]> git.proxmox.com Git - mirror_ubuntu-artful-kernel.git/blob - include/linux/sched.h
block: remove per-queue plugging
[mirror_ubuntu-artful-kernel.git] / include / linux / sched.h
1 #ifndef _LINUX_SCHED_H
2 #define _LINUX_SCHED_H
3
4 /*
5 * cloning flags:
6 */
7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */
8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */
9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */
10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */
11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */
12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */
13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */
14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */
15 #define CLONE_THREAD 0x00010000 /* Same thread group? */
16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */
17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */
18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */
19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */
20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */
21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */
22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */
23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */
24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state)
25 and is now available for re-use. */
26 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
27 #define CLONE_NEWIPC 0x08000000 /* New ipcs */
28 #define CLONE_NEWUSER 0x10000000 /* New user namespace */
29 #define CLONE_NEWPID 0x20000000 /* New pid namespace */
30 #define CLONE_NEWNET 0x40000000 /* New network namespace */
31 #define CLONE_IO 0x80000000 /* Clone io context */
32
33 /*
34 * Scheduling policies
35 */
36 #define SCHED_NORMAL 0
37 #define SCHED_FIFO 1
38 #define SCHED_RR 2
39 #define SCHED_BATCH 3
40 /* SCHED_ISO: reserved but not implemented yet */
41 #define SCHED_IDLE 5
42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
43 #define SCHED_RESET_ON_FORK 0x40000000
44
45 #ifdef __KERNEL__
46
47 struct sched_param {
48 int sched_priority;
49 };
50
51 #include <asm/param.h> /* for HZ */
52
53 #include <linux/capability.h>
54 #include <linux/threads.h>
55 #include <linux/kernel.h>
56 #include <linux/types.h>
57 #include <linux/timex.h>
58 #include <linux/jiffies.h>
59 #include <linux/rbtree.h>
60 #include <linux/thread_info.h>
61 #include <linux/cpumask.h>
62 #include <linux/errno.h>
63 #include <linux/nodemask.h>
64 #include <linux/mm_types.h>
65
66 #include <asm/system.h>
67 #include <asm/page.h>
68 #include <asm/ptrace.h>
69 #include <asm/cputime.h>
70
71 #include <linux/smp.h>
72 #include <linux/sem.h>
73 #include <linux/signal.h>
74 #include <linux/compiler.h>
75 #include <linux/completion.h>
76 #include <linux/pid.h>
77 #include <linux/percpu.h>
78 #include <linux/topology.h>
79 #include <linux/proportions.h>
80 #include <linux/seccomp.h>
81 #include <linux/rcupdate.h>
82 #include <linux/rculist.h>
83 #include <linux/rtmutex.h>
84
85 #include <linux/time.h>
86 #include <linux/param.h>
87 #include <linux/resource.h>
88 #include <linux/timer.h>
89 #include <linux/hrtimer.h>
90 #include <linux/task_io_accounting.h>
91 #include <linux/latencytop.h>
92 #include <linux/cred.h>
93
94 #include <asm/processor.h>
95
96 struct exec_domain;
97 struct futex_pi_state;
98 struct robust_list_head;
99 struct bio_list;
100 struct fs_struct;
101 struct perf_event_context;
102 struct blk_plug;
103
104 /*
105 * List of flags we want to share for kernel threads,
106 * if only because they are not used by them anyway.
107 */
108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
109
110 /*
111 * These are the constant used to fake the fixed-point load-average
112 * counting. Some notes:
113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives
114 * a load-average precision of 10 bits integer + 11 bits fractional
115 * - if you want to count load-averages more often, you need more
116 * precision, or rounding will get you. With 2-second counting freq,
117 * the EXP_n values would be 1981, 2034 and 2043 if still using only
118 * 11 bit fractions.
119 */
120 extern unsigned long avenrun[]; /* Load averages */
121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
122
123 #define FSHIFT 11 /* nr of bits of precision */
124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
127 #define EXP_5 2014 /* 1/exp(5sec/5min) */
128 #define EXP_15 2037 /* 1/exp(5sec/15min) */
129
130 #define CALC_LOAD(load,exp,n) \
131 load *= exp; \
132 load += n*(FIXED_1-exp); \
133 load >>= FSHIFT;
134
135 extern unsigned long total_forks;
136 extern int nr_threads;
137 DECLARE_PER_CPU(unsigned long, process_counts);
138 extern int nr_processes(void);
139 extern unsigned long nr_running(void);
140 extern unsigned long nr_uninterruptible(void);
141 extern unsigned long nr_iowait(void);
142 extern unsigned long nr_iowait_cpu(int cpu);
143 extern unsigned long this_cpu_load(void);
144
145
146 extern void calc_global_load(unsigned long ticks);
147
148 extern unsigned long get_parent_ip(unsigned long addr);
149
150 struct seq_file;
151 struct cfs_rq;
152 struct task_group;
153 #ifdef CONFIG_SCHED_DEBUG
154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
155 extern void proc_sched_set_task(struct task_struct *p);
156 extern void
157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
158 #else
159 static inline void
160 proc_sched_show_task(struct task_struct *p, struct seq_file *m)
161 {
162 }
163 static inline void proc_sched_set_task(struct task_struct *p)
164 {
165 }
166 static inline void
167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
168 {
169 }
170 #endif
171
172 /*
173 * Task state bitmask. NOTE! These bits are also
174 * encoded in fs/proc/array.c: get_task_state().
175 *
176 * We have two separate sets of flags: task->state
177 * is about runnability, while task->exit_state are
178 * about the task exiting. Confusing, but this way
179 * modifying one set can't modify the other one by
180 * mistake.
181 */
182 #define TASK_RUNNING 0
183 #define TASK_INTERRUPTIBLE 1
184 #define TASK_UNINTERRUPTIBLE 2
185 #define __TASK_STOPPED 4
186 #define __TASK_TRACED 8
187 /* in tsk->exit_state */
188 #define EXIT_ZOMBIE 16
189 #define EXIT_DEAD 32
190 /* in tsk->state again */
191 #define TASK_DEAD 64
192 #define TASK_WAKEKILL 128
193 #define TASK_WAKING 256
194 #define TASK_STATE_MAX 512
195
196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
197
198 extern char ___assert_task_state[1 - 2*!!(
199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
200
201 /* Convenience macros for the sake of set_task_state */
202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
205
206 /* Convenience macros for the sake of wake_up */
207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
209
210 /* get_task_state() */
211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
213 __TASK_TRACED)
214
215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0)
217 #define task_is_dead(task) ((task)->exit_state != 0)
218 #define task_is_stopped_or_traced(task) \
219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
220 #define task_contributes_to_load(task) \
221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
222 (task->flags & PF_FREEZING) == 0)
223
224 #define __set_task_state(tsk, state_value) \
225 do { (tsk)->state = (state_value); } while (0)
226 #define set_task_state(tsk, state_value) \
227 set_mb((tsk)->state, (state_value))
228
229 /*
230 * set_current_state() includes a barrier so that the write of current->state
231 * is correctly serialised wrt the caller's subsequent test of whether to
232 * actually sleep:
233 *
234 * set_current_state(TASK_UNINTERRUPTIBLE);
235 * if (do_i_need_to_sleep())
236 * schedule();
237 *
238 * If the caller does not need such serialisation then use __set_current_state()
239 */
240 #define __set_current_state(state_value) \
241 do { current->state = (state_value); } while (0)
242 #define set_current_state(state_value) \
243 set_mb(current->state, (state_value))
244
245 /* Task command name length */
246 #define TASK_COMM_LEN 16
247
248 #include <linux/spinlock.h>
249
250 /*
251 * This serializes "schedule()" and also protects
252 * the run-queue from deletions/modifications (but
253 * _adding_ to the beginning of the run-queue has
254 * a separate lock).
255 */
256 extern rwlock_t tasklist_lock;
257 extern spinlock_t mmlist_lock;
258
259 struct task_struct;
260
261 #ifdef CONFIG_PROVE_RCU
262 extern int lockdep_tasklist_lock_is_held(void);
263 #endif /* #ifdef CONFIG_PROVE_RCU */
264
265 extern void sched_init(void);
266 extern void sched_init_smp(void);
267 extern asmlinkage void schedule_tail(struct task_struct *prev);
268 extern void init_idle(struct task_struct *idle, int cpu);
269 extern void init_idle_bootup_task(struct task_struct *idle);
270
271 extern int runqueue_is_locked(int cpu);
272
273 extern cpumask_var_t nohz_cpu_mask;
274 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
275 extern void select_nohz_load_balancer(int stop_tick);
276 extern int get_nohz_timer_target(void);
277 #else
278 static inline void select_nohz_load_balancer(int stop_tick) { }
279 #endif
280
281 /*
282 * Only dump TASK_* tasks. (0 for all tasks)
283 */
284 extern void show_state_filter(unsigned long state_filter);
285
286 static inline void show_state(void)
287 {
288 show_state_filter(0);
289 }
290
291 extern void show_regs(struct pt_regs *);
292
293 /*
294 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
295 * task), SP is the stack pointer of the first frame that should be shown in the back
296 * trace (or NULL if the entire call-chain of the task should be shown).
297 */
298 extern void show_stack(struct task_struct *task, unsigned long *sp);
299
300 void io_schedule(void);
301 long io_schedule_timeout(long timeout);
302
303 extern void cpu_init (void);
304 extern void trap_init(void);
305 extern void update_process_times(int user);
306 extern void scheduler_tick(void);
307
308 extern void sched_show_task(struct task_struct *p);
309
310 #ifdef CONFIG_LOCKUP_DETECTOR
311 extern void touch_softlockup_watchdog(void);
312 extern void touch_softlockup_watchdog_sync(void);
313 extern void touch_all_softlockup_watchdogs(void);
314 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
315 void __user *buffer,
316 size_t *lenp, loff_t *ppos);
317 extern unsigned int softlockup_panic;
318 extern int softlockup_thresh;
319 void lockup_detector_init(void);
320 #else
321 static inline void touch_softlockup_watchdog(void)
322 {
323 }
324 static inline void touch_softlockup_watchdog_sync(void)
325 {
326 }
327 static inline void touch_all_softlockup_watchdogs(void)
328 {
329 }
330 static inline void lockup_detector_init(void)
331 {
332 }
333 #endif
334
335 #ifdef CONFIG_DETECT_HUNG_TASK
336 extern unsigned int sysctl_hung_task_panic;
337 extern unsigned long sysctl_hung_task_check_count;
338 extern unsigned long sysctl_hung_task_timeout_secs;
339 extern unsigned long sysctl_hung_task_warnings;
340 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
341 void __user *buffer,
342 size_t *lenp, loff_t *ppos);
343 #else
344 /* Avoid need for ifdefs elsewhere in the code */
345 enum { sysctl_hung_task_timeout_secs = 0 };
346 #endif
347
348 /* Attach to any functions which should be ignored in wchan output. */
349 #define __sched __attribute__((__section__(".sched.text")))
350
351 /* Linker adds these: start and end of __sched functions */
352 extern char __sched_text_start[], __sched_text_end[];
353
354 /* Is this address in the __sched functions? */
355 extern int in_sched_functions(unsigned long addr);
356
357 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
358 extern signed long schedule_timeout(signed long timeout);
359 extern signed long schedule_timeout_interruptible(signed long timeout);
360 extern signed long schedule_timeout_killable(signed long timeout);
361 extern signed long schedule_timeout_uninterruptible(signed long timeout);
362 asmlinkage void schedule(void);
363 extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
364
365 struct nsproxy;
366 struct user_namespace;
367
368 /*
369 * Default maximum number of active map areas, this limits the number of vmas
370 * per mm struct. Users can overwrite this number by sysctl but there is a
371 * problem.
372 *
373 * When a program's coredump is generated as ELF format, a section is created
374 * per a vma. In ELF, the number of sections is represented in unsigned short.
375 * This means the number of sections should be smaller than 65535 at coredump.
376 * Because the kernel adds some informative sections to a image of program at
377 * generating coredump, we need some margin. The number of extra sections is
378 * 1-3 now and depends on arch. We use "5" as safe margin, here.
379 */
380 #define MAPCOUNT_ELF_CORE_MARGIN (5)
381 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
382
383 extern int sysctl_max_map_count;
384
385 #include <linux/aio.h>
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 extern void arch_unmap_area(struct mm_struct *, unsigned long);
397 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
398 #else
399 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
400 #endif
401
402
403 extern void set_dumpable(struct mm_struct *mm, int value);
404 extern int get_dumpable(struct mm_struct *mm);
405
406 /* mm flags */
407 /* dumpable bits */
408 #define MMF_DUMPABLE 0 /* core dump is permitted */
409 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
410
411 #define MMF_DUMPABLE_BITS 2
412 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
413
414 /* coredump filter bits */
415 #define MMF_DUMP_ANON_PRIVATE 2
416 #define MMF_DUMP_ANON_SHARED 3
417 #define MMF_DUMP_MAPPED_PRIVATE 4
418 #define MMF_DUMP_MAPPED_SHARED 5
419 #define MMF_DUMP_ELF_HEADERS 6
420 #define MMF_DUMP_HUGETLB_PRIVATE 7
421 #define MMF_DUMP_HUGETLB_SHARED 8
422
423 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
424 #define MMF_DUMP_FILTER_BITS 7
425 #define MMF_DUMP_FILTER_MASK \
426 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
427 #define MMF_DUMP_FILTER_DEFAULT \
428 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
429 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
430
431 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
432 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
433 #else
434 # define MMF_DUMP_MASK_DEFAULT_ELF 0
435 #endif
436 /* leave room for more dump flags */
437 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
438 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
439
440 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
441
442 struct sighand_struct {
443 atomic_t count;
444 struct k_sigaction action[_NSIG];
445 spinlock_t siglock;
446 wait_queue_head_t signalfd_wqh;
447 };
448
449 struct pacct_struct {
450 int ac_flag;
451 long ac_exitcode;
452 unsigned long ac_mem;
453 cputime_t ac_utime, ac_stime;
454 unsigned long ac_minflt, ac_majflt;
455 };
456
457 struct cpu_itimer {
458 cputime_t expires;
459 cputime_t incr;
460 u32 error;
461 u32 incr_error;
462 };
463
464 /**
465 * struct task_cputime - collected CPU time counts
466 * @utime: time spent in user mode, in &cputime_t units
467 * @stime: time spent in kernel mode, in &cputime_t units
468 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
469 *
470 * This structure groups together three kinds of CPU time that are
471 * tracked for threads and thread groups. Most things considering
472 * CPU time want to group these counts together and treat all three
473 * of them in parallel.
474 */
475 struct task_cputime {
476 cputime_t utime;
477 cputime_t stime;
478 unsigned long long sum_exec_runtime;
479 };
480 /* Alternate field names when used to cache expirations. */
481 #define prof_exp stime
482 #define virt_exp utime
483 #define sched_exp sum_exec_runtime
484
485 #define INIT_CPUTIME \
486 (struct task_cputime) { \
487 .utime = cputime_zero, \
488 .stime = cputime_zero, \
489 .sum_exec_runtime = 0, \
490 }
491
492 /*
493 * Disable preemption until the scheduler is running.
494 * Reset by start_kernel()->sched_init()->init_idle().
495 *
496 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
497 * before the scheduler is active -- see should_resched().
498 */
499 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
500
501 /**
502 * struct thread_group_cputimer - thread group interval timer counts
503 * @cputime: thread group interval timers.
504 * @running: non-zero when there are timers running and
505 * @cputime receives updates.
506 * @lock: lock for fields in this struct.
507 *
508 * This structure contains the version of task_cputime, above, that is
509 * used for thread group CPU timer calculations.
510 */
511 struct thread_group_cputimer {
512 struct task_cputime cputime;
513 int running;
514 spinlock_t lock;
515 };
516
517 struct autogroup;
518
519 /*
520 * NOTE! "signal_struct" does not have it's own
521 * locking, because a shared signal_struct always
522 * implies a shared sighand_struct, so locking
523 * sighand_struct is always a proper superset of
524 * the locking of signal_struct.
525 */
526 struct signal_struct {
527 atomic_t sigcnt;
528 atomic_t live;
529 int nr_threads;
530
531 wait_queue_head_t wait_chldexit; /* for wait4() */
532
533 /* current thread group signal load-balancing target: */
534 struct task_struct *curr_target;
535
536 /* shared signal handling: */
537 struct sigpending shared_pending;
538
539 /* thread group exit support */
540 int group_exit_code;
541 /* overloaded:
542 * - notify group_exit_task when ->count is equal to notify_count
543 * - everyone except group_exit_task is stopped during signal delivery
544 * of fatal signals, group_exit_task processes the signal.
545 */
546 int notify_count;
547 struct task_struct *group_exit_task;
548
549 /* thread group stop support, overloads group_exit_code too */
550 int group_stop_count;
551 unsigned int flags; /* see SIGNAL_* flags below */
552
553 /* POSIX.1b Interval Timers */
554 struct list_head posix_timers;
555
556 /* ITIMER_REAL timer for the process */
557 struct hrtimer real_timer;
558 struct pid *leader_pid;
559 ktime_t it_real_incr;
560
561 /*
562 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
563 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
564 * values are defined to 0 and 1 respectively
565 */
566 struct cpu_itimer it[2];
567
568 /*
569 * Thread group totals for process CPU timers.
570 * See thread_group_cputimer(), et al, for details.
571 */
572 struct thread_group_cputimer cputimer;
573
574 /* Earliest-expiration cache. */
575 struct task_cputime cputime_expires;
576
577 struct list_head cpu_timers[3];
578
579 struct pid *tty_old_pgrp;
580
581 /* boolean value for session group leader */
582 int leader;
583
584 struct tty_struct *tty; /* NULL if no tty */
585
586 #ifdef CONFIG_SCHED_AUTOGROUP
587 struct autogroup *autogroup;
588 #endif
589 /*
590 * Cumulative resource counters for dead threads in the group,
591 * and for reaped dead child processes forked by this group.
592 * Live threads maintain their own counters and add to these
593 * in __exit_signal, except for the group leader.
594 */
595 cputime_t utime, stime, cutime, cstime;
596 cputime_t gtime;
597 cputime_t cgtime;
598 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
599 cputime_t prev_utime, prev_stime;
600 #endif
601 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
602 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
603 unsigned long inblock, oublock, cinblock, coublock;
604 unsigned long maxrss, cmaxrss;
605 struct task_io_accounting ioac;
606
607 /*
608 * Cumulative ns of schedule CPU time fo dead threads in the
609 * group, not including a zombie group leader, (This only differs
610 * from jiffies_to_ns(utime + stime) if sched_clock uses something
611 * other than jiffies.)
612 */
613 unsigned long long sum_sched_runtime;
614
615 /*
616 * We don't bother to synchronize most readers of this at all,
617 * because there is no reader checking a limit that actually needs
618 * to get both rlim_cur and rlim_max atomically, and either one
619 * alone is a single word that can safely be read normally.
620 * getrlimit/setrlimit use task_lock(current->group_leader) to
621 * protect this instead of the siglock, because they really
622 * have no need to disable irqs.
623 */
624 struct rlimit rlim[RLIM_NLIMITS];
625
626 #ifdef CONFIG_BSD_PROCESS_ACCT
627 struct pacct_struct pacct; /* per-process accounting information */
628 #endif
629 #ifdef CONFIG_TASKSTATS
630 struct taskstats *stats;
631 #endif
632 #ifdef CONFIG_AUDIT
633 unsigned audit_tty;
634 struct tty_audit_buf *tty_audit_buf;
635 #endif
636
637 int oom_adj; /* OOM kill score adjustment (bit shift) */
638 int oom_score_adj; /* OOM kill score adjustment */
639 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
640 * Only settable by CAP_SYS_RESOURCE. */
641
642 struct mutex cred_guard_mutex; /* guard against foreign influences on
643 * credential calculations
644 * (notably. ptrace) */
645 };
646
647 /* Context switch must be unlocked if interrupts are to be enabled */
648 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
649 # define __ARCH_WANT_UNLOCKED_CTXSW
650 #endif
651
652 /*
653 * Bits in flags field of signal_struct.
654 */
655 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
656 #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
657 #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
658 #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
659 /*
660 * Pending notifications to parent.
661 */
662 #define SIGNAL_CLD_STOPPED 0x00000010
663 #define SIGNAL_CLD_CONTINUED 0x00000020
664 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
665
666 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
667
668 /* If true, all threads except ->group_exit_task have pending SIGKILL */
669 static inline int signal_group_exit(const struct signal_struct *sig)
670 {
671 return (sig->flags & SIGNAL_GROUP_EXIT) ||
672 (sig->group_exit_task != NULL);
673 }
674
675 /*
676 * Some day this will be a full-fledged user tracking system..
677 */
678 struct user_struct {
679 atomic_t __count; /* reference count */
680 atomic_t processes; /* How many processes does this user have? */
681 atomic_t files; /* How many open files does this user have? */
682 atomic_t sigpending; /* How many pending signals does this user have? */
683 #ifdef CONFIG_INOTIFY_USER
684 atomic_t inotify_watches; /* How many inotify watches does this user have? */
685 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
686 #endif
687 #ifdef CONFIG_FANOTIFY
688 atomic_t fanotify_listeners;
689 #endif
690 #ifdef CONFIG_EPOLL
691 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
692 #endif
693 #ifdef CONFIG_POSIX_MQUEUE
694 /* protected by mq_lock */
695 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
696 #endif
697 unsigned long locked_shm; /* How many pages of mlocked shm ? */
698
699 #ifdef CONFIG_KEYS
700 struct key *uid_keyring; /* UID specific keyring */
701 struct key *session_keyring; /* UID's default session keyring */
702 #endif
703
704 /* Hash table maintenance information */
705 struct hlist_node uidhash_node;
706 uid_t uid;
707 struct user_namespace *user_ns;
708
709 #ifdef CONFIG_PERF_EVENTS
710 atomic_long_t locked_vm;
711 #endif
712 };
713
714 extern int uids_sysfs_init(void);
715
716 extern struct user_struct *find_user(uid_t);
717
718 extern struct user_struct root_user;
719 #define INIT_USER (&root_user)
720
721
722 struct backing_dev_info;
723 struct reclaim_state;
724
725 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
726 struct sched_info {
727 /* cumulative counters */
728 unsigned long pcount; /* # of times run on this cpu */
729 unsigned long long run_delay; /* time spent waiting on a runqueue */
730
731 /* timestamps */
732 unsigned long long last_arrival,/* when we last ran on a cpu */
733 last_queued; /* when we were last queued to run */
734 #ifdef CONFIG_SCHEDSTATS
735 /* BKL stats */
736 unsigned int bkl_count;
737 #endif
738 };
739 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
740
741 #ifdef CONFIG_TASK_DELAY_ACCT
742 struct task_delay_info {
743 spinlock_t lock;
744 unsigned int flags; /* Private per-task flags */
745
746 /* For each stat XXX, add following, aligned appropriately
747 *
748 * struct timespec XXX_start, XXX_end;
749 * u64 XXX_delay;
750 * u32 XXX_count;
751 *
752 * Atomicity of updates to XXX_delay, XXX_count protected by
753 * single lock above (split into XXX_lock if contention is an issue).
754 */
755
756 /*
757 * XXX_count is incremented on every XXX operation, the delay
758 * associated with the operation is added to XXX_delay.
759 * XXX_delay contains the accumulated delay time in nanoseconds.
760 */
761 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
762 u64 blkio_delay; /* wait for sync block io completion */
763 u64 swapin_delay; /* wait for swapin block io completion */
764 u32 blkio_count; /* total count of the number of sync block */
765 /* io operations performed */
766 u32 swapin_count; /* total count of the number of swapin block */
767 /* io operations performed */
768
769 struct timespec freepages_start, freepages_end;
770 u64 freepages_delay; /* wait for memory reclaim */
771 u32 freepages_count; /* total count of memory reclaim */
772 };
773 #endif /* CONFIG_TASK_DELAY_ACCT */
774
775 static inline int sched_info_on(void)
776 {
777 #ifdef CONFIG_SCHEDSTATS
778 return 1;
779 #elif defined(CONFIG_TASK_DELAY_ACCT)
780 extern int delayacct_on;
781 return delayacct_on;
782 #else
783 return 0;
784 #endif
785 }
786
787 enum cpu_idle_type {
788 CPU_IDLE,
789 CPU_NOT_IDLE,
790 CPU_NEWLY_IDLE,
791 CPU_MAX_IDLE_TYPES
792 };
793
794 /*
795 * sched-domains (multiprocessor balancing) declarations:
796 */
797
798 /*
799 * Increase resolution of nice-level calculations:
800 */
801 #define SCHED_LOAD_SHIFT 10
802 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
803
804 #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
805
806 #ifdef CONFIG_SMP
807 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
808 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
809 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
810 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
811 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
812 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
813 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
814 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
815 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
816 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
817 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
818 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
819 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
820
821 enum powersavings_balance_level {
822 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
823 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
824 * first for long running threads
825 */
826 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
827 * cpu package for power savings
828 */
829 MAX_POWERSAVINGS_BALANCE_LEVELS
830 };
831
832 extern int sched_mc_power_savings, sched_smt_power_savings;
833
834 static inline int sd_balance_for_mc_power(void)
835 {
836 if (sched_smt_power_savings)
837 return SD_POWERSAVINGS_BALANCE;
838
839 if (!sched_mc_power_savings)
840 return SD_PREFER_SIBLING;
841
842 return 0;
843 }
844
845 static inline int sd_balance_for_package_power(void)
846 {
847 if (sched_mc_power_savings | sched_smt_power_savings)
848 return SD_POWERSAVINGS_BALANCE;
849
850 return SD_PREFER_SIBLING;
851 }
852
853 extern int __weak arch_sd_sibiling_asym_packing(void);
854
855 /*
856 * Optimise SD flags for power savings:
857 * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
858 * Keep default SD flags if sched_{smt,mc}_power_saving=0
859 */
860
861 static inline int sd_power_saving_flags(void)
862 {
863 if (sched_mc_power_savings | sched_smt_power_savings)
864 return SD_BALANCE_NEWIDLE;
865
866 return 0;
867 }
868
869 struct sched_group {
870 struct sched_group *next; /* Must be a circular list */
871
872 /*
873 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
874 * single CPU.
875 */
876 unsigned int cpu_power, cpu_power_orig;
877 unsigned int group_weight;
878
879 /*
880 * The CPUs this group covers.
881 *
882 * NOTE: this field is variable length. (Allocated dynamically
883 * by attaching extra space to the end of the structure,
884 * depending on how many CPUs the kernel has booted up with)
885 *
886 * It is also be embedded into static data structures at build
887 * time. (See 'struct static_sched_group' in kernel/sched.c)
888 */
889 unsigned long cpumask[0];
890 };
891
892 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
893 {
894 return to_cpumask(sg->cpumask);
895 }
896
897 enum sched_domain_level {
898 SD_LV_NONE = 0,
899 SD_LV_SIBLING,
900 SD_LV_MC,
901 SD_LV_BOOK,
902 SD_LV_CPU,
903 SD_LV_NODE,
904 SD_LV_ALLNODES,
905 SD_LV_MAX
906 };
907
908 struct sched_domain_attr {
909 int relax_domain_level;
910 };
911
912 #define SD_ATTR_INIT (struct sched_domain_attr) { \
913 .relax_domain_level = -1, \
914 }
915
916 struct sched_domain {
917 /* These fields must be setup */
918 struct sched_domain *parent; /* top domain must be null terminated */
919 struct sched_domain *child; /* bottom domain must be null terminated */
920 struct sched_group *groups; /* the balancing groups of the domain */
921 unsigned long min_interval; /* Minimum balance interval ms */
922 unsigned long max_interval; /* Maximum balance interval ms */
923 unsigned int busy_factor; /* less balancing by factor if busy */
924 unsigned int imbalance_pct; /* No balance until over watermark */
925 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
926 unsigned int busy_idx;
927 unsigned int idle_idx;
928 unsigned int newidle_idx;
929 unsigned int wake_idx;
930 unsigned int forkexec_idx;
931 unsigned int smt_gain;
932 int flags; /* See SD_* */
933 enum sched_domain_level level;
934
935 /* Runtime fields. */
936 unsigned long last_balance; /* init to jiffies. units in jiffies */
937 unsigned int balance_interval; /* initialise to 1. units in ms. */
938 unsigned int nr_balance_failed; /* initialise to 0 */
939
940 u64 last_update;
941
942 #ifdef CONFIG_SCHEDSTATS
943 /* load_balance() stats */
944 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
945 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
946 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
947 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
948 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
949 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
952
953 /* Active load balancing */
954 unsigned int alb_count;
955 unsigned int alb_failed;
956 unsigned int alb_pushed;
957
958 /* SD_BALANCE_EXEC stats */
959 unsigned int sbe_count;
960 unsigned int sbe_balanced;
961 unsigned int sbe_pushed;
962
963 /* SD_BALANCE_FORK stats */
964 unsigned int sbf_count;
965 unsigned int sbf_balanced;
966 unsigned int sbf_pushed;
967
968 /* try_to_wake_up() stats */
969 unsigned int ttwu_wake_remote;
970 unsigned int ttwu_move_affine;
971 unsigned int ttwu_move_balance;
972 #endif
973 #ifdef CONFIG_SCHED_DEBUG
974 char *name;
975 #endif
976
977 unsigned int span_weight;
978 /*
979 * Span of all CPUs in this domain.
980 *
981 * NOTE: this field is variable length. (Allocated dynamically
982 * by attaching extra space to the end of the structure,
983 * depending on how many CPUs the kernel has booted up with)
984 *
985 * It is also be embedded into static data structures at build
986 * time. (See 'struct static_sched_domain' in kernel/sched.c)
987 */
988 unsigned long span[0];
989 };
990
991 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
992 {
993 return to_cpumask(sd->span);
994 }
995
996 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
997 struct sched_domain_attr *dattr_new);
998
999 /* Allocate an array of sched domains, for partition_sched_domains(). */
1000 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1001 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1002
1003 /* Test a flag in parent sched domain */
1004 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1005 {
1006 if (sd->parent && (sd->parent->flags & flag))
1007 return 1;
1008
1009 return 0;
1010 }
1011
1012 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1013 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1014
1015 #else /* CONFIG_SMP */
1016
1017 struct sched_domain_attr;
1018
1019 static inline void
1020 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1021 struct sched_domain_attr *dattr_new)
1022 {
1023 }
1024 #endif /* !CONFIG_SMP */
1025
1026
1027 struct io_context; /* See blkdev.h */
1028
1029
1030 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1031 extern void prefetch_stack(struct task_struct *t);
1032 #else
1033 static inline void prefetch_stack(struct task_struct *t) { }
1034 #endif
1035
1036 struct audit_context; /* See audit.c */
1037 struct mempolicy;
1038 struct pipe_inode_info;
1039 struct uts_namespace;
1040
1041 struct rq;
1042 struct sched_domain;
1043
1044 /*
1045 * wake flags
1046 */
1047 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1048 #define WF_FORK 0x02 /* child wakeup after fork */
1049
1050 #define ENQUEUE_WAKEUP 1
1051 #define ENQUEUE_WAKING 2
1052 #define ENQUEUE_HEAD 4
1053
1054 #define DEQUEUE_SLEEP 1
1055
1056 struct sched_class {
1057 const struct sched_class *next;
1058
1059 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1060 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1061 void (*yield_task) (struct rq *rq);
1062
1063 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1064
1065 struct task_struct * (*pick_next_task) (struct rq *rq);
1066 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1067
1068 #ifdef CONFIG_SMP
1069 int (*select_task_rq)(struct rq *rq, struct task_struct *p,
1070 int sd_flag, int flags);
1071
1072 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1073 void (*post_schedule) (struct rq *this_rq);
1074 void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1075 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1076
1077 void (*set_cpus_allowed)(struct task_struct *p,
1078 const struct cpumask *newmask);
1079
1080 void (*rq_online)(struct rq *rq);
1081 void (*rq_offline)(struct rq *rq);
1082 #endif
1083
1084 void (*set_curr_task) (struct rq *rq);
1085 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1086 void (*task_fork) (struct task_struct *p);
1087
1088 void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1089 int running);
1090 void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1091 int running);
1092 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1093 int oldprio, int running);
1094
1095 unsigned int (*get_rr_interval) (struct rq *rq,
1096 struct task_struct *task);
1097
1098 #ifdef CONFIG_FAIR_GROUP_SCHED
1099 void (*task_move_group) (struct task_struct *p, int on_rq);
1100 #endif
1101 };
1102
1103 struct load_weight {
1104 unsigned long weight, inv_weight;
1105 };
1106
1107 #ifdef CONFIG_SCHEDSTATS
1108 struct sched_statistics {
1109 u64 wait_start;
1110 u64 wait_max;
1111 u64 wait_count;
1112 u64 wait_sum;
1113 u64 iowait_count;
1114 u64 iowait_sum;
1115
1116 u64 sleep_start;
1117 u64 sleep_max;
1118 s64 sum_sleep_runtime;
1119
1120 u64 block_start;
1121 u64 block_max;
1122 u64 exec_max;
1123 u64 slice_max;
1124
1125 u64 nr_migrations_cold;
1126 u64 nr_failed_migrations_affine;
1127 u64 nr_failed_migrations_running;
1128 u64 nr_failed_migrations_hot;
1129 u64 nr_forced_migrations;
1130
1131 u64 nr_wakeups;
1132 u64 nr_wakeups_sync;
1133 u64 nr_wakeups_migrate;
1134 u64 nr_wakeups_local;
1135 u64 nr_wakeups_remote;
1136 u64 nr_wakeups_affine;
1137 u64 nr_wakeups_affine_attempts;
1138 u64 nr_wakeups_passive;
1139 u64 nr_wakeups_idle;
1140 };
1141 #endif
1142
1143 struct sched_entity {
1144 struct load_weight load; /* for load-balancing */
1145 struct rb_node run_node;
1146 struct list_head group_node;
1147 unsigned int on_rq;
1148
1149 u64 exec_start;
1150 u64 sum_exec_runtime;
1151 u64 vruntime;
1152 u64 prev_sum_exec_runtime;
1153
1154 u64 nr_migrations;
1155
1156 #ifdef CONFIG_SCHEDSTATS
1157 struct sched_statistics statistics;
1158 #endif
1159
1160 #ifdef CONFIG_FAIR_GROUP_SCHED
1161 struct sched_entity *parent;
1162 /* rq on which this entity is (to be) queued: */
1163 struct cfs_rq *cfs_rq;
1164 /* rq "owned" by this entity/group: */
1165 struct cfs_rq *my_q;
1166 #endif
1167 };
1168
1169 struct sched_rt_entity {
1170 struct list_head run_list;
1171 unsigned long timeout;
1172 unsigned int time_slice;
1173 int nr_cpus_allowed;
1174
1175 struct sched_rt_entity *back;
1176 #ifdef CONFIG_RT_GROUP_SCHED
1177 struct sched_rt_entity *parent;
1178 /* rq on which this entity is (to be) queued: */
1179 struct rt_rq *rt_rq;
1180 /* rq "owned" by this entity/group: */
1181 struct rt_rq *my_q;
1182 #endif
1183 };
1184
1185 struct rcu_node;
1186
1187 enum perf_event_task_context {
1188 perf_invalid_context = -1,
1189 perf_hw_context = 0,
1190 perf_sw_context,
1191 perf_nr_task_contexts,
1192 };
1193
1194 struct task_struct {
1195 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1196 void *stack;
1197 atomic_t usage;
1198 unsigned int flags; /* per process flags, defined below */
1199 unsigned int ptrace;
1200
1201 int lock_depth; /* BKL lock depth */
1202
1203 #ifdef CONFIG_SMP
1204 #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1205 int oncpu;
1206 #endif
1207 #endif
1208
1209 int prio, static_prio, normal_prio;
1210 unsigned int rt_priority;
1211 const struct sched_class *sched_class;
1212 struct sched_entity se;
1213 struct sched_rt_entity rt;
1214
1215 #ifdef CONFIG_PREEMPT_NOTIFIERS
1216 /* list of struct preempt_notifier: */
1217 struct hlist_head preempt_notifiers;
1218 #endif
1219
1220 /*
1221 * fpu_counter contains the number of consecutive context switches
1222 * that the FPU is used. If this is over a threshold, the lazy fpu
1223 * saving becomes unlazy to save the trap. This is an unsigned char
1224 * so that after 256 times the counter wraps and the behavior turns
1225 * lazy again; this to deal with bursty apps that only use FPU for
1226 * a short time
1227 */
1228 unsigned char fpu_counter;
1229 #ifdef CONFIG_BLK_DEV_IO_TRACE
1230 unsigned int btrace_seq;
1231 #endif
1232
1233 unsigned int policy;
1234 cpumask_t cpus_allowed;
1235
1236 #ifdef CONFIG_PREEMPT_RCU
1237 int rcu_read_lock_nesting;
1238 char rcu_read_unlock_special;
1239 struct list_head rcu_node_entry;
1240 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1241 #ifdef CONFIG_TREE_PREEMPT_RCU
1242 struct rcu_node *rcu_blocked_node;
1243 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1244 #ifdef CONFIG_RCU_BOOST
1245 struct rt_mutex *rcu_boost_mutex;
1246 #endif /* #ifdef CONFIG_RCU_BOOST */
1247
1248 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1249 struct sched_info sched_info;
1250 #endif
1251
1252 struct list_head tasks;
1253 #ifdef CONFIG_SMP
1254 struct plist_node pushable_tasks;
1255 #endif
1256
1257 struct mm_struct *mm, *active_mm;
1258 #if defined(SPLIT_RSS_COUNTING)
1259 struct task_rss_stat rss_stat;
1260 #endif
1261 /* task state */
1262 int exit_state;
1263 int exit_code, exit_signal;
1264 int pdeath_signal; /* The signal sent when the parent dies */
1265 /* ??? */
1266 unsigned int personality;
1267 unsigned did_exec:1;
1268 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1269 * execve */
1270 unsigned in_iowait:1;
1271
1272
1273 /* Revert to default priority/policy when forking */
1274 unsigned sched_reset_on_fork:1;
1275
1276 pid_t pid;
1277 pid_t tgid;
1278
1279 #ifdef CONFIG_CC_STACKPROTECTOR
1280 /* Canary value for the -fstack-protector gcc feature */
1281 unsigned long stack_canary;
1282 #endif
1283
1284 /*
1285 * pointers to (original) parent process, youngest child, younger sibling,
1286 * older sibling, respectively. (p->father can be replaced with
1287 * p->real_parent->pid)
1288 */
1289 struct task_struct *real_parent; /* real parent process */
1290 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1291 /*
1292 * children/sibling forms the list of my natural children
1293 */
1294 struct list_head children; /* list of my children */
1295 struct list_head sibling; /* linkage in my parent's children list */
1296 struct task_struct *group_leader; /* threadgroup leader */
1297
1298 /*
1299 * ptraced is the list of tasks this task is using ptrace on.
1300 * This includes both natural children and PTRACE_ATTACH targets.
1301 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1302 */
1303 struct list_head ptraced;
1304 struct list_head ptrace_entry;
1305
1306 /* PID/PID hash table linkage. */
1307 struct pid_link pids[PIDTYPE_MAX];
1308 struct list_head thread_group;
1309
1310 struct completion *vfork_done; /* for vfork() */
1311 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1312 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1313
1314 cputime_t utime, stime, utimescaled, stimescaled;
1315 cputime_t gtime;
1316 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1317 cputime_t prev_utime, prev_stime;
1318 #endif
1319 unsigned long nvcsw, nivcsw; /* context switch counts */
1320 struct timespec start_time; /* monotonic time */
1321 struct timespec real_start_time; /* boot based time */
1322 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1323 unsigned long min_flt, maj_flt;
1324
1325 struct task_cputime cputime_expires;
1326 struct list_head cpu_timers[3];
1327
1328 /* process credentials */
1329 const struct cred __rcu *real_cred; /* objective and real subjective task
1330 * credentials (COW) */
1331 const struct cred __rcu *cred; /* effective (overridable) subjective task
1332 * credentials (COW) */
1333 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1334
1335 char comm[TASK_COMM_LEN]; /* executable name excluding path
1336 - access with [gs]et_task_comm (which lock
1337 it with task_lock())
1338 - initialized normally by setup_new_exec */
1339 /* file system info */
1340 int link_count, total_link_count;
1341 #ifdef CONFIG_SYSVIPC
1342 /* ipc stuff */
1343 struct sysv_sem sysvsem;
1344 #endif
1345 #ifdef CONFIG_DETECT_HUNG_TASK
1346 /* hung task detection */
1347 unsigned long last_switch_count;
1348 #endif
1349 /* CPU-specific state of this task */
1350 struct thread_struct thread;
1351 /* filesystem information */
1352 struct fs_struct *fs;
1353 /* open file information */
1354 struct files_struct *files;
1355 /* namespaces */
1356 struct nsproxy *nsproxy;
1357 /* signal handlers */
1358 struct signal_struct *signal;
1359 struct sighand_struct *sighand;
1360
1361 sigset_t blocked, real_blocked;
1362 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1363 struct sigpending pending;
1364
1365 unsigned long sas_ss_sp;
1366 size_t sas_ss_size;
1367 int (*notifier)(void *priv);
1368 void *notifier_data;
1369 sigset_t *notifier_mask;
1370 struct audit_context *audit_context;
1371 #ifdef CONFIG_AUDITSYSCALL
1372 uid_t loginuid;
1373 unsigned int sessionid;
1374 #endif
1375 seccomp_t seccomp;
1376
1377 /* Thread group tracking */
1378 u32 parent_exec_id;
1379 u32 self_exec_id;
1380 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1381 * mempolicy */
1382 spinlock_t alloc_lock;
1383
1384 #ifdef CONFIG_GENERIC_HARDIRQS
1385 /* IRQ handler threads */
1386 struct irqaction *irqaction;
1387 #endif
1388
1389 /* Protection of the PI data structures: */
1390 raw_spinlock_t pi_lock;
1391
1392 #ifdef CONFIG_RT_MUTEXES
1393 /* PI waiters blocked on a rt_mutex held by this task */
1394 struct plist_head pi_waiters;
1395 /* Deadlock detection and priority inheritance handling */
1396 struct rt_mutex_waiter *pi_blocked_on;
1397 #endif
1398
1399 #ifdef CONFIG_DEBUG_MUTEXES
1400 /* mutex deadlock detection */
1401 struct mutex_waiter *blocked_on;
1402 #endif
1403 #ifdef CONFIG_TRACE_IRQFLAGS
1404 unsigned int irq_events;
1405 unsigned long hardirq_enable_ip;
1406 unsigned long hardirq_disable_ip;
1407 unsigned int hardirq_enable_event;
1408 unsigned int hardirq_disable_event;
1409 int hardirqs_enabled;
1410 int hardirq_context;
1411 unsigned long softirq_disable_ip;
1412 unsigned long softirq_enable_ip;
1413 unsigned int softirq_disable_event;
1414 unsigned int softirq_enable_event;
1415 int softirqs_enabled;
1416 int softirq_context;
1417 #endif
1418 #ifdef CONFIG_LOCKDEP
1419 # define MAX_LOCK_DEPTH 48UL
1420 u64 curr_chain_key;
1421 int lockdep_depth;
1422 unsigned int lockdep_recursion;
1423 struct held_lock held_locks[MAX_LOCK_DEPTH];
1424 gfp_t lockdep_reclaim_gfp;
1425 #endif
1426
1427 /* journalling filesystem info */
1428 void *journal_info;
1429
1430 /* stacked block device info */
1431 struct bio_list *bio_list;
1432
1433 #ifdef CONFIG_BLOCK
1434 /* stack plugging */
1435 struct blk_plug *plug;
1436 #endif
1437
1438 /* VM state */
1439 struct reclaim_state *reclaim_state;
1440
1441 struct backing_dev_info *backing_dev_info;
1442
1443 struct io_context *io_context;
1444
1445 unsigned long ptrace_message;
1446 siginfo_t *last_siginfo; /* For ptrace use. */
1447 struct task_io_accounting ioac;
1448 #if defined(CONFIG_TASK_XACCT)
1449 u64 acct_rss_mem1; /* accumulated rss usage */
1450 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1451 cputime_t acct_timexpd; /* stime + utime since last update */
1452 #endif
1453 #ifdef CONFIG_CPUSETS
1454 nodemask_t mems_allowed; /* Protected by alloc_lock */
1455 int mems_allowed_change_disable;
1456 int cpuset_mem_spread_rotor;
1457 int cpuset_slab_spread_rotor;
1458 #endif
1459 #ifdef CONFIG_CGROUPS
1460 /* Control Group info protected by css_set_lock */
1461 struct css_set __rcu *cgroups;
1462 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1463 struct list_head cg_list;
1464 #endif
1465 #ifdef CONFIG_FUTEX
1466 struct robust_list_head __user *robust_list;
1467 #ifdef CONFIG_COMPAT
1468 struct compat_robust_list_head __user *compat_robust_list;
1469 #endif
1470 struct list_head pi_state_list;
1471 struct futex_pi_state *pi_state_cache;
1472 #endif
1473 #ifdef CONFIG_PERF_EVENTS
1474 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1475 struct mutex perf_event_mutex;
1476 struct list_head perf_event_list;
1477 #endif
1478 #ifdef CONFIG_NUMA
1479 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1480 short il_next;
1481 #endif
1482 atomic_t fs_excl; /* holding fs exclusive resources */
1483 struct rcu_head rcu;
1484
1485 /*
1486 * cache last used pipe for splice
1487 */
1488 struct pipe_inode_info *splice_pipe;
1489 #ifdef CONFIG_TASK_DELAY_ACCT
1490 struct task_delay_info *delays;
1491 #endif
1492 #ifdef CONFIG_FAULT_INJECTION
1493 int make_it_fail;
1494 #endif
1495 struct prop_local_single dirties;
1496 #ifdef CONFIG_LATENCYTOP
1497 int latency_record_count;
1498 struct latency_record latency_record[LT_SAVECOUNT];
1499 #endif
1500 /*
1501 * time slack values; these are used to round up poll() and
1502 * select() etc timeout values. These are in nanoseconds.
1503 */
1504 unsigned long timer_slack_ns;
1505 unsigned long default_timer_slack_ns;
1506
1507 struct list_head *scm_work_list;
1508 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1509 /* Index of current stored address in ret_stack */
1510 int curr_ret_stack;
1511 /* Stack of return addresses for return function tracing */
1512 struct ftrace_ret_stack *ret_stack;
1513 /* time stamp for last schedule */
1514 unsigned long long ftrace_timestamp;
1515 /*
1516 * Number of functions that haven't been traced
1517 * because of depth overrun.
1518 */
1519 atomic_t trace_overrun;
1520 /* Pause for the tracing */
1521 atomic_t tracing_graph_pause;
1522 #endif
1523 #ifdef CONFIG_TRACING
1524 /* state flags for use by tracers */
1525 unsigned long trace;
1526 /* bitmask of trace recursion */
1527 unsigned long trace_recursion;
1528 #endif /* CONFIG_TRACING */
1529 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1530 struct memcg_batch_info {
1531 int do_batch; /* incremented when batch uncharge started */
1532 struct mem_cgroup *memcg; /* target memcg of uncharge */
1533 unsigned long bytes; /* uncharged usage */
1534 unsigned long memsw_bytes; /* uncharged mem+swap usage */
1535 } memcg_batch;
1536 #endif
1537 };
1538
1539 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1540 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1541
1542 /*
1543 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1544 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1545 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1546 * values are inverted: lower p->prio value means higher priority.
1547 *
1548 * The MAX_USER_RT_PRIO value allows the actual maximum
1549 * RT priority to be separate from the value exported to
1550 * user-space. This allows kernel threads to set their
1551 * priority to a value higher than any user task. Note:
1552 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1553 */
1554
1555 #define MAX_USER_RT_PRIO 100
1556 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1557
1558 #define MAX_PRIO (MAX_RT_PRIO + 40)
1559 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1560
1561 static inline int rt_prio(int prio)
1562 {
1563 if (unlikely(prio < MAX_RT_PRIO))
1564 return 1;
1565 return 0;
1566 }
1567
1568 static inline int rt_task(struct task_struct *p)
1569 {
1570 return rt_prio(p->prio);
1571 }
1572
1573 static inline struct pid *task_pid(struct task_struct *task)
1574 {
1575 return task->pids[PIDTYPE_PID].pid;
1576 }
1577
1578 static inline struct pid *task_tgid(struct task_struct *task)
1579 {
1580 return task->group_leader->pids[PIDTYPE_PID].pid;
1581 }
1582
1583 /*
1584 * Without tasklist or rcu lock it is not safe to dereference
1585 * the result of task_pgrp/task_session even if task == current,
1586 * we can race with another thread doing sys_setsid/sys_setpgid.
1587 */
1588 static inline struct pid *task_pgrp(struct task_struct *task)
1589 {
1590 return task->group_leader->pids[PIDTYPE_PGID].pid;
1591 }
1592
1593 static inline struct pid *task_session(struct task_struct *task)
1594 {
1595 return task->group_leader->pids[PIDTYPE_SID].pid;
1596 }
1597
1598 struct pid_namespace;
1599
1600 /*
1601 * the helpers to get the task's different pids as they are seen
1602 * from various namespaces
1603 *
1604 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1605 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1606 * current.
1607 * task_xid_nr_ns() : id seen from the ns specified;
1608 *
1609 * set_task_vxid() : assigns a virtual id to a task;
1610 *
1611 * see also pid_nr() etc in include/linux/pid.h
1612 */
1613 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1614 struct pid_namespace *ns);
1615
1616 static inline pid_t task_pid_nr(struct task_struct *tsk)
1617 {
1618 return tsk->pid;
1619 }
1620
1621 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1622 struct pid_namespace *ns)
1623 {
1624 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1625 }
1626
1627 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1628 {
1629 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1630 }
1631
1632
1633 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1634 {
1635 return tsk->tgid;
1636 }
1637
1638 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1639
1640 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1641 {
1642 return pid_vnr(task_tgid(tsk));
1643 }
1644
1645
1646 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1647 struct pid_namespace *ns)
1648 {
1649 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1650 }
1651
1652 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1653 {
1654 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1655 }
1656
1657
1658 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1659 struct pid_namespace *ns)
1660 {
1661 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1662 }
1663
1664 static inline pid_t task_session_vnr(struct task_struct *tsk)
1665 {
1666 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1667 }
1668
1669 /* obsolete, do not use */
1670 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1671 {
1672 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1673 }
1674
1675 /**
1676 * pid_alive - check that a task structure is not stale
1677 * @p: Task structure to be checked.
1678 *
1679 * Test if a process is not yet dead (at most zombie state)
1680 * If pid_alive fails, then pointers within the task structure
1681 * can be stale and must not be dereferenced.
1682 */
1683 static inline int pid_alive(struct task_struct *p)
1684 {
1685 return p->pids[PIDTYPE_PID].pid != NULL;
1686 }
1687
1688 /**
1689 * is_global_init - check if a task structure is init
1690 * @tsk: Task structure to be checked.
1691 *
1692 * Check if a task structure is the first user space task the kernel created.
1693 */
1694 static inline int is_global_init(struct task_struct *tsk)
1695 {
1696 return tsk->pid == 1;
1697 }
1698
1699 /*
1700 * is_container_init:
1701 * check whether in the task is init in its own pid namespace.
1702 */
1703 extern int is_container_init(struct task_struct *tsk);
1704
1705 extern struct pid *cad_pid;
1706
1707 extern void free_task(struct task_struct *tsk);
1708 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1709
1710 extern void __put_task_struct(struct task_struct *t);
1711
1712 static inline void put_task_struct(struct task_struct *t)
1713 {
1714 if (atomic_dec_and_test(&t->usage))
1715 __put_task_struct(t);
1716 }
1717
1718 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1719 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1720
1721 /*
1722 * Per process flags
1723 */
1724 #define PF_KSOFTIRQD 0x00000001 /* I am ksoftirqd */
1725 #define PF_STARTING 0x00000002 /* being created */
1726 #define PF_EXITING 0x00000004 /* getting shut down */
1727 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1728 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1729 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1730 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1731 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1732 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1733 #define PF_DUMPCORE 0x00000200 /* dumped core */
1734 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1735 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1736 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1737 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1738 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1739 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1740 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1741 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1742 #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1743 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1744 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1745 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1746 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1747 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1748 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1749 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1750 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1751 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1752 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1753 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1754 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1755
1756 /*
1757 * Only the _current_ task can read/write to tsk->flags, but other
1758 * tasks can access tsk->flags in readonly mode for example
1759 * with tsk_used_math (like during threaded core dumping).
1760 * There is however an exception to this rule during ptrace
1761 * or during fork: the ptracer task is allowed to write to the
1762 * child->flags of its traced child (same goes for fork, the parent
1763 * can write to the child->flags), because we're guaranteed the
1764 * child is not running and in turn not changing child->flags
1765 * at the same time the parent does it.
1766 */
1767 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1768 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1769 #define clear_used_math() clear_stopped_child_used_math(current)
1770 #define set_used_math() set_stopped_child_used_math(current)
1771 #define conditional_stopped_child_used_math(condition, child) \
1772 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1773 #define conditional_used_math(condition) \
1774 conditional_stopped_child_used_math(condition, current)
1775 #define copy_to_stopped_child_used_math(child) \
1776 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1777 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1778 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1779 #define used_math() tsk_used_math(current)
1780
1781 #ifdef CONFIG_PREEMPT_RCU
1782
1783 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1784 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1785 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1786
1787 static inline void rcu_copy_process(struct task_struct *p)
1788 {
1789 p->rcu_read_lock_nesting = 0;
1790 p->rcu_read_unlock_special = 0;
1791 #ifdef CONFIG_TREE_PREEMPT_RCU
1792 p->rcu_blocked_node = NULL;
1793 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1794 #ifdef CONFIG_RCU_BOOST
1795 p->rcu_boost_mutex = NULL;
1796 #endif /* #ifdef CONFIG_RCU_BOOST */
1797 INIT_LIST_HEAD(&p->rcu_node_entry);
1798 }
1799
1800 #else
1801
1802 static inline void rcu_copy_process(struct task_struct *p)
1803 {
1804 }
1805
1806 #endif
1807
1808 #ifdef CONFIG_SMP
1809 extern int set_cpus_allowed_ptr(struct task_struct *p,
1810 const struct cpumask *new_mask);
1811 #else
1812 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1813 const struct cpumask *new_mask)
1814 {
1815 if (!cpumask_test_cpu(0, new_mask))
1816 return -EINVAL;
1817 return 0;
1818 }
1819 #endif
1820
1821 #ifndef CONFIG_CPUMASK_OFFSTACK
1822 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1823 {
1824 return set_cpus_allowed_ptr(p, &new_mask);
1825 }
1826 #endif
1827
1828 /*
1829 * Do not use outside of architecture code which knows its limitations.
1830 *
1831 * sched_clock() has no promise of monotonicity or bounded drift between
1832 * CPUs, use (which you should not) requires disabling IRQs.
1833 *
1834 * Please use one of the three interfaces below.
1835 */
1836 extern unsigned long long notrace sched_clock(void);
1837 /*
1838 * See the comment in kernel/sched_clock.c
1839 */
1840 extern u64 cpu_clock(int cpu);
1841 extern u64 local_clock(void);
1842 extern u64 sched_clock_cpu(int cpu);
1843
1844
1845 extern void sched_clock_init(void);
1846
1847 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1848 static inline void sched_clock_tick(void)
1849 {
1850 }
1851
1852 static inline void sched_clock_idle_sleep_event(void)
1853 {
1854 }
1855
1856 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1857 {
1858 }
1859 #else
1860 /*
1861 * Architectures can set this to 1 if they have specified
1862 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1863 * but then during bootup it turns out that sched_clock()
1864 * is reliable after all:
1865 */
1866 extern int sched_clock_stable;
1867
1868 extern void sched_clock_tick(void);
1869 extern void sched_clock_idle_sleep_event(void);
1870 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1871 #endif
1872
1873 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1874 /*
1875 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1876 * The reason for this explicit opt-in is not to have perf penalty with
1877 * slow sched_clocks.
1878 */
1879 extern void enable_sched_clock_irqtime(void);
1880 extern void disable_sched_clock_irqtime(void);
1881 #else
1882 static inline void enable_sched_clock_irqtime(void) {}
1883 static inline void disable_sched_clock_irqtime(void) {}
1884 #endif
1885
1886 extern unsigned long long
1887 task_sched_runtime(struct task_struct *task);
1888 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1889
1890 /* sched_exec is called by processes performing an exec */
1891 #ifdef CONFIG_SMP
1892 extern void sched_exec(void);
1893 #else
1894 #define sched_exec() {}
1895 #endif
1896
1897 extern void sched_clock_idle_sleep_event(void);
1898 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1899
1900 #ifdef CONFIG_HOTPLUG_CPU
1901 extern void idle_task_exit(void);
1902 #else
1903 static inline void idle_task_exit(void) {}
1904 #endif
1905
1906 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1907 extern void wake_up_idle_cpu(int cpu);
1908 #else
1909 static inline void wake_up_idle_cpu(int cpu) { }
1910 #endif
1911
1912 extern unsigned int sysctl_sched_latency;
1913 extern unsigned int sysctl_sched_min_granularity;
1914 extern unsigned int sysctl_sched_wakeup_granularity;
1915 extern unsigned int sysctl_sched_child_runs_first;
1916
1917 enum sched_tunable_scaling {
1918 SCHED_TUNABLESCALING_NONE,
1919 SCHED_TUNABLESCALING_LOG,
1920 SCHED_TUNABLESCALING_LINEAR,
1921 SCHED_TUNABLESCALING_END,
1922 };
1923 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1924
1925 #ifdef CONFIG_SCHED_DEBUG
1926 extern unsigned int sysctl_sched_migration_cost;
1927 extern unsigned int sysctl_sched_nr_migrate;
1928 extern unsigned int sysctl_sched_time_avg;
1929 extern unsigned int sysctl_timer_migration;
1930 extern unsigned int sysctl_sched_shares_window;
1931
1932 int sched_proc_update_handler(struct ctl_table *table, int write,
1933 void __user *buffer, size_t *length,
1934 loff_t *ppos);
1935 #endif
1936 #ifdef CONFIG_SCHED_DEBUG
1937 static inline unsigned int get_sysctl_timer_migration(void)
1938 {
1939 return sysctl_timer_migration;
1940 }
1941 #else
1942 static inline unsigned int get_sysctl_timer_migration(void)
1943 {
1944 return 1;
1945 }
1946 #endif
1947 extern unsigned int sysctl_sched_rt_period;
1948 extern int sysctl_sched_rt_runtime;
1949
1950 int sched_rt_handler(struct ctl_table *table, int write,
1951 void __user *buffer, size_t *lenp,
1952 loff_t *ppos);
1953
1954 extern unsigned int sysctl_sched_compat_yield;
1955
1956 #ifdef CONFIG_SCHED_AUTOGROUP
1957 extern unsigned int sysctl_sched_autogroup_enabled;
1958
1959 extern void sched_autogroup_create_attach(struct task_struct *p);
1960 extern void sched_autogroup_detach(struct task_struct *p);
1961 extern void sched_autogroup_fork(struct signal_struct *sig);
1962 extern void sched_autogroup_exit(struct signal_struct *sig);
1963 #ifdef CONFIG_PROC_FS
1964 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1965 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1966 #endif
1967 #else
1968 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1969 static inline void sched_autogroup_detach(struct task_struct *p) { }
1970 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1971 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1972 #endif
1973
1974 #ifdef CONFIG_RT_MUTEXES
1975 extern int rt_mutex_getprio(struct task_struct *p);
1976 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1977 extern void rt_mutex_adjust_pi(struct task_struct *p);
1978 #else
1979 static inline int rt_mutex_getprio(struct task_struct *p)
1980 {
1981 return p->normal_prio;
1982 }
1983 # define rt_mutex_adjust_pi(p) do { } while (0)
1984 #endif
1985
1986 extern void set_user_nice(struct task_struct *p, long nice);
1987 extern int task_prio(const struct task_struct *p);
1988 extern int task_nice(const struct task_struct *p);
1989 extern int can_nice(const struct task_struct *p, const int nice);
1990 extern int task_curr(const struct task_struct *p);
1991 extern int idle_cpu(int cpu);
1992 extern int sched_setscheduler(struct task_struct *, int,
1993 const struct sched_param *);
1994 extern int sched_setscheduler_nocheck(struct task_struct *, int,
1995 const struct sched_param *);
1996 extern struct task_struct *idle_task(int cpu);
1997 extern struct task_struct *curr_task(int cpu);
1998 extern void set_curr_task(int cpu, struct task_struct *p);
1999
2000 void yield(void);
2001
2002 /*
2003 * The default (Linux) execution domain.
2004 */
2005 extern struct exec_domain default_exec_domain;
2006
2007 union thread_union {
2008 struct thread_info thread_info;
2009 unsigned long stack[THREAD_SIZE/sizeof(long)];
2010 };
2011
2012 #ifndef __HAVE_ARCH_KSTACK_END
2013 static inline int kstack_end(void *addr)
2014 {
2015 /* Reliable end of stack detection:
2016 * Some APM bios versions misalign the stack
2017 */
2018 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2019 }
2020 #endif
2021
2022 extern union thread_union init_thread_union;
2023 extern struct task_struct init_task;
2024
2025 extern struct mm_struct init_mm;
2026
2027 extern struct pid_namespace init_pid_ns;
2028
2029 /*
2030 * find a task by one of its numerical ids
2031 *
2032 * find_task_by_pid_ns():
2033 * finds a task by its pid in the specified namespace
2034 * find_task_by_vpid():
2035 * finds a task by its virtual pid
2036 *
2037 * see also find_vpid() etc in include/linux/pid.h
2038 */
2039
2040 extern struct task_struct *find_task_by_vpid(pid_t nr);
2041 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2042 struct pid_namespace *ns);
2043
2044 extern void __set_special_pids(struct pid *pid);
2045
2046 /* per-UID process charging. */
2047 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2048 static inline struct user_struct *get_uid(struct user_struct *u)
2049 {
2050 atomic_inc(&u->__count);
2051 return u;
2052 }
2053 extern void free_uid(struct user_struct *);
2054 extern void release_uids(struct user_namespace *ns);
2055
2056 #include <asm/current.h>
2057
2058 extern void do_timer(unsigned long ticks);
2059
2060 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2061 extern int wake_up_process(struct task_struct *tsk);
2062 extern void wake_up_new_task(struct task_struct *tsk,
2063 unsigned long clone_flags);
2064 #ifdef CONFIG_SMP
2065 extern void kick_process(struct task_struct *tsk);
2066 #else
2067 static inline void kick_process(struct task_struct *tsk) { }
2068 #endif
2069 extern void sched_fork(struct task_struct *p, int clone_flags);
2070 extern void sched_dead(struct task_struct *p);
2071
2072 extern void proc_caches_init(void);
2073 extern void flush_signals(struct task_struct *);
2074 extern void __flush_signals(struct task_struct *);
2075 extern void ignore_signals(struct task_struct *);
2076 extern void flush_signal_handlers(struct task_struct *, int force_default);
2077 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2078
2079 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2080 {
2081 unsigned long flags;
2082 int ret;
2083
2084 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2085 ret = dequeue_signal(tsk, mask, info);
2086 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2087
2088 return ret;
2089 }
2090
2091 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2092 sigset_t *mask);
2093 extern void unblock_all_signals(void);
2094 extern void release_task(struct task_struct * p);
2095 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2096 extern int force_sigsegv(int, struct task_struct *);
2097 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2098 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2099 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2100 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2101 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2102 extern int kill_pid(struct pid *pid, int sig, int priv);
2103 extern int kill_proc_info(int, struct siginfo *, pid_t);
2104 extern int do_notify_parent(struct task_struct *, int);
2105 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2106 extern void force_sig(int, struct task_struct *);
2107 extern int send_sig(int, struct task_struct *, int);
2108 extern int zap_other_threads(struct task_struct *p);
2109 extern struct sigqueue *sigqueue_alloc(void);
2110 extern void sigqueue_free(struct sigqueue *);
2111 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2112 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2113 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2114
2115 static inline int kill_cad_pid(int sig, int priv)
2116 {
2117 return kill_pid(cad_pid, sig, priv);
2118 }
2119
2120 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2121 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2122 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2123 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2124
2125 /*
2126 * True if we are on the alternate signal stack.
2127 */
2128 static inline int on_sig_stack(unsigned long sp)
2129 {
2130 #ifdef CONFIG_STACK_GROWSUP
2131 return sp >= current->sas_ss_sp &&
2132 sp - current->sas_ss_sp < current->sas_ss_size;
2133 #else
2134 return sp > current->sas_ss_sp &&
2135 sp - current->sas_ss_sp <= current->sas_ss_size;
2136 #endif
2137 }
2138
2139 static inline int sas_ss_flags(unsigned long sp)
2140 {
2141 return (current->sas_ss_size == 0 ? SS_DISABLE
2142 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2143 }
2144
2145 /*
2146 * Routines for handling mm_structs
2147 */
2148 extern struct mm_struct * mm_alloc(void);
2149
2150 /* mmdrop drops the mm and the page tables */
2151 extern void __mmdrop(struct mm_struct *);
2152 static inline void mmdrop(struct mm_struct * mm)
2153 {
2154 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2155 __mmdrop(mm);
2156 }
2157
2158 /* mmput gets rid of the mappings and all user-space */
2159 extern void mmput(struct mm_struct *);
2160 /* Grab a reference to a task's mm, if it is not already going away */
2161 extern struct mm_struct *get_task_mm(struct task_struct *task);
2162 /* Remove the current tasks stale references to the old mm_struct */
2163 extern void mm_release(struct task_struct *, struct mm_struct *);
2164 /* Allocate a new mm structure and copy contents from tsk->mm */
2165 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2166
2167 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2168 struct task_struct *, struct pt_regs *);
2169 extern void flush_thread(void);
2170 extern void exit_thread(void);
2171
2172 extern void exit_files(struct task_struct *);
2173 extern void __cleanup_sighand(struct sighand_struct *);
2174
2175 extern void exit_itimers(struct signal_struct *);
2176 extern void flush_itimer_signals(void);
2177
2178 extern NORET_TYPE void do_group_exit(int);
2179
2180 extern void daemonize(const char *, ...);
2181 extern int allow_signal(int);
2182 extern int disallow_signal(int);
2183
2184 extern int do_execve(const char *,
2185 const char __user * const __user *,
2186 const char __user * const __user *, struct pt_regs *);
2187 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2188 struct task_struct *fork_idle(int);
2189
2190 extern void set_task_comm(struct task_struct *tsk, char *from);
2191 extern char *get_task_comm(char *to, struct task_struct *tsk);
2192
2193 #ifdef CONFIG_SMP
2194 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2195 #else
2196 static inline unsigned long wait_task_inactive(struct task_struct *p,
2197 long match_state)
2198 {
2199 return 1;
2200 }
2201 #endif
2202
2203 #define next_task(p) \
2204 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2205
2206 #define for_each_process(p) \
2207 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2208
2209 extern bool current_is_single_threaded(void);
2210
2211 /*
2212 * Careful: do_each_thread/while_each_thread is a double loop so
2213 * 'break' will not work as expected - use goto instead.
2214 */
2215 #define do_each_thread(g, t) \
2216 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2217
2218 #define while_each_thread(g, t) \
2219 while ((t = next_thread(t)) != g)
2220
2221 static inline int get_nr_threads(struct task_struct *tsk)
2222 {
2223 return tsk->signal->nr_threads;
2224 }
2225
2226 /* de_thread depends on thread_group_leader not being a pid based check */
2227 #define thread_group_leader(p) (p == p->group_leader)
2228
2229 /* Do to the insanities of de_thread it is possible for a process
2230 * to have the pid of the thread group leader without actually being
2231 * the thread group leader. For iteration through the pids in proc
2232 * all we care about is that we have a task with the appropriate
2233 * pid, we don't actually care if we have the right task.
2234 */
2235 static inline int has_group_leader_pid(struct task_struct *p)
2236 {
2237 return p->pid == p->tgid;
2238 }
2239
2240 static inline
2241 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2242 {
2243 return p1->tgid == p2->tgid;
2244 }
2245
2246 static inline struct task_struct *next_thread(const struct task_struct *p)
2247 {
2248 return list_entry_rcu(p->thread_group.next,
2249 struct task_struct, thread_group);
2250 }
2251
2252 static inline int thread_group_empty(struct task_struct *p)
2253 {
2254 return list_empty(&p->thread_group);
2255 }
2256
2257 #define delay_group_leader(p) \
2258 (thread_group_leader(p) && !thread_group_empty(p))
2259
2260 static inline int task_detached(struct task_struct *p)
2261 {
2262 return p->exit_signal == -1;
2263 }
2264
2265 /*
2266 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2267 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2268 * pins the final release of task.io_context. Also protects ->cpuset and
2269 * ->cgroup.subsys[].
2270 *
2271 * Nests both inside and outside of read_lock(&tasklist_lock).
2272 * It must not be nested with write_lock_irq(&tasklist_lock),
2273 * neither inside nor outside.
2274 */
2275 static inline void task_lock(struct task_struct *p)
2276 {
2277 spin_lock(&p->alloc_lock);
2278 }
2279
2280 static inline void task_unlock(struct task_struct *p)
2281 {
2282 spin_unlock(&p->alloc_lock);
2283 }
2284
2285 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2286 unsigned long *flags);
2287
2288 #define lock_task_sighand(tsk, flags) \
2289 ({ struct sighand_struct *__ss; \
2290 __cond_lock(&(tsk)->sighand->siglock, \
2291 (__ss = __lock_task_sighand(tsk, flags))); \
2292 __ss; \
2293 }) \
2294
2295 static inline void unlock_task_sighand(struct task_struct *tsk,
2296 unsigned long *flags)
2297 {
2298 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2299 }
2300
2301 #ifndef __HAVE_THREAD_FUNCTIONS
2302
2303 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2304 #define task_stack_page(task) ((task)->stack)
2305
2306 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2307 {
2308 *task_thread_info(p) = *task_thread_info(org);
2309 task_thread_info(p)->task = p;
2310 }
2311
2312 static inline unsigned long *end_of_stack(struct task_struct *p)
2313 {
2314 return (unsigned long *)(task_thread_info(p) + 1);
2315 }
2316
2317 #endif
2318
2319 static inline int object_is_on_stack(void *obj)
2320 {
2321 void *stack = task_stack_page(current);
2322
2323 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2324 }
2325
2326 extern void thread_info_cache_init(void);
2327
2328 #ifdef CONFIG_DEBUG_STACK_USAGE
2329 static inline unsigned long stack_not_used(struct task_struct *p)
2330 {
2331 unsigned long *n = end_of_stack(p);
2332
2333 do { /* Skip over canary */
2334 n++;
2335 } while (!*n);
2336
2337 return (unsigned long)n - (unsigned long)end_of_stack(p);
2338 }
2339 #endif
2340
2341 /* set thread flags in other task's structures
2342 * - see asm/thread_info.h for TIF_xxxx flags available
2343 */
2344 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2345 {
2346 set_ti_thread_flag(task_thread_info(tsk), flag);
2347 }
2348
2349 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2350 {
2351 clear_ti_thread_flag(task_thread_info(tsk), flag);
2352 }
2353
2354 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2355 {
2356 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2357 }
2358
2359 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2360 {
2361 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2362 }
2363
2364 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2365 {
2366 return test_ti_thread_flag(task_thread_info(tsk), flag);
2367 }
2368
2369 static inline void set_tsk_need_resched(struct task_struct *tsk)
2370 {
2371 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2372 }
2373
2374 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2375 {
2376 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2377 }
2378
2379 static inline int test_tsk_need_resched(struct task_struct *tsk)
2380 {
2381 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2382 }
2383
2384 static inline int restart_syscall(void)
2385 {
2386 set_tsk_thread_flag(current, TIF_SIGPENDING);
2387 return -ERESTARTNOINTR;
2388 }
2389
2390 static inline int signal_pending(struct task_struct *p)
2391 {
2392 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2393 }
2394
2395 static inline int __fatal_signal_pending(struct task_struct *p)
2396 {
2397 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2398 }
2399
2400 static inline int fatal_signal_pending(struct task_struct *p)
2401 {
2402 return signal_pending(p) && __fatal_signal_pending(p);
2403 }
2404
2405 static inline int signal_pending_state(long state, struct task_struct *p)
2406 {
2407 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2408 return 0;
2409 if (!signal_pending(p))
2410 return 0;
2411
2412 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2413 }
2414
2415 static inline int need_resched(void)
2416 {
2417 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2418 }
2419
2420 /*
2421 * cond_resched() and cond_resched_lock(): latency reduction via
2422 * explicit rescheduling in places that are safe. The return
2423 * value indicates whether a reschedule was done in fact.
2424 * cond_resched_lock() will drop the spinlock before scheduling,
2425 * cond_resched_softirq() will enable bhs before scheduling.
2426 */
2427 extern int _cond_resched(void);
2428
2429 #define cond_resched() ({ \
2430 __might_sleep(__FILE__, __LINE__, 0); \
2431 _cond_resched(); \
2432 })
2433
2434 extern int __cond_resched_lock(spinlock_t *lock);
2435
2436 #ifdef CONFIG_PREEMPT
2437 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2438 #else
2439 #define PREEMPT_LOCK_OFFSET 0
2440 #endif
2441
2442 #define cond_resched_lock(lock) ({ \
2443 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2444 __cond_resched_lock(lock); \
2445 })
2446
2447 extern int __cond_resched_softirq(void);
2448
2449 #define cond_resched_softirq() ({ \
2450 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2451 __cond_resched_softirq(); \
2452 })
2453
2454 /*
2455 * Does a critical section need to be broken due to another
2456 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2457 * but a general need for low latency)
2458 */
2459 static inline int spin_needbreak(spinlock_t *lock)
2460 {
2461 #ifdef CONFIG_PREEMPT
2462 return spin_is_contended(lock);
2463 #else
2464 return 0;
2465 #endif
2466 }
2467
2468 /*
2469 * Thread group CPU time accounting.
2470 */
2471 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2472 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2473
2474 static inline void thread_group_cputime_init(struct signal_struct *sig)
2475 {
2476 spin_lock_init(&sig->cputimer.lock);
2477 }
2478
2479 /*
2480 * Reevaluate whether the task has signals pending delivery.
2481 * Wake the task if so.
2482 * This is required every time the blocked sigset_t changes.
2483 * callers must hold sighand->siglock.
2484 */
2485 extern void recalc_sigpending_and_wake(struct task_struct *t);
2486 extern void recalc_sigpending(void);
2487
2488 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2489
2490 /*
2491 * Wrappers for p->thread_info->cpu access. No-op on UP.
2492 */
2493 #ifdef CONFIG_SMP
2494
2495 static inline unsigned int task_cpu(const struct task_struct *p)
2496 {
2497 return task_thread_info(p)->cpu;
2498 }
2499
2500 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2501
2502 #else
2503
2504 static inline unsigned int task_cpu(const struct task_struct *p)
2505 {
2506 return 0;
2507 }
2508
2509 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2510 {
2511 }
2512
2513 #endif /* CONFIG_SMP */
2514
2515 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2516 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2517
2518 extern void normalize_rt_tasks(void);
2519
2520 #ifdef CONFIG_CGROUP_SCHED
2521
2522 extern struct task_group root_task_group;
2523
2524 extern struct task_group *sched_create_group(struct task_group *parent);
2525 extern void sched_destroy_group(struct task_group *tg);
2526 extern void sched_move_task(struct task_struct *tsk);
2527 #ifdef CONFIG_FAIR_GROUP_SCHED
2528 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2529 extern unsigned long sched_group_shares(struct task_group *tg);
2530 #endif
2531 #ifdef CONFIG_RT_GROUP_SCHED
2532 extern int sched_group_set_rt_runtime(struct task_group *tg,
2533 long rt_runtime_us);
2534 extern long sched_group_rt_runtime(struct task_group *tg);
2535 extern int sched_group_set_rt_period(struct task_group *tg,
2536 long rt_period_us);
2537 extern long sched_group_rt_period(struct task_group *tg);
2538 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2539 #endif
2540 #endif
2541
2542 extern int task_can_switch_user(struct user_struct *up,
2543 struct task_struct *tsk);
2544
2545 #ifdef CONFIG_TASK_XACCT
2546 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2547 {
2548 tsk->ioac.rchar += amt;
2549 }
2550
2551 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2552 {
2553 tsk->ioac.wchar += amt;
2554 }
2555
2556 static inline void inc_syscr(struct task_struct *tsk)
2557 {
2558 tsk->ioac.syscr++;
2559 }
2560
2561 static inline void inc_syscw(struct task_struct *tsk)
2562 {
2563 tsk->ioac.syscw++;
2564 }
2565 #else
2566 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2567 {
2568 }
2569
2570 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2571 {
2572 }
2573
2574 static inline void inc_syscr(struct task_struct *tsk)
2575 {
2576 }
2577
2578 static inline void inc_syscw(struct task_struct *tsk)
2579 {
2580 }
2581 #endif
2582
2583 #ifndef TASK_SIZE_OF
2584 #define TASK_SIZE_OF(tsk) TASK_SIZE
2585 #endif
2586
2587 /*
2588 * Call the function if the target task is executing on a CPU right now:
2589 */
2590 extern void task_oncpu_function_call(struct task_struct *p,
2591 void (*func) (void *info), void *info);
2592
2593
2594 #ifdef CONFIG_MM_OWNER
2595 extern void mm_update_next_owner(struct mm_struct *mm);
2596 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2597 #else
2598 static inline void mm_update_next_owner(struct mm_struct *mm)
2599 {
2600 }
2601
2602 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2603 {
2604 }
2605 #endif /* CONFIG_MM_OWNER */
2606
2607 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2608 unsigned int limit)
2609 {
2610 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2611 }
2612
2613 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2614 unsigned int limit)
2615 {
2616 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2617 }
2618
2619 static inline unsigned long rlimit(unsigned int limit)
2620 {
2621 return task_rlimit(current, limit);
2622 }
2623
2624 static inline unsigned long rlimit_max(unsigned int limit)
2625 {
2626 return task_rlimit_max(current, limit);
2627 }
2628
2629 #endif /* __KERNEL__ */
2630
2631 #endif