switch counts */
1369 struct timespec start_time; /* monotonic time */
1370 struct timespec real_start_time; /* boot based time */
1371 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1372 unsigned long min_flt, maj_flt;
1373
1374 struct task_cputime cputime_expires;
1375 struct list_head cpu_timers[3];
1376
1377 /* process credentials */
1378 const struct cred __rcu *real_cred; /* objective and real subjective task
1379 * credentials (COW) */
1380 const struct cred __rcu *cred; /* effective (overridable) subjective task
1381 * credentials (COW) */
1382 char comm[TASK_COMM_LEN]; /* executable name excluding path
1383 - access with [gs]et_task_comm (which lock
1384 it with task_lock())
1385 - initialized normally by setup_new_exec */
1386 /* file system info */
1387 int link_count, total_link_count;
1388 #ifdef CONFIG_SYSVIPC
1389 /* ipc stuff */
1390 struct sysv_sem sysvsem;
1391 #endif
1392 #ifdef CONFIG_DETECT_HUNG_TASK
1393 /* hung task detection */
1394 unsigned long last_switch_count;
1395 #endif
1396 /* CPU-specific state of this task */
1397 struct thread_struct thread;
1398 /* filesystem information */
1399 struct fs_struct *fs;
1400 /* open file information */
1401 struct files_struct *files;
1402 /* namespaces */
1403 struct nsproxy *nsproxy;
1404 /* signal handlers */
1405 struct signal_struct *signal;
1406 struct sighand_struct *sighand;
1407
1408 sigset_t blocked, real_blocked;
1409 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1410 struct sigpending pending;
1411
1412 unsigned long sas_ss_sp;
1413 size_t sas_ss_size;
1414 int (*notifier)(void *priv);
1415 void *notifier_data;
1416 sigset_t *notifier_mask;
1417 struct callback_head *task_works;
1418
1419 struct audit_context *audit_context;
1420 #ifdef CONFIG_AUDITSYSCALL
1421 kuid_t loginuid;
1422 unsigned int sessionid;
1423 #endif
1424 struct seccomp seccomp;
1425
1426 /* Thread group tracking */
1427 u32 parent_exec_id;
1428 u32 self_exec_id;
1429 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1430 * mempolicy */
1431 spinlock_t alloc_lock;
1432
1433 /* Protection of the PI data structures: */
1434 raw_spinlock_t pi_lock;
1435
1436 #ifdef CONFIG_RT_MUTEXES
1437 /* PI waiters blocked on a rt_mutex held by this task */
1438 struct rb_root pi_waiters;
1439 struct rb_node *pi_waiters_leftmost;
1440 /* Deadlock detection and priority inheritance handling */
1441 struct rt_mutex_waiter *pi_blocked_on;
1442 /* Top pi_waiters task */
1443 struct task_struct *pi_top_task;
1444 #endif
1445
1446 #ifdef CONFIG_DEBUG_MUTEXES
1447 /* mutex deadlock detection */
1448 struct mutex_waiter *blocked_on;
1449 #endif
1450 #ifdef CONFIG_TRACE_IRQFLAGS
1451 unsigned int irq_events;
1452 unsigned long hardirq_enable_ip;
1453 unsigned long hardirq_disable_ip;
1454 unsigned int hardirq_enable_event;
1455 unsigned int hardirq_disable_event;
1456 int hardirqs_enabled;
1457 int hardirq_context;
1458 unsigned long softirq_disable_ip;
1459 unsigned long softirq_enable_ip;
1460 unsigned int softirq_disable_event;
1461 unsigned int softirq_enable_event;
1462 int softirqs_enabled;
1463 int softirq_context;
1464 #endif
1465 #ifdef CONFIG_LOCKDEP
1466 # define MAX_LOCK_DEPTH 48UL
1467 u64 curr_chain_key;
1468 int lockdep_depth;
1469 unsigned int lockdep_recursion;
1470 struct held_lock held_locks[MAX_LOCK_DEPTH];
1471 gfp_t lockdep_reclaim_gfp;
1472 #endif
1473
1474 /* journalling filesystem info */
1475 void *journal_info;
1476
1477 /* stacked block device info */
1478 struct bio_list *bio_list;
1479
1480 #ifdef CONFIG_BLOCK
1481 /* stack plugging */
1482 struct blk_plug *plug;
1483 #endif
1484
1485 /* VM state */
1486 struct reclaim_state *reclaim_state;
1487
1488 struct backing_dev_info *backing_dev_info;
1489
1490 struct io_context *io_context;
1491
1492 unsigned long ptrace_message;
1493 siginfo_t *last_siginfo; /* For ptrace use. */
1494 struct task_io_accounting ioac;
1495 #if defined(CONFIG_TASK_XACCT)
1496 u64 acct_rss_mem1; /* accumulated rss usage */
1497 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1498 cputime_t acct_timexpd; /* stime + utime since last update */
1499 #endif
1500 #ifdef CONFIG_CPUSETS
1501 nodemask_t mems_allowed; /* Protected by alloc_lock */
1502 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */
1503 int cpuset_mem_spread_rotor;
1504 int cpuset_slab_spread_rotor;
1505 #endif
1506 #ifdef CONFIG_CGROUPS
1507 /* Control Group info protected by css_set_lock */
1508 struct css_set __rcu *cgroups;
1509 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1510 struct list_head cg_list;
1511 #endif
1512 #ifdef CONFIG_FUTEX
1513 struct robust_list_head __user *robust_list;
1514 #ifdef CONFIG_COMPAT
1515 struct compat_robust_list_head __user *compat_robust_list;
1516 #endif
1517 struct list_head pi_state_list;
1518 struct futex_pi_state *pi_state_cache;
1519 #endif
1520 #ifdef CONFIG_PERF_EVENTS
1521 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1522 struct mutex perf_event_mutex;
1523 struct list_head perf_event_list;
1524 #endif
1525 #ifdef CONFIG_DEBUG_PREEMPT
1526 unsigned long preempt_disable_ip;
1527 #endif
1528 #ifdef CONFIG_NUMA
1529 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1530 short il_next;
1531 short pref_node_fork;
1532 #endif
1533 #ifdef CONFIG_NUMA_BALANCING
1534 int numa_scan_seq;
1535 unsigned int numa_scan_period;
1536 unsigned int numa_scan_period_max;
1537 int numa_preferred_nid;
1538 unsigned long numa_migrate_retry;
1539 u64 node_stamp; /* migration stamp */
1540 u64 last_task_numa_placement;
1541 u64 last_sum_exec_runtime;
1542 struct callback_head numa_work;
1543
1544 struct list_head numa_entry;
1545 struct numa_group *numa_group;
1546
1547 /*
1548 * Exponential decaying average of faults on a per-node basis.
1549 * Scheduling placement decisions are made based on the these counts.
1550 * The values remain static for the duration of a PTE scan
1551 */
1552 unsigned long *numa_faults_memory;
1553 unsigned long total_numa_faults;
1554
1555 /*
1556 * numa_faults_buffer records faults per node during the current
1557 * scan window. When the scan completes, the counts in
1558 * numa_faults_memory decay and these values are copied.
1559 */
1560 unsigned long *numa_faults_buffer_memory;
1561
1562 /*
1563 * Track the nodes the process was running on when a NUMA hinting
1564 * fault was incurred.
1565 */
1566 unsigned long *numa_faults_cpu;
1567 unsigned long *numa_faults_buffer_cpu;
1568
1569 /*
1570 * numa_faults_locality tracks if faults recorded during the last
1571 * scan window were remote/local. The task scan period is adapted
1572 * based on the locality of the faults with different weights
1573 * depending on whether they were shared or private faults
1574 */
1575 unsigned long numa_faults_locality[2];
1576
1577 unsigned long numa_pages_migrated;
1578 #endif /* CONFIG_NUMA_BALANCING */
1579
1580 struct rcu_head rcu;
1581
1582 /*
1583 * cache last used pipe for splice
1584 */
1585 struct pipe_inode_info *splice_pipe;
1586
1587 struct page_frag task_frag;
1588
1589 #ifdef CONFIG_TASK_DELAY_ACCT
1590 struct task_delay_info *delays;
1591 #endif
1592 #ifdef CONFIG_FAULT_INJECTION
1593 int make_it_fail;
1594 #endif
1595 /*
1596 * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1597 * balance_dirty_pages() for some dirty throttling pause
1598 */
1599 int nr_dirtied;
1600 int nr_dirtied_pause;
1601 unsigned long dirty_paused_when; /* start of a write-and-pause period */
1602
1603 #ifdef CONFIG_LATENCYTOP
1604 int latency_record_count;
1605 struct latency_record latency_record[LT_SAVECOUNT];
1606 #endif
1607 /*
1608 * time slack values; these are used to round up poll() and
1609 * select() etc timeout values. These are in nanoseconds.
1610 */
1611 unsigned long timer_slack_ns;
1612 unsigned long default_timer_slack_ns;
1613
1614 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1615 /* Index of current stored address in ret_stack */
1616 int curr_ret_stack;
1617 /* Stack of return addresses for return function tracing */
1618 struct ftrace_ret_stack *ret_stack;
1619 /* time stamp for last schedule */
1620 unsigned long long ftrace_timestamp;
1621 /*
1622 * Number of functions that haven't been traced
1623 * because of depth overrun.
1624 */
1625 atomic_t trace_overrun;
1626 /* Pause for the tracing */
1627 atomic_t tracing_graph_pause;
1628 #endif
1629 #ifdef CONFIG_TRACING
1630 /* state flags for use by tracers */
1631 unsigned long trace;
1632 /* bitmask and counter of trace recursion */
1633 unsigned long trace_recursion;
1634 #endif /* CONFIG_TRACING */
1635 #ifdef CONFIG_MEMCG /* memcg uses this to do batch job */
1636 struct memcg_batch_info {
1637 int do_batch; /* incremented when batch uncharge started */
1638 struct mem_cgroup *memcg; /* target memcg of uncharge */
1639 unsigned long nr_pages; /* uncharged usage */
1640 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1641 } memcg_batch;
1642 unsigned int memcg_kmem_skip_account;
1643 struct memcg_oom_info {
1644 struct mem_cgroup *memcg;
1645 gfp_t gfp_mask;
1646 int order;
1647 unsigned int may_oom:1;
1648 } memcg_oom;
1649 #endif
1650 #ifdef CONFIG_UPROBES
1651 struct uprobe_task *utask;
1652 #endif
1653 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1654 unsigned int sequential_io;
1655 unsigned int sequential_io_avg;
1656 #endif
1657 };
1658
1659 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1660 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1661
1662 #define TNF_MIGRATED 0x01
1663 #define TNF_NO_GROUP 0x02
1664 #define TNF_SHARED 0x04
1665 #define TNF_FAULT_LOCAL 0x08
1666
1667 #ifdef CONFIG_NUMA_BALANCING
1668 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1669 extern pid_t task_numa_group_id(struct task_struct *p);
1670 extern void set_numabalancing_state(bool enabled);
1671 extern void task_numa_free(struct task_struct *p);
1672 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1673 int src_nid, int dst_cpu);
1674 #else
1675 static inline void task_numa_fault(int last_node, int node, int pages,
1676 int flags)
1677 {
1678 }
1679 static inline pid_t task_numa_group_id(struct task_struct *p)
1680 {
1681 return 0;
1682 }
1683 static inline void set_numabalancing_state(bool enabled)
1684 {
1685 }
1686 static inline void task_numa_free(struct task_struct *p)
1687 {
1688 }
1689 static inline bool should_numa_migrate_memory(struct task_struct *p,
1690 struct page *page, int src_nid, int dst_cpu)
1691 {
1692 return true;
1693 }
1694 #endif
1695
1696 static inline struct pid *task_pid(struct task_struct *task)
1697 {
1698 return task->pids[PIDTYPE_PID].pid;
1699 }
1700
1701 static inline struct pid *task_tgid(struct task_struct *task)
1702 {
1703 return task->group_leader->pids[PIDTYPE_PID].pid;
1704 }
1705
1706 /*
1707 * Without tasklist or rcu lock it is not safe to dereference
1708 * the result of task_pgrp/task_session even if task == current,
1709 * we can race with another thread doing sys_setsid/sys_setpgid.
1710 */
1711 static inline struct pid *task_pgrp(struct task_struct *task)
1712 {
1713 return task->group_leader->pids[PIDTYPE_PGID].pid;
1714 }
1715
1716 static inline struct pid *task_session(struct task_struct *task)
1717 {
1718 return task->group_leader->pids[PIDTYPE_SID].pid;
1719 }
1720
1721 struct pid_namespace;
1722
1723 /*
1724 * the helpers to get the task's different pids as they are seen
1725 * from various namespaces
1726 *
1727 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1728 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1729 * current.
1730 * task_xid_nr_ns() : id seen from the ns specified;
1731 *
1732 * set_task_vxid() : assigns a virtual id to a task;
1733 *
1734 * see also pid_nr() etc in include/linux/pid.h
1735 */
1736 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1737 struct pid_namespace *ns);
1738
1739 static inline pid_t task_pid_nr(struct task_struct *tsk)
1740 {
1741 return tsk->pid;
1742 }
1743
1744 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1745 struct pid_namespace *ns)
1746 {
1747 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1748 }
1749
1750 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1751 {
1752 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1753 }
1754
1755
1756 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1757 {
1758 return tsk->tgid;
1759 }
1760
1761 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1762
1763 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1764 {
1765 return pid_vnr(task_tgid(tsk));
1766 }
1767
1768
1769 static inline int pid_alive(const struct task_struct *p);
1770 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
1771 {
1772 pid_t pid = 0;
1773
1774 rcu_read_lock();
1775 if (pid_alive(tsk))
1776 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
1777 rcu_read_unlock();
1778
1779 return pid;
1780 }
1781
1782 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
1783 {
1784 return task_ppid_nr_ns(tsk, &init_pid_ns);
1785 }
1786
1787 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1788 struct pid_namespace *ns)
1789 {
1790 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1791 }
1792
1793 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1794 {
1795 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1796 }
1797
1798
1799 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1800 struct pid_namespace *ns)
1801 {
1802 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1803 }
1804
1805 static inline pid_t task_session_vnr(struct task_struct *tsk)
1806 {
1807 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1808 }
1809
1810 /* obsolete, do not use */
1811 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1812 {
1813 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1814 }
1815
1816 /**
1817 * pid_alive - check that a task structure is not stale
1818 * @p: Task structure to be checked.
1819 *
1820 * Test if a process is not yet dead (at most zombie state)
1821 * If pid_alive fails, then pointers within the task structure
1822 * can be stale and must not be dereferenced.
1823 *
1824 * Return: 1 if the process is alive. 0 otherwise.
1825 */
1826 static inline int pid_alive(const struct task_struct *p)
1827 {
1828 return p->pids[PIDTYPE_PID].pid != NULL;
1829 }
1830
1831 /**
1832 * is_global_init - check if a task structure is init
1833 * @tsk: Task structure to be checked.
1834 *
1835 * Check if a task structure is the first user space task the kernel created.
1836 *
1837 * Return: 1 if the task structure is init. 0 otherwise.
1838 */
1839 static inline int is_global_init(struct task_struct *tsk)
1840 {
1841 return tsk->pid == 1;
1842 }
1843
1844 extern struct pid *cad_pid;
1845
1846 extern void free_task(struct task_struct *tsk);
1847 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1848
1849 extern void __put_task_struct(struct task_struct *t);
1850
1851 static inline void put_task_struct(struct task_struct *t)
1852 {
1853 if (atomic_dec_and_test(&t->usage))
1854 __put_task_struct(t);
1855 }
1856
1857 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1858 extern void task_cputime(struct task_struct *t,
1859 cputime_t *utime, cputime_t *stime);
1860 extern void task_cputime_scaled(struct task_struct *t,
1861 cputime_t *utimescaled, cputime_t *stimescaled);
1862 extern cputime_t task_gtime(struct task_struct *t);
1863 #else
1864 static inline void task_cputime(struct task_struct *t,
1865 cputime_t *utime, cputime_t *stime)
1866 {
1867 if (utime)
1868 *utime = t->utime;
1869 if (stime)
1870 *stime = t->stime;
1871 }
1872
1873 static inline void task_cputime_scaled(struct task_struct *t,
1874 cputime_t *utimescaled,
1875 cputime_t *stimescaled)
1876 {
1877 if (utimescaled)
1878 *utimescaled = t->utimescaled;
1879 if (stimescaled)
1880 *stimescaled = t->stimescaled;
1881 }
1882
1883 static inline cputime_t task_gtime(struct task_struct *t)
1884 {
1885 return t->gtime;
1886 }
1887 #endif
1888 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1889 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
1890
1891 /*
1892 * Per process flags
1893 */
1894 #define PF_EXITING 0x00000004 /* getting shut down */
1895 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1896 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1897 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1898 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1899 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1900 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1901 #define PF_DUMPCORE 0x00000200 /* dumped core */
1902 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1903 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1904 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */
1905 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1906 #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */
1907 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1908 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1909 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1910 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1911 #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */
1912 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1913 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1914 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1915 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1916 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1917 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1918 #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */
1919 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1920 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1921 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1922 #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */
1923
1924 /*
1925 * Only the _current_ task can read/write to tsk->flags, but other
1926 * tasks can access tsk->flags in readonly mode for example
1927 * with tsk_used_math (like during threaded core dumping).
1928 * There is however an exception to this rule during ptrace
1929 * or during fork: the ptracer task is allowed to write to the
1930 * child->flags of its traced child (same goes for fork, the parent
1931 * can write to the child->flags), because we're guaranteed the
1932 * child is not running and in turn not changing child->flags
1933 * at the same time the parent does it.
1934 */
1935 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1936 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1937 #define clear_used_math() clear_stopped_child_used_math(current)
1938 #define set_used_math() set_stopped_child_used_math(current)
1939 #define conditional_stopped_child_used_math(condition, child) \
1940 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1941 #define conditional_used_math(condition) \
1942 conditional_stopped_child_used_math(condition, current)
1943 #define copy_to_stopped_child_used_math(child) \
1944 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1945 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1946 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1947 #define used_math() tsk_used_math(current)
1948
1949 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags */
1950 static inline gfp_t memalloc_noio_flags(gfp_t flags)
1951 {
1952 if (unlikely(current->flags & PF_MEMALLOC_NOIO))
1953 flags &= ~__GFP_IO;
1954 return flags;
1955 }
1956
1957 static inline unsigned int memalloc_noio_save(void)
1958 {
1959 unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
1960 current->flags |= PF_MEMALLOC_NOIO;
1961 return flags;
1962 }
1963
1964 static inline void memalloc_noio_restore(unsigned int flags)
1965 {
1966 current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
1967 }
1968
1969 /* Per-process atomic flags. */
1970 #define PFA_NO_NEW_PRIVS 0x00000001 /* May not gain new privileges. */
1971
1972 static inline bool task_no_new_privs(struct task_struct *p)
1973 {
1974 return test_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1975 }
1976
1977 static inline void task_set_no_new_privs(struct task_struct *p)
1978 {
1979 set_bit(PFA_NO_NEW_PRIVS, &p->atomic_flags);
1980 }
1981
1982 /*
1983 * task->jobctl flags
1984 */
1985 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */
1986
1987 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */
1988 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */
1989 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */
1990 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */
1991 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */
1992 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */
1993 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */
1994
1995 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT)
1996 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT)
1997 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT)
1998 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT)
1999 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT)
2000 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT)
2001 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT)
2002
2003 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2004 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2005
2006 extern bool task_set_jobctl_pending(struct task_struct *task,
2007 unsigned int mask);
2008 extern void task_clear_jobctl_trapping(struct task_struct *task);
2009 extern void task_clear_jobctl_pending(struct task_struct *task,
2010 unsigned int mask);
2011
2012 #ifdef CONFIG_PREEMPT_RCU
2013
2014 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
2015 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
2016
2017 static inline void rcu_copy_process(struct task_struct *p)
2018 {
2019 p->rcu_read_lock_nesting = 0;
2020 p->rcu_read_unlock_special = 0;
2021 #ifdef CONFIG_TREE_PREEMPT_RCU
2022 p->rcu_blocked_node = NULL;
2023 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
2024 #ifdef CONFIG_RCU_BOOST
2025 p->rcu_boost_mutex = NULL;
2026 #endif /* #ifdef CONFIG_RCU_BOOST */
2027 INIT_LIST_HEAD(&p->rcu_node_entry);
2028 }
2029
2030 #else
2031
2032 static inline void rcu_copy_process(struct task_struct *p)
2033 {
2034 }
2035
2036 #endif
2037
2038 static inline void tsk_restore_flags(struct task_struct *task,
2039 unsigned long orig_flags, unsigned long flags)
2040 {
2041 task->flags &= ~flags;
2042 task->flags |= orig_flags & flags;
2043 }
2044
2045 #ifdef CONFIG_SMP
2046 extern void do_set_cpus_allowed(struct task_struct *p,
2047 const struct cpumask *new_mask);
2048
2049 extern int set_cpus_allowed_ptr(struct task_struct *p,
2050 const struct cpumask *new_mask);
2051 #else
2052 static inline void do_set_cpus_allowed(struct task_struct *p,
2053 const struct cpumask *new_mask)
2054 {
2055 }
2056 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2057 const struct cpumask *new_mask)
2058 {
2059 if (!cpumask_test_cpu(0, new_mask))
2060 return -EINVAL;
2061 return 0;
2062 }
2063 #endif
2064
2065 #ifdef CONFIG_NO_HZ_COMMON
2066 void calc_load_enter_idle(void);
2067 void calc_load_exit_idle(void);
2068 #else
2069 static inline void calc_load_enter_idle(void) { }
2070 static inline void calc_load_exit_idle(void) { }
2071 #endif /* CONFIG_NO_HZ_COMMON */
2072
2073 #ifndef CONFIG_CPUMASK_OFFSTACK
2074 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
2075 {
2076 return set_cpus_allowed_ptr(p, &new_mask);
2077 }
2078 #endif
2079
2080 /*
2081 * Do not use outside of architecture code which knows its limitations.
2082 *
2083 * sched_clock() has no promise of monotonicity or bounded drift between
2084 * CPUs, use (which you should not) requires disabling IRQs.
2085 *
2086 * Please use one of the three interfaces below.
2087 */
2088 extern unsigned long long notrace sched_clock(void);
2089 /*
2090 * See the comment in kernel/sched/clock.c
2091 */
2092 extern u64 cpu_clock(int cpu);
2093 extern u64 local_clock(void);
2094 extern u64 sched_clock_cpu(int cpu);
2095
2096
2097 extern void sched_clock_init(void);
2098
2099 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2100 static inline void sched_clock_tick(void)
2101 {
2102 }
2103
2104 static inline void sched_clock_idle_sleep_event(void)
2105 {
2106 }
2107
2108 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2109 {
2110 }
2111 #else
2112 /*
2113 * Architectures can set this to 1 if they have specified
2114 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2115 * but then during bootup it turns out that sched_clock()
2116 * is reliable after all:
2117 */
2118 extern int sched_clock_stable(void);
2119 extern void set_sched_clock_stable(void);
2120 extern void clear_sched_clock_stable(void);
2121
2122 extern void sched_clock_tick(void);
2123 extern void sched_clock_idle_sleep_event(void);
2124 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2125 #endif
2126
2127 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2128 /*
2129 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2130 * The reason for this explicit opt-in is not to have perf penalty with
2131 * slow sched_clocks.
2132 */
2133 extern void enable_sched_clock_irqtime(void);
2134 extern void disable_sched_clock_irqtime(void);
2135 #else
2136 static inline void enable_sched_clock_irqtime(void) {}
2137 static inline void disable_sched_clock_irqtime(void) {}
2138 #endif
2139
2140 extern unsigned long long
2141 task_sched_runtime(struct task_struct *task);
2142
2143 /* sched_exec is called by processes performing an exec */
2144 #ifdef CONFIG_SMP
2145 extern void sched_exec(void);
2146 #else
2147 #define sched_exec() {}
2148 #endif
2149
2150 extern void sched_clock_idle_sleep_event(void);
2151 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2152
2153 #ifdef CONFIG_HOTPLUG_CPU
2154 extern void idle_task_exit(void);
2155 #else
2156 static inline void idle_task_exit(void) {}
2157 #endif
2158
2159 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2160 extern void wake_up_nohz_cpu(int cpu);
2161 #else
2162 static inline void wake_up_nohz_cpu(int cpu) { }
2163 #endif
2164
2165 #ifdef CONFIG_NO_HZ_FULL
2166 extern bool sched_can_stop_tick(void);
2167 extern u64 scheduler_tick_max_deferment(void);
2168 #else
2169 static inline bool sched_can_stop_tick(void) { return false; }
2170 #endif
2171
2172 #ifdef CONFIG_SCHED_AUTOGROUP
2173 extern void sched_autogroup_create_attach(struct task_struct *p);
2174 extern void sched_autogroup_detach(struct task_struct *p);
2175 extern void sched_autogroup_fork(struct signal_struct *sig);
2176 extern void sched_autogroup_exit(struct signal_struct *sig);
2177 #ifdef CONFIG_PROC_FS
2178 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2179 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2180 #endif
2181 #else
2182 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2183 static inline void sched_autogroup_detach(struct task_struct *p) { }
2184 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2185 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2186 #endif
2187
2188 extern int yield_to(struct task_struct *p, bool preempt);
2189 extern void set_user_nice(struct task_struct *p, long nice);
2190 extern int task_prio(const struct task_struct *p);
2191 /**
2192 * task_nice - return the nice value of a given task.
2193 * @p: the task in question.
2194 *
2195 * Return: The nice value [ -20 ... 0 ... 19 ].
2196 */
2197 static inline int task_nice(const struct task_struct *p)
2198 {
2199 return PRIO_TO_NICE((p)->static_prio);
2200 }
2201 extern int can_nice(const struct task_struct *p, const int nice);
2202 extern int task_curr(const struct task_struct *p);
2203 extern int idle_cpu(int cpu);
2204 extern int sched_setscheduler(struct task_struct *, int,
2205 const struct sched_param *);
2206 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2207 const struct sched_param *);
2208 extern int sched_setattr(struct task_struct *,
2209 const struct sched_attr *);
2210 extern struct task_struct *idle_task(int cpu);
2211 /**
2212 * is_idle_task - is the specified task an idle task?
2213 * @p: the task in question.
2214 *
2215 * Return: 1 if @p is an idle task. 0 otherwise.
2216 */
2217 static inline bool is_idle_task(const struct task_struct *p)
2218 {
2219 return p->pid == 0;
2220 }
2221 extern struct task_struct *curr_task(int cpu);
2222 extern void set_curr_task(int cpu, struct task_struct *p);
2223
2224 void yield(void);
2225
2226 /*
2227 * The default (Linux) execution domain.
2228 */
2229 extern struct exec_domain default_exec_domain;
2230
2231 union thread_union {
2232 struct thread_info thread_info;
2233 unsigned long stack[THREAD_SIZE/sizeof(long)];
2234 };
2235
2236 #ifndef __HAVE_ARCH_KSTACK_END
2237 static inline int kstack_end(void *addr)
2238 {
2239 /* Reliable end of stack detection:
2240 * Some APM bios versions misalign the stack
2241 */
2242 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2243 }
2244 #endif
2245
2246 extern union thread_union init_thread_union;
2247 extern struct task_struct init_task;
2248
2249 extern struct mm_struct init_mm;
2250
2251 extern struct pid_namespace init_pid_ns;
2252
2253 /*
2254 * find a task by one of its numerical ids
2255 *
2256 * find_task_by_pid_ns():
2257 * finds a task by its pid in the specified namespace
2258 * find_task_by_vpid():
2259 * finds a task by its virtual pid
2260 *
2261 * see also find_vpid() etc in include/linux/pid.h
2262 */
2263
2264 extern struct task_struct *find_task_by_vpid(pid_t nr);
2265 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2266 struct pid_namespace *ns);
2267
2268 /* per-UID process charging. */
2269 extern struct user_struct * alloc_uid(kuid_t);
2270 static inline struct user_struct *get_uid(struct user_struct *u)
2271 {
2272 atomic_inc(&u->__count);
2273 return u;
2274 }
2275 extern void free_uid(struct user_struct *);
2276
2277 #include <asm/current.h>
2278
2279 extern void xtime_update(unsigned long ticks);
2280
2281 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2282 extern int wake_up_process(struct task_struct *tsk);
2283 extern void wake_up_new_task(struct task_struct *tsk);
2284 #ifdef CONFIG_SMP
2285 extern void kick_process(struct task_struct *tsk);
2286 #else
2287 static inline void kick_process(struct task_struct *tsk) { }
2288 #endif
2289 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2290 extern void sched_dead(struct task_struct *p);
2291
2292 extern void proc_caches_init(void);
2293 extern void flush_signals(struct task_struct *);
2294 extern void __flush_signals(struct task_struct *);
2295 extern void ignore_signals(struct task_struct *);
2296 extern void flush_signal_handlers(struct task_struct *, int force_default);
2297 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2298
2299 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2300 {
2301 unsigned long flags;
2302 int ret;
2303
2304 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2305 ret = dequeue_signal(tsk, mask, info);
2306 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2307
2308 return ret;
2309 }
2310
2311 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2312 sigset_t *mask);
2313 extern void unblock_all_signals(void);
2314 extern void release_task(struct task_struct * p);
2315 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2316 extern int force_sigsegv(int, struct task_struct *);
2317 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2318 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2319 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2320 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2321 const struct cred *, u32);
2322 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2323 extern int kill_pid(struct pid *pid, int sig, int priv);
2324 extern int kill_proc_info(int, struct siginfo *, pid_t);
2325 extern __must_check bool do_notify_parent(struct task_struct *, int);
2326 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2327 extern void force_sig(int, struct task_struct *);
2328 extern int send_sig(int, struct task_struct *, int);
2329 extern int zap_other_threads(struct task_struct *p);
2330 extern struct sigqueue *sigqueue_alloc(void);
2331 extern void sigqueue_free(struct sigqueue *);
2332 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2333 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2334
2335 static inline void restore_saved_sigmask(void)
2336 {
2337 if (test_and_clear_restore_sigmask())
2338 __set_current_blocked(&current->saved_sigmask);
2339 }
2340
2341 static inline sigset_t *sigmask_to_save(void)
2342 {
2343 sigset_t *res = &current->blocked;
2344 if (unlikely(test_restore_sigmask()))
2345 res = &current->saved_sigmask;
2346 return res;
2347 }
2348
2349 static inline int kill_cad_pid(int sig, int priv)
2350 {
2351 return kill_pid(cad_pid, sig, priv);
2352 }
2353
2354 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2355 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2356 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2357 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2358
2359 /*
2360 * True if we are on the alternate signal stack.
2361 */
2362 static inline int on_sig_stack(unsigned long sp)
2363 {
2364 #ifdef CONFIG_STACK_GROWSUP
2365 return sp >= current->sas_ss_sp &&
2366 sp - current->sas_ss_sp < current->sas_ss_size;
2367 #else
2368 return sp > current->sas_ss_sp &&
2369 sp - current->sas_ss_sp <= current->sas_ss_size;
2370 #endif
2371 }
2372
2373 static inline int sas_ss_flags(unsigned long sp)
2374 {
2375 return (current->sas_ss_size == 0 ? SS_DISABLE
2376 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2377 }
2378
2379 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2380 {
2381 if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2382 #ifdef CONFIG_STACK_GROWSUP
2383 return current->sas_ss_sp;
2384 #else
2385 return current->sas_ss_sp + current->sas_ss_size;
2386 #endif
2387 return sp;
2388 }
2389
2390 /*
2391 * Routines for handling mm_structs
2392 */
2393 extern struct mm_struct * mm_alloc(void);
2394
2395 /* mmdrop drops the mm and the page tables */
2396 extern void __mmdrop(struct mm_struct *);
2397 static inline void mmdrop(struct mm_struct * mm)
2398 {
2399 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2400 __mmdrop(mm);
2401 }
2402
2403 /* mmput gets rid of the mappings and all user-space */
2404 extern void mmput(struct mm_struct *);
2405 /* Grab a reference to a task's mm, if it is not already going away */
2406 extern struct mm_struct *get_task_mm(struct task_struct *task);
2407 /*
2408 * Grab a reference to a task's mm, if it is not already going away
2409 * and ptrace_may_access with the mode parameter passed to it
2410 * succeeds.
2411 */
2412 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2413 /* Remove the current tasks stale references to the old mm_struct */
2414 extern void mm_release(struct task_struct *, struct mm_struct *);
2415
2416 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2417 struct task_struct *);
2418 extern void flush_thread(void);
2419 extern void exit_thread(void);
2420
2421 extern void exit_files(struct task_struct *);
2422 extern void __cleanup_sighand(struct sighand_struct *);
2423
2424 extern void exit_itimers(struct signal_struct *);
2425 extern void flush_itimer_signals(void);
2426
2427 extern void do_group_exit(int);
2428
2429 extern int do_execve(struct filename *,
2430 const char __user * const __user *,
2431 const char __user * const __user *);
2432 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2433 struct task_struct *fork_idle(int);
2434 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2435
2436 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2437 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2438 {
2439 __set_task_comm(tsk, from, false);
2440 }
2441 extern char *get_task_comm(char *to, struct task_struct *tsk);
2442
2443 #ifdef CONFIG_SMP
2444 void scheduler_ipi(void);
2445 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2446 #else
2447 static inline void scheduler_ipi(void) { }
2448 static inline unsigned long wait_task_inactive(struct task_struct *p,
2449 long match_state)
2450 {
2451 return 1;
2452 }
2453 #endif
2454
2455 #define next_task(p) \
2456 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2457
2458 #define for_each_process(p) \
2459 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2460
2461 extern bool current_is_single_threaded(void);
2462
2463 /*
2464 * Careful: do_each_thread/while_each_thread is a double loop so
2465 * 'break' will not work as expected - use goto instead.
2466 */
2467 #define do_each_thread(g, t) \
2468 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2469
2470 #define while_each_thread(g, t) \
2471 while ((t = next_thread(t)) != g)
2472
2473 #define __for_each_thread(signal, t) \
2474 list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2475
2476 #define for_each_thread(p, t) \
2477 __for_each_thread((p)->signal, t)
2478
2479 /* Careful: this is a double loop, 'break' won't work as expected. */
2480 #define for_each_process_thread(p, t) \
2481 for_each_process(p) for_each_thread(p, t)
2482
2483 static inline int get_nr_threads(struct task_struct *tsk)
2484 {
2485 return tsk->signal->nr_threads;
2486 }
2487
2488 static inline bool thread_group_leader(struct task_struct *p)
2489 {
2490 return p->exit_signal >= 0;
2491 }
2492
2493 /* Do to the insanities of de_thread it is possible for a process
2494 * to have the pid of the thread group leader without actually being
2495 * the thread group leader. For iteration through the pids in proc
2496 * all we care about is that we have a task with the appropriate
2497 * pid, we don't actually care if we have the right task.
2498 */
2499 static inline bool has_group_leader_pid(struct task_struct *p)
2500 {
2501 return task_pid(p) == p->signal->leader_pid;
2502 }
2503
2504 static inline
2505 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2506 {
2507 return p1->signal == p2->signal;
2508 }
2509
2510 static inline struct task_struct *next_thread(const struct task_struct *p)
2511 {
2512 return list_entry_rcu(p->thread_group.next,
2513 struct task_struct, thread_group);
2514 }
2515
2516 static inline int thread_group_empty(struct task_struct *p)
2517 {
2518 return list_empty(&p->thread_group);
2519 }
2520
2521 #define delay_group_leader(p) \
2522 (thread_group_leader(p) && !thread_group_empty(p))
2523
2524 /*
2525 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2526 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2527 * pins the final release of task.io_context. Also protects ->cpuset and
2528 * ->cgroup.subsys[]. And ->vfork_done.
2529 *
2530 * Nests both inside and outside of read_lock(&tasklist_lock).
2531 * It must not be nested with write_lock_irq(&tasklist_lock),
2532 * neither inside nor outside.
2533 */
2534 static inline void task_lock(struct task_struct *p)
2535 {
2536 spin_lock(&p->alloc_lock);
2537 }
2538
2539 static inline void task_unlock(struct task_struct *p)
2540 {
2541 spin_unlock(&p->alloc_lock);
2542 }
2543
2544 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2545 unsigned long *flags);
2546
2547 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2548 unsigned long *flags)
2549 {
2550 struct sighand_struct *ret;
2551
2552 ret = __lock_task_sighand(tsk, flags);
2553 (void)__cond_lock(&tsk->sighand->siglock, ret);
2554 return ret;
2555 }
2556
2557 static inline void unlock_task_sighand(struct task_struct *tsk,
2558 unsigned long *flags)
2559 {
2560 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2561 }
2562
2563 #ifdef CONFIG_CGROUPS
2564 static inline void threadgroup_change_begin(struct task_struct *tsk)
2565 {
2566 down_read(&tsk->signal->group_rwsem);
2567 }
2568 static inline void threadgroup_change_end(struct task_struct *tsk)
2569 {
2570 up_read(&tsk->signal->group_rwsem);
2571 }
2572
2573 /**
2574 * threadgroup_lock - lock threadgroup
2575 * @tsk: member task of the threadgroup to lock
2576 *
2577 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter
2578 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or
2579 * change ->group_leader/pid. This is useful for cases where the threadgroup
2580 * needs to stay stable across blockable operations.
2581 *
2582 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for
2583 * synchronization. While held, no new task will be added to threadgroup
2584 * and no existing live task will have its PF_EXITING set.
2585 *
2586 * de_thread() does threadgroup_change_{begin|end}() when a non-leader
2587 * sub-thread becomes a new leader.
2588 */
2589 static inline void threadgroup_lock(struct task_struct *tsk)
2590 {
2591 down_write(&tsk->signal->group_rwsem);
2592 }
2593
2594 /**
2595 * threadgroup_unlock - unlock threadgroup
2596 * @tsk: member task of the threadgroup to unlock
2597 *
2598 * Reverse threadgroup_lock().
2599 */
2600 static inline void threadgroup_unlock(struct task_struct *tsk)
2601 {
2602 up_write(&tsk->signal->group_rwsem);
2603 }
2604 #else
2605 static inline void threadgroup_change_begin(struct task_struct *tsk) {}
2606 static inline void threadgroup_change_end(struct task_struct *tsk) {}
2607 static inline void threadgroup_lock(struct task_struct *tsk) {}
2608 static inline void threadgroup_unlock(struct task_struct *tsk) {}
2609 #endif
2610
2611 #ifndef __HAVE_THREAD_FUNCTIONS
2612
2613 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2614 #define task_stack_page(task) ((task)->stack)
2615
2616 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2617 {
2618 *task_thread_info(p) = *task_thread_info(org);
2619 task_thread_info(p)->task = p;
2620 }
2621
2622 static inline unsigned long *end_of_stack(struct task_struct *p)
2623 {
2624 return (unsigned long *)(task_thread_info(p) + 1);
2625 }
2626
2627 #endif
2628
2629 static inline int object_is_on_stack(void *obj)
2630 {
2631 void *stack = task_stack_page(current);
2632
2633 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2634 }
2635
2636 extern void thread_info_cache_init(void);
2637
2638 #ifdef CONFIG_DEBUG_STACK_USAGE
2639 static inline unsigned long stack_not_used(struct task_struct *p)
2640 {
2641 unsigned long *n = end_of_stack(p);
2642
2643 do { /* Skip over canary */
2644 n++;
2645 } while (!*n);
2646
2647 return (unsigned long)n - (unsigned long)end_of_stack(p);
2648 }
2649 #endif
2650
2651 /* set thread flags in other task's structures
2652 * - see asm/thread_info.h for TIF_xxxx flags available
2653 */
2654 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2655 {
2656 set_ti_thread_flag(task_thread_info(tsk), flag);
2657 }
2658
2659 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2660 {
2661 clear_ti_thread_flag(task_thread_info(tsk), flag);
2662 }
2663
2664 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2665 {
2666 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2667 }
2668
2669 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2670 {
2671 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2672 }
2673
2674 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2675 {
2676 return test_ti_thread_flag(task_thread_info(tsk), flag);
2677 }
2678
2679 static inline void set_tsk_need_resched(struct task_struct *tsk)
2680 {
2681 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2682 }
2683
2684 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2685 {
2686 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2687 }
2688
2689 static inline int test_tsk_need_resched(struct task_struct *tsk)
2690 {
2691 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2692 }
2693
2694 static inline int restart_syscall(void)
2695 {
2696 set_tsk_thread_flag(current, TIF_SIGPENDING);
2697 return -ERESTARTNOINTR;
2698 }
2699
2700 static inline int signal_pending(struct task_struct *p)
2701 {
2702 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2703 }
2704
2705 static inline int __fatal_signal_pending(struct task_struct *p)
2706 {
2707 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2708 }
2709
2710 static inline int fatal_signal_pending(struct task_struct *p)
2711 {
2712 return signal_pending(p) && __fatal_signal_pending(p);
2713 }
2714
2715 static inline int signal_pending_state(long state, struct task_struct *p)
2716 {
2717 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2718 return 0;
2719 if (!signal_pending(p))
2720 return 0;
2721
2722 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2723 }
2724
2725 /*
2726 * cond_resched() and cond_resched_lock(): latency reduction via
2727 * explicit rescheduling in places that are safe. The return
2728 * value indicates whether a reschedule was done in fact.
2729 * cond_resched_lock() will drop the spinlock before scheduling,
2730 * cond_resched_softirq() will enable bhs before scheduling.
2731 */
2732 extern int _cond_resched(void);
2733
2734 #define cond_resched() ({ \
2735 __might_sleep(__FILE__, __LINE__, 0); \
2736 _cond_resched(); \
2737 })
2738
2739 extern int __cond_resched_lock(spinlock_t *lock);
2740
2741 #ifdef CONFIG_PREEMPT_COUNT
2742 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2743 #else
2744 #define PREEMPT_LOCK_OFFSET 0
2745 #endif
2746
2747 #define cond_resched_lock(lock) ({ \
2748 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2749 __cond_resched_lock(lock); \
2750 })
2751
2752 extern int __cond_resched_softirq(void);
2753
2754 #define cond_resched_softirq() ({ \
2755 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2756 __cond_resched_softirq(); \
2757 })
2758
2759 static inline void cond_resched_rcu(void)
2760 {
2761 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
2762 rcu_read_unlock();
2763 cond_resched();
2764 rcu_read_lock();
2765 #endif
2766 }
2767
2768 /*
2769 * Does a critical section need to be broken due to another
2770 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2771 * but a general need for low latency)
2772 */
2773 static inline int spin_needbreak(spinlock_t *lock)
2774 {
2775 #ifdef CONFIG_PREEMPT
2776 return spin_is_contended(lock);
2777 #else
2778 return 0;
2779 #endif
2780 }
2781
2782 /*
2783 * Idle thread specific functions to determine the need_resched
2784 * polling state.
2785 */
2786 #ifdef TIF_POLLING_NRFLAG
2787 static inline int tsk_is_polling(struct task_struct *p)
2788 {
2789 return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
2790 }
2791
2792 static inline void __current_set_polling(void)
2793 {
2794 set_thread_flag(TIF_POLLING_NRFLAG);
2795 }
2796
2797 static inline bool __must_check current_set_polling_and_test(void)
2798 {
2799 __current_set_polling();
2800
2801 /*
2802 * Polling state must be visible before we test NEED_RESCHED,
2803 * paired by resched_task()
2804 */
2805 smp_mb__after_atomic();
2806
2807 return unlikely(tif_need_resched());
2808 }
2809
2810 static inline void __current_clr_polling(void)
2811 {
2812 clear_thread_flag(TIF_POLLING_NRFLAG);
2813 }
2814
2815 static inline bool __must_check current_clr_polling_and_test(void)
2816 {
2817 __current_clr_polling();
2818
2819 /*
2820 * Polling state must be visible before we test NEED_RESCHED,
2821 * paired by resched_task()
2822 */
2823 smp_mb__after_atomic();
2824
2825 return unlikely(tif_need_resched());
2826 }
2827
2828 #else
2829 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
2830 static inline void __current_set_polling(void) { }
2831 static inline void __current_clr_polling(void) { }
2832
2833 static inline bool __must_check current_set_polling_and_test(void)
2834 {
2835 return unlikely(tif_need_resched());
2836 }
2837 static inline bool __must_check current_clr_polling_and_test(void)
2838 {
2839 return unlikely(tif_need_resched());
2840 }
2841 #endif
2842
2843 static inline void current_clr_polling(void)
2844 {
2845 __current_clr_polling();
2846
2847 /*
2848 * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
2849 * Once the bit is cleared, we'll get IPIs with every new
2850 * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
2851 * fold.
2852 */
2853 smp_mb(); /* paired with resched_task() */
2854
2855 preempt_fold_need_resched();
2856 }
2857
2858 static __always_inline bool need_resched(void)
2859 {
2860 return unlikely(tif_need_resched());
2861 }
2862
2863 /*
2864 * Thread group CPU time accounting.
2865 */
2866 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2867 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2868
2869 static inline void thread_group_cputime_init(struct signal_struct *sig)
2870 {
2871 raw_spin_lock_init(&sig->cputimer.lock);
2872 }
2873
2874 /*
2875 * Reevaluate whether the task has signals pending delivery.
2876 * Wake the task if so.
2877 * This is required every time the blocked sigset_t changes.
2878 * callers must hold sighand->siglock.
2879 */
2880 extern void recalc_sigpending_and_wake(struct task_struct *t);
2881 extern void recalc_sigpending(void);
2882
2883 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
2884
2885 static inline void signal_wake_up(struct task_struct *t, bool resume)
2886 {
2887 signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
2888 }
2889 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
2890 {
2891 signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
2892 }
2893
2894 /*
2895 * Wrappers for p->thread_info->cpu access. No-op on UP.
2896 */
2897 #ifdef CONFIG_SMP
2898
2899 static inline unsigned int task_cpu(const struct task_struct *p)
2900 {
2901 return task_thread_info(p)->cpu;
2902 }
2903
2904 static inline int task_node(const struct task_struct *p)
2905 {
2906 return cpu_to_node(task_cpu(p));
2907 }
2908
2909 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2910
2911 #else
2912
2913 static inline unsigned int task_cpu(const struct task_struct *p)
2914 {
2915 return 0;
2916 }
2917
2918 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2919 {
2920 }
2921
2922 #endif /* CONFIG_SMP */
2923
2924 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2925 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2926
2927 #ifdef CONFIG_CGROUP_SCHED
2928 extern struct task_group root_task_group;
2929 #endif /* CONFIG_CGROUP_SCHED */
2930
2931 extern int task_can_switch_user(struct user_struct *up,
2932 struct task_struct *tsk);
2933
2934 #ifdef CONFIG_TASK_XACCT
2935 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2936 {
2937 tsk->ioac.rchar += amt;
2938 }
2939
2940 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2941 {
2942 tsk->ioac.wchar += amt;
2943 }
2944
2945 static inline void inc_syscr(struct task_struct *tsk)
2946 {
2947 tsk->ioac.syscr++;
2948 }
2949
2950 static inline void inc_syscw(struct task_struct *tsk)
2951 {
2952 tsk->ioac.syscw++;
2953 }
2954 #else
2955 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2956 {
2957 }
2958
2959 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2960 {
2961 }
2962
2963 static inline void inc_syscr(struct task_struct *tsk)
2964 {
2965 }
2966
2967 static inline void inc_syscw(struct task_struct *tsk)
2968 {
2969 }
2970 #endif
2971
2972 #ifndef TASK_SIZE_OF
2973 #define TASK_SIZE_OF(tsk) TASK_SIZE
2974 #endif
2975
2976 #ifdef CONFIG_MEMCG
2977 extern void mm_update_next_owner(struct mm_struct *mm);
2978 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2979 #else
2980 static inline void mm_update_next_owner(struct mm_struct *mm)
2981 {
2982 }
2983
2984 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2985 {
2986 }
2987 #endif /* CONFIG_MEMCG */
2988
2989 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2990 unsigned int limit)
2991 {
2992 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2993 }
2994
2995 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2996 unsigned int limit)
2997 {
2998 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2999 }
3000
3001 static inline unsigned long rlimit(unsigned int limit)
3002 {
3003 return task_rlimit(current, limit);
3004 }
3005
3006 static inline unsigned long rlimit_max(unsigned int limit)
3007 {
3008 return task_rlimit_max(current, limit);
3009 }
3010
3011 #endif