projects / deliverable / linux.git / blob
? search:
summary | shortlog | log | commit | commitdiff | tree
blame | history | raw | HEAD
Merge branch 'for-2.6.40/splice' of git://git.kernel.dk/linux-2.6-block
[deliverable/linux.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 void lockup_detector_init(void);
319 #else
320 static inline void touch_softlockup_watchdog(void)
321 {
322 }
323 static inline void touch_softlockup_watchdog_sync(void)
324 {
325 }
326 static inline void touch_all_softlockup_watchdogs(void)
327 {
328 }
329 static inline void lockup_detector_init(void)
330 {
331 }
332 #endif
333
334 #ifdef CONFIG_DETECT_HUNG_TASK
335 extern unsigned int sysctl_hung_task_panic;
336 extern unsigned long sysctl_hung_task_check_count;
337 extern unsigned long sysctl_hung_task_timeout_secs;
338 extern unsigned long sysctl_hung_task_warnings;
339 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write,
340 void __user *buffer,
341 size_t *lenp, loff_t *ppos);
342 #else
343 /* Avoid need for ifdefs elsewhere in the code */
344 enum { sysctl_hung_task_timeout_secs = 0 };
345 #endif
346
347 /* Attach to any functions which should be ignored in wchan output. */
348 #define __sched __attribute__((__section__(".sched.text")))
349
350 /* Linker adds these: start and end of __sched functions */
351 extern char __sched_text_start[], __sched_text_end[];
352
353 /* Is this address in the __sched functions? */
354 extern int in_sched_functions(unsigned long addr);
355
356 #define MAX_SCHEDULE_TIMEOUT LONG_MAX
357 extern signed long schedule_timeout(signed long timeout);
358 extern signed long schedule_timeout_interruptible(signed long timeout);
359 extern signed long schedule_timeout_killable(signed long timeout);
360 extern signed long schedule_timeout_uninterruptible(signed long timeout);
361 asmlinkage void schedule(void);
362 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner);
363
364 struct nsproxy;
365 struct user_namespace;
366
367 /*
368 * Default maximum number of active map areas, this limits the number of vmas
369 * per mm struct. Users can overwrite this number by sysctl but there is a
370 * problem.
371 *
372 * When a program's coredump is generated as ELF format, a section is created
373 * per a vma. In ELF, the number of sections is represented in unsigned short.
374 * This means the number of sections should be smaller than 65535 at coredump.
375 * Because the kernel adds some informative sections to a image of program at
376 * generating coredump, we need some margin. The number of extra sections is
377 * 1-3 now and depends on arch. We use "5" as safe margin, here.
378 */
379 #define MAPCOUNT_ELF_CORE_MARGIN (5)
380 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
381
382 extern int sysctl_max_map_count;
383
384 #include <linux/aio.h>
385
386 #ifdef CONFIG_MMU
387 extern void arch_pick_mmap_layout(struct mm_struct *mm);
388 extern unsigned long
389 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
390 unsigned long, unsigned long);
391 extern unsigned long
392 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
393 unsigned long len, unsigned long pgoff,
394 unsigned long flags);
395 extern void arch_unmap_area(struct mm_struct *, unsigned long);
396 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
397 #else
398 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
399 #endif
400
401
402 extern void set_dumpable(struct mm_struct *mm, int value);
403 extern int get_dumpable(struct mm_struct *mm);
404
405 /* mm flags */
406 /* dumpable bits */
407 #define MMF_DUMPABLE 0 /* core dump is permitted */
408 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
409
410 #define MMF_DUMPABLE_BITS 2
411 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
412
413 /* coredump filter bits */
414 #define MMF_DUMP_ANON_PRIVATE 2
415 #define MMF_DUMP_ANON_SHARED 3
416 #define MMF_DUMP_MAPPED_PRIVATE 4
417 #define MMF_DUMP_MAPPED_SHARED 5
418 #define MMF_DUMP_ELF_HEADERS 6
419 #define MMF_DUMP_HUGETLB_PRIVATE 7
420 #define MMF_DUMP_HUGETLB_SHARED 8
421
422 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
423 #define MMF_DUMP_FILTER_BITS 7
424 #define MMF_DUMP_FILTER_MASK \
425 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
426 #define MMF_DUMP_FILTER_DEFAULT \
427 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
428 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
429
430 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
431 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
432 #else
433 # define MMF_DUMP_MASK_DEFAULT_ELF 0
434 #endif
435 /* leave room for more dump flags */
436 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
437 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */
438
439 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
440
441 struct sighand_struct {
442 atomic_t count;
443 struct k_sigaction action[_NSIG];
444 spinlock_t siglock;
445 wait_queue_head_t signalfd_wqh;
446 };
447
448 struct pacct_struct {
449 int ac_flag;
450 long ac_exitcode;
451 unsigned long ac_mem;
452 cputime_t ac_utime, ac_stime;
453 unsigned long ac_minflt, ac_majflt;
454 };
455
456 struct cpu_itimer {
457 cputime_t expires;
458 cputime_t incr;
459 u32 error;
460 u32 incr_error;
461 };
462
463 /**
464 * struct task_cputime - collected CPU time counts
465 * @utime: time spent in user mode, in &cputime_t units
466 * @stime: time spent in kernel mode, in &cputime_t units
467 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
468 *
469 * This structure groups together three kinds of CPU time that are
470 * tracked for threads and thread groups. Most things considering
471 * CPU time want to group these counts together and treat all three
472 * of them in parallel.
473 */
474 struct task_cputime {
475 cputime_t utime;
476 cputime_t stime;
477 unsigned long long sum_exec_runtime;
478 };
479 /* Alternate field names when used to cache expirations. */
480 #define prof_exp stime
481 #define virt_exp utime
482 #define sched_exp sum_exec_runtime
483
484 #define INIT_CPUTIME \
485 (struct task_cputime) { \
486 .utime = cputime_zero, \
487 .stime = cputime_zero, \
488 .sum_exec_runtime = 0, \
489 }
490
491 /*
492 * Disable preemption until the scheduler is running.
493 * Reset by start_kernel()->sched_init()->init_idle().
494 *
495 * We include PREEMPT_ACTIVE to avoid cond_resched() from working
496 * before the scheduler is active -- see should_resched().
497 */
498 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
499
500 /**
501 * struct thread_group_cputimer - thread group interval timer counts
502 * @cputime: thread group interval timers.
503 * @running: non-zero when there are timers running and
504 * @cputime receives updates.
505 * @lock: lock for fields in this struct.
506 *
507 * This structure contains the version of task_cputime, above, that is
508 * used for thread group CPU timer calculations.
509 */
510 struct thread_group_cputimer {
511 struct task_cputime cputime;
512 int running;
513 spinlock_t lock;
514 };
515
516 struct autogroup;
517
518 /*
519 * NOTE! "signal_struct" does not have its own
520 * locking, because a shared signal_struct always
521 * implies a shared sighand_struct, so locking
522 * sighand_struct is always a proper superset of
523 * the locking of signal_struct.
524 */
525 struct signal_struct {
526 atomic_t sigcnt;
527 atomic_t live;
528 int nr_threads;
529
530 wait_queue_head_t wait_chldexit; /* for wait4() */
531
532 /* current thread group signal load-balancing target: */
533 struct task_struct *curr_target;
534
535 /* shared signal handling: */
536 struct sigpending shared_pending;
537
538 /* thread group exit support */
539 int group_exit_code;
540 /* overloaded:
541 * - notify group_exit_task when ->count is equal to notify_count
542 * - everyone except group_exit_task is stopped during signal delivery
543 * of fatal signals, group_exit_task processes the signal.
544 */
545 int notify_count;
546 struct task_struct *group_exit_task;
547
548 /* thread group stop support, overloads group_exit_code too */
549 int group_stop_count;
550 unsigned int flags; /* see SIGNAL_* flags below */
551
552 /* POSIX.1b Interval Timers */
553 struct list_head posix_timers;
554
555 /* ITIMER_REAL timer for the process */
556 struct hrtimer real_timer;
557 struct pid *leader_pid;
558 ktime_t it_real_incr;
559
560 /*
561 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
562 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
563 * values are defined to 0 and 1 respectively
564 */
565 struct cpu_itimer it[2];
566
567 /*
568 * Thread group totals for process CPU timers.
569 * See thread_group_cputimer(), et al, for details.
570 */
571 struct thread_group_cputimer cputimer;
572
573 /* Earliest-expiration cache. */
574 struct task_cputime cputime_expires;
575
576 struct list_head cpu_timers[3];
577
578 struct pid *tty_old_pgrp;
579
580 /* boolean value for session group leader */
581 int leader;
582
583 struct tty_struct *tty; /* NULL if no tty */
584
585 #ifdef CONFIG_SCHED_AUTOGROUP
586 struct autogroup *autogroup;
587 #endif
588 /*
589 * Cumulative resource counters for dead threads in the group,
590 * and for reaped dead child processes forked by this group.
591 * Live threads maintain their own counters and add to these
592 * in __exit_signal, except for the group leader.
593 */
594 cputime_t utime, stime, cutime, cstime;
595 cputime_t gtime;
596 cputime_t cgtime;
597 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
598 cputime_t prev_utime, prev_stime;
599 #endif
600 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
601 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
602 unsigned long inblock, oublock, cinblock, coublock;
603 unsigned long maxrss, cmaxrss;
604 struct task_io_accounting ioac;
605
606 /*
607 * Cumulative ns of schedule CPU time fo dead threads in the
608 * group, not including a zombie group leader, (This only differs
609 * from jiffies_to_ns(utime + stime) if sched_clock uses something
610 * other than jiffies.)
611 */
612 unsigned long long sum_sched_runtime;
613
614 /*
615 * We don't bother to synchronize most readers of this at all,
616 * because there is no reader checking a limit that actually needs
617 * to get both rlim_cur and rlim_max atomically, and either one
618 * alone is a single word that can safely be read normally.
619 * getrlimit/setrlimit use task_lock(current->group_leader) to
620 * protect this instead of the siglock, because they really
621 * have no need to disable irqs.
622 */
623 struct rlimit rlim[RLIM_NLIMITS];
624
625 #ifdef CONFIG_BSD_PROCESS_ACCT
626 struct pacct_struct pacct; /* per-process accounting information */
627 #endif
628 #ifdef CONFIG_TASKSTATS
629 struct taskstats *stats;
630 #endif
631 #ifdef CONFIG_AUDIT
632 unsigned audit_tty;
633 struct tty_audit_buf *tty_audit_buf;
634 #endif
635
636 int oom_adj; /* OOM kill score adjustment (bit shift) */
637 int oom_score_adj; /* OOM kill score adjustment */
638 int oom_score_adj_min; /* OOM kill score adjustment minimum value.
639 * Only settable by CAP_SYS_RESOURCE. */
640
641 struct mutex cred_guard_mutex; /* guard against foreign influences on
642 * credential calculations
643 * (notably. ptrace) */
644 };
645
646 /* Context switch must be unlocked if interrupts are to be enabled */
647 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
648 # define __ARCH_WANT_UNLOCKED_CTXSW
649 #endif
650
651 /*
652 * Bits in flags field of signal_struct.
653 */
654 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
655 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */
656 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */
657 /*
658 * Pending notifications to parent.
659 */
660 #define SIGNAL_CLD_STOPPED 0x00000010
661 #define SIGNAL_CLD_CONTINUED 0x00000020
662 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
663
664 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
665
666 /* If true, all threads except ->group_exit_task have pending SIGKILL */
667 static inline int signal_group_exit(const struct signal_struct *sig)
668 {
669 return (sig->flags & SIGNAL_GROUP_EXIT) ||
670 (sig->group_exit_task != NULL);
671 }
672
673 /*
674 * Some day this will be a full-fledged user tracking system..
675 */
676 struct user_struct {
677 atomic_t __count; /* reference count */
678 atomic_t processes; /* How many processes does this user have? */
679 atomic_t files; /* How many open files does this user have? */
680 atomic_t sigpending; /* How many pending signals does this user have? */
681 #ifdef CONFIG_INOTIFY_USER
682 atomic_t inotify_watches; /* How many inotify watches does this user have? */
683 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
684 #endif
685 #ifdef CONFIG_FANOTIFY
686 atomic_t fanotify_listeners;
687 #endif
688 #ifdef CONFIG_EPOLL
689 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
690 #endif
691 #ifdef CONFIG_POSIX_MQUEUE
692 /* protected by mq_lock */
693 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
694 #endif
695 unsigned long locked_shm; /* How many pages of mlocked shm ? */
696
697 #ifdef CONFIG_KEYS
698 struct key *uid_keyring; /* UID specific keyring */
699 struct key *session_keyring; /* UID's default session keyring */
700 #endif
701
702 /* Hash table maintenance information */
703 struct hlist_node uidhash_node;
704 uid_t uid;
705 struct user_namespace *user_ns;
706
707 #ifdef CONFIG_PERF_EVENTS
708 atomic_long_t locked_vm;
709 #endif
710 };
711
712 extern int uids_sysfs_init(void);
713
714 extern struct user_struct *find_user(uid_t);
715
716 extern struct user_struct root_user;
717 #define INIT_USER (&root_user)
718
719
720 struct backing_dev_info;
721 struct reclaim_state;
722
723 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
724 struct sched_info {
725 /* cumulative counters */
726 unsigned long pcount; /* # of times run on this cpu */
727 unsigned long long run_delay; /* time spent waiting on a runqueue */
728
729 /* timestamps */
730 unsigned long long last_arrival,/* when we last ran on a cpu */
731 last_queued; /* when we were last queued to run */
732 };
733 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
734
735 #ifdef CONFIG_TASK_DELAY_ACCT
736 struct task_delay_info {
737 spinlock_t lock;
738 unsigned int flags; /* Private per-task flags */
739
740 /* For each stat XXX, add following, aligned appropriately
741 *
742 * struct timespec XXX_start, XXX_end;
743 * u64 XXX_delay;
744 * u32 XXX_count;
745 *
746 * Atomicity of updates to XXX_delay, XXX_count protected by
747 * single lock above (split into XXX_lock if contention is an issue).
748 */
749
750 /*
751 * XXX_count is incremented on every XXX operation, the delay
752 * associated with the operation is added to XXX_delay.
753 * XXX_delay contains the accumulated delay time in nanoseconds.
754 */
755 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
756 u64 blkio_delay; /* wait for sync block io completion */
757 u64 swapin_delay; /* wait for swapin block io completion */
758 u32 blkio_count; /* total count of the number of sync block */
759 /* io operations performed */
760 u32 swapin_count; /* total count of the number of swapin block */
761 /* io operations performed */
762
763 struct timespec freepages_start, freepages_end;
764 u64 freepages_delay; /* wait for memory reclaim */
765 u32 freepages_count; /* total count of memory reclaim */
766 };
767 #endif /* CONFIG_TASK_DELAY_ACCT */
768
769 static inline int sched_info_on(void)
770 {
771 #ifdef CONFIG_SCHEDSTATS
772 return 1;
773 #elif defined(CONFIG_TASK_DELAY_ACCT)
774 extern int delayacct_on;
775 return delayacct_on;
776 #else
777 return 0;
778 #endif
779 }
780
781 enum cpu_idle_type {
782 CPU_IDLE,
783 CPU_NOT_IDLE,
784 CPU_NEWLY_IDLE,
785 CPU_MAX_IDLE_TYPES
786 };
787
788 /*
789 * Increase resolution of nice-level calculations for 64-bit architectures.
790 * The extra resolution improves shares distribution and load balancing of
791 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup
792 * hierarchies, especially on larger systems. This is not a user-visible change
793 * and does not change the user-interface for setting shares/weights.
794 *
795 * We increase resolution only if we have enough bits to allow this increased
796 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution
797 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the
798 * increased costs.
799 */
800 #if BITS_PER_LONG > 32
801 # define SCHED_LOAD_RESOLUTION 10
802 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION)
803 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION)
804 #else
805 # define SCHED_LOAD_RESOLUTION 0
806 # define scale_load(w) (w)
807 # define scale_load_down(w) (w)
808 #endif
809
810 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION)
811 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
812
813 /*
814 * Increase resolution of cpu_power calculations
815 */
816 #define SCHED_POWER_SHIFT 10
817 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT)
818
819 /*
820 * sched-domains (multiprocessor balancing) declarations:
821 */
822 #ifdef CONFIG_SMP
823 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
824 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
825 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
826 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
827 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
828 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
829 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
830 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
831 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
832 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
833 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
834 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */
835 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
836
837 enum powersavings_balance_level {
838 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
839 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
840 * first for long running threads
841 */
842 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
843 * cpu package for power savings
844 */
845 MAX_POWERSAVINGS_BALANCE_LEVELS
846 };
847
848 extern int sched_mc_power_savings, sched_smt_power_savings;
849
850 static inline int sd_balance_for_mc_power(void)
851 {
852 if (sched_smt_power_savings)
853 return SD_POWERSAVINGS_BALANCE;
854
855 if (!sched_mc_power_savings)
856 return SD_PREFER_SIBLING;
857
858 return 0;
859 }
860
861 static inline int sd_balance_for_package_power(void)
862 {
863 if (sched_mc_power_savings | sched_smt_power_savings)
864 return SD_POWERSAVINGS_BALANCE;
865
866 return SD_PREFER_SIBLING;
867 }
868
869 extern int __weak arch_sd_sibiling_asym_packing(void);
870
871 /*
872 * Optimise SD flags for power savings:
873 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings.
874 * Keep default SD flags if sched_{smt,mc}_power_saving=0
875 */
876
877 static inline int sd_power_saving_flags(void)
878 {
879 if (sched_mc_power_savings | sched_smt_power_savings)
880 return SD_BALANCE_NEWIDLE;
881
882 return 0;
883 }
884
885 struct sched_group {
886 struct sched_group *next; /* Must be a circular list */
887 atomic_t ref;
888
889 /*
890 * CPU power of this group, SCHED_LOAD_SCALE being max power for a
891 * single CPU.
892 */
893 unsigned int cpu_power, cpu_power_orig;
894 unsigned int group_weight;
895
896 /*
897 * The CPUs this group covers.
898 *
899 * NOTE: this field is variable length. (Allocated dynamically
900 * by attaching extra space to the end of the structure,
901 * depending on how many CPUs the kernel has booted up with)
902 */
903 unsigned long cpumask[0];
904 };
905
906 static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
907 {
908 return to_cpumask(sg->cpumask);
909 }
910
911 struct sched_domain_attr {
912 int relax_domain_level;
913 };
914
915 #define SD_ATTR_INIT (struct sched_domain_attr) { \
916 .relax_domain_level = -1, \
917 }
918
919 extern int sched_domain_level_max;
920
921 struct sched_domain {
922 /* These fields must be setup */
923 struct sched_domain *parent; /* top domain must be null terminated */
924 struct sched_domain *child; /* bottom domain must be null terminated */
925 struct sched_group *groups; /* the balancing groups of the domain */
926 unsigned long min_interval; /* Minimum balance interval ms */
927 unsigned long max_interval; /* Maximum balance interval ms */
928 unsigned int busy_factor; /* less balancing by factor if busy */
929 unsigned int imbalance_pct; /* No balance until over watermark */
930 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
931 unsigned int busy_idx;
932 unsigned int idle_idx;
933 unsigned int newidle_idx;
934 unsigned int wake_idx;
935 unsigned int forkexec_idx;
936 unsigned int smt_gain;
937 int flags; /* See SD_* */
938 int level;
939
940 /* Runtime fields. */
941 unsigned long last_balance; /* init to jiffies. units in jiffies */
942 unsigned int balance_interval; /* initialise to 1. units in ms. */
943 unsigned int nr_balance_failed; /* initialise to 0 */
944
945 u64 last_update;
946
947 #ifdef CONFIG_SCHEDSTATS
948 /* load_balance() stats */
949 unsigned int lb_count[CPU_MAX_IDLE_TYPES];
950 unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
951 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
952 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
953 unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
954 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
955 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
956 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
957
958 /* Active load balancing */
959 unsigned int alb_count;
960 unsigned int alb_failed;
961 unsigned int alb_pushed;
962
963 /* SD_BALANCE_EXEC stats */
964 unsigned int sbe_count;
965 unsigned int sbe_balanced;
966 unsigned int sbe_pushed;
967
968 /* SD_BALANCE_FORK stats */
969 unsigned int sbf_count;
970 unsigned int sbf_balanced;
971 unsigned int sbf_pushed;
972
973 /* try_to_wake_up() stats */
974 unsigned int ttwu_wake_remote;
975 unsigned int ttwu_move_affine;
976 unsigned int ttwu_move_balance;
977 #endif
978 #ifdef CONFIG_SCHED_DEBUG
979 char *name;
980 #endif
981 union {
982 void *private; /* used during construction */
983 struct rcu_head rcu; /* used during destruction */
984 };
985
986 unsigned int span_weight;
987 /*
988 * Span of all CPUs in this domain.
989 *
990 * NOTE: this field is variable length. (Allocated dynamically
991 * by attaching extra space to the end of the structure,
992 * depending on how many CPUs the kernel has booted up with)
993 */
994 unsigned long span[0];
995 };
996
997 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
998 {
999 return to_cpumask(sd->span);
1000 }
1001
1002 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1003 struct sched_domain_attr *dattr_new);
1004
1005 /* Allocate an array of sched domains, for partition_sched_domains(). */
1006 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1007 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1008
1009 /* Test a flag in parent sched domain */
1010 static inline int test_sd_parent(struct sched_domain *sd, int flag)
1011 {
1012 if (sd->parent && (sd->parent->flags & flag))
1013 return 1;
1014
1015 return 0;
1016 }
1017
1018 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1019 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1020
1021 #else /* CONFIG_SMP */
1022
1023 struct sched_domain_attr;
1024
1025 static inline void
1026 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1027 struct sched_domain_attr *dattr_new)
1028 {
1029 }
1030 #endif /* !CONFIG_SMP */
1031
1032
1033 struct io_context; /* See blkdev.h */
1034
1035
1036 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1037 extern void prefetch_stack(struct task_struct *t);
1038 #else
1039 static inline void prefetch_stack(struct task_struct *t) { }
1040 #endif
1041
1042 struct audit_context; /* See audit.c */
1043 struct mempolicy;
1044 struct pipe_inode_info;
1045 struct uts_namespace;
1046
1047 struct rq;
1048 struct sched_domain;
1049
1050 /*
1051 * wake flags
1052 */
1053 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1054 #define WF_FORK 0x02 /* child wakeup after fork */
1055
1056 #define ENQUEUE_WAKEUP 1
1057 #define ENQUEUE_HEAD 2
1058 #ifdef CONFIG_SMP
1059 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */
1060 #else
1061 #define ENQUEUE_WAKING 0
1062 #endif
1063
1064 #define DEQUEUE_SLEEP 1
1065
1066 struct sched_class {
1067 const struct sched_class *next;
1068
1069 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags);
1070 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags);
1071 void (*yield_task) (struct rq *rq);
1072 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt);
1073
1074 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1075
1076 struct task_struct * (*pick_next_task) (struct rq *rq);
1077 void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1078
1079 #ifdef CONFIG_SMP
1080 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1081
1082 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1083 void (*post_schedule) (struct rq *this_rq);
1084 void (*task_waking) (struct task_struct *task);
1085 void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1086
1087 void (*set_cpus_allowed)(struct task_struct *p,
1088 const struct cpumask *newmask);
1089
1090 void (*rq_online)(struct rq *rq);
1091 void (*rq_offline)(struct rq *rq);
1092 #endif
1093
1094 void (*set_curr_task) (struct rq *rq);
1095 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1096 void (*task_fork) (struct task_struct *p);
1097
1098 void (*switched_from) (struct rq *this_rq, struct task_struct *task);
1099 void (*switched_to) (struct rq *this_rq, struct task_struct *task);
1100 void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1101 int oldprio);
1102
1103 unsigned int (*get_rr_interval) (struct rq *rq,
1104 struct task_struct *task);
1105
1106 #ifdef CONFIG_FAIR_GROUP_SCHED
1107 void (*task_move_group) (struct task_struct *p, int on_rq);
1108 #endif
1109 };
1110
1111 struct load_weight {
1112 unsigned long weight, inv_weight;
1113 };
1114
1115 #ifdef CONFIG_SCHEDSTATS
1116 struct sched_statistics {
1117 u64 wait_start;
1118 u64 wait_max;
1119 u64 wait_count;
1120 u64 wait_sum;
1121 u64 iowait_count;
1122 u64 iowait_sum;
1123
1124 u64 sleep_start;
1125 u64 sleep_max;
1126 s64 sum_sleep_runtime;
1127
1128 u64 block_start;
1129 u64 block_max;
1130 u64 exec_max;
1131 u64 slice_max;
1132
1133 u64 nr_migrations_cold;
1134 u64 nr_failed_migrations_affine;
1135 u64 nr_failed_migrations_running;
1136 u64 nr_failed_migrations_hot;
1137 u64 nr_forced_migrations;
1138
1139 u64 nr_wakeups;
1140 u64 nr_wakeups_sync;
1141 u64 nr_wakeups_migrate;
1142 u64 nr_wakeups_local;
1143 u64 nr_wakeups_remote;
1144 u64 nr_wakeups_affine;
1145 u64 nr_wakeups_affine_attempts;
1146 u64 nr_wakeups_passive;
1147 u64 nr_wakeups_idle;
1148 };
1149 #endif
1150
1151 struct sched_entity {
1152 struct load_weight load; /* for load-balancing */
1153 struct rb_node run_node;
1154 struct list_head group_node;
1155 unsigned int on_rq;
1156
1157 u64 exec_start;
1158 u64 sum_exec_runtime;
1159 u64 vruntime;
1160 u64 prev_sum_exec_runtime;
1161
1162 u64 nr_migrations;
1163
1164 #ifdef CONFIG_SCHEDSTATS
1165 struct sched_statistics statistics;
1166 #endif
1167
1168 #ifdef CONFIG_FAIR_GROUP_SCHED
1169 struct sched_entity *parent;
1170 /* rq on which this entity is (to be) queued: */
1171 struct cfs_rq *cfs_rq;
1172 /* rq "owned" by this entity/group: */
1173 struct cfs_rq *my_q;
1174 #endif
1175 };
1176
1177 struct sched_rt_entity {
1178 struct list_head run_list;
1179 unsigned long timeout;
1180 unsigned int time_slice;
1181 int nr_cpus_allowed;
1182
1183 struct sched_rt_entity *back;
1184 #ifdef CONFIG_RT_GROUP_SCHED
1185 struct sched_rt_entity *parent;
1186 /* rq on which this entity is (to be) queued: */
1187 struct rt_rq *rt_rq;
1188 /* rq "owned" by this entity/group: */
1189 struct rt_rq *my_q;
1190 #endif
1191 };
1192
1193 struct rcu_node;
1194
1195 enum perf_event_task_context {
1196 perf_invalid_context = -1,
1197 perf_hw_context = 0,
1198 perf_sw_context,
1199 perf_nr_task_contexts,
1200 };
1201
1202 struct task_struct {
1203 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1204 void *stack;
1205 atomic_t usage;
1206 unsigned int flags; /* per process flags, defined below */
1207 unsigned int ptrace;
1208
1209 #ifdef CONFIG_SMP
1210 struct task_struct *wake_entry;
1211 int on_cpu;
1212 #endif
1213 int on_rq;
1214
1215 int prio, static_prio, normal_prio;
1216 unsigned int rt_priority;
1217 const struct sched_class *sched_class;
1218 struct sched_entity se;
1219 struct sched_rt_entity rt;
1220
1221 #ifdef CONFIG_PREEMPT_NOTIFIERS
1222 /* list of struct preempt_notifier: */
1223 struct hlist_head preempt_notifiers;
1224 #endif
1225
1226 /*
1227 * fpu_counter contains the number of consecutive context switches
1228 * that the FPU is used. If this is over a threshold, the lazy fpu
1229 * saving becomes unlazy to save the trap. This is an unsigned char
1230 * so that after 256 times the counter wraps and the behavior turns
1231 * lazy again; this to deal with bursty apps that only use FPU for
1232 * a short time
1233 */
1234 unsigned char fpu_counter;
1235 #ifdef CONFIG_BLK_DEV_IO_TRACE
1236 unsigned int btrace_seq;
1237 #endif
1238
1239 unsigned int policy;
1240 cpumask_t cpus_allowed;
1241
1242 #ifdef CONFIG_PREEMPT_RCU
1243 int rcu_read_lock_nesting;
1244 char rcu_read_unlock_special;
1245 struct list_head rcu_node_entry;
1246 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1247 #ifdef CONFIG_TREE_PREEMPT_RCU
1248 struct rcu_node *rcu_blocked_node;
1249 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1250 #ifdef CONFIG_RCU_BOOST
1251 struct rt_mutex *rcu_boost_mutex;
1252 #endif /* #ifdef CONFIG_RCU_BOOST */
1253
1254 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1255 struct sched_info sched_info;
1256 #endif
1257
1258 struct list_head tasks;
1259 #ifdef CONFIG_SMP
1260 struct plist_node pushable_tasks;
1261 #endif
1262
1263 struct mm_struct *mm, *active_mm;
1264 #ifdef CONFIG_COMPAT_BRK
1265 unsigned brk_randomized:1;
1266 #endif
1267 #if defined(SPLIT_RSS_COUNTING)
1268 struct task_rss_stat rss_stat;
1269 #endif
1270 /* task state */
1271 int exit_state;
1272 int exit_code, exit_signal;
1273 int pdeath_signal; /* The signal sent when the parent dies */
1274 unsigned int group_stop; /* GROUP_STOP_*, siglock protected */
1275 /* ??? */
1276 unsigned int personality;
1277 unsigned did_exec:1;
1278 unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1279 * execve */
1280 unsigned in_iowait:1;
1281
1282
1283 /* Revert to default priority/policy when forking */
1284 unsigned sched_reset_on_fork:1;
1285 unsigned sched_contributes_to_load:1;
1286
1287 pid_t pid;
1288 pid_t tgid;
1289
1290 #ifdef CONFIG_CC_STACKPROTECTOR
1291 /* Canary value for the -fstack-protector gcc feature */
1292 unsigned long stack_canary;
1293 #endif
1294
1295 /*
1296 * pointers to (original) parent process, youngest child, younger sibling,
1297 * older sibling, respectively. (p->father can be replaced with
1298 * p->real_parent->pid)
1299 */
1300 struct task_struct *real_parent; /* real parent process */
1301 struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1302 /*
1303 * children/sibling forms the list of my natural children
1304 */
1305 struct list_head children; /* list of my children */
1306 struct list_head sibling; /* linkage in my parent's children list */
1307 struct task_struct *group_leader; /* threadgroup leader */
1308
1309 /*
1310 * ptraced is the list of tasks this task is using ptrace on.
1311 * This includes both natural children and PTRACE_ATTACH targets.
1312 * p->ptrace_entry is p's link on the p->parent->ptraced list.
1313 */
1314 struct list_head ptraced;
1315 struct list_head ptrace_entry;
1316
1317 /* PID/PID hash table linkage. */
1318 struct pid_link pids[PIDTYPE_MAX];
1319 struct list_head thread_group;
1320
1321 struct completion *vfork_done; /* for vfork() */
1322 int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1323 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1324
1325 cputime_t utime, stime, utimescaled, stimescaled;
1326 cputime_t gtime;
1327 #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1328 cputime_t prev_utime, prev_stime;
1329 #endif
1330 unsigned long nvcsw, nivcsw; /* context switch counts */
1331 struct timespec start_time; /* monotonic time */
1332 struct timespec real_start_time; /* boot based time */
1333 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1334 unsigned long min_flt, maj_flt;
1335
1336 struct task_cputime cputime_expires;
1337 struct list_head cpu_timers[3];
1338
1339 /* process credentials */
1340 const struct cred __rcu *real_cred; /* objective and real subjective task
1341 * credentials (COW) */
1342 const struct cred __rcu *cred; /* effective (overridable) subjective task
1343 * credentials (COW) */
1344 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1345
1346 char comm[TASK_COMM_LEN]; /* executable name excluding path
1347 - access with [gs]et_task_comm (which lock
1348 it with task_lock())
1349 - initialized normally by setup_new_exec */
1350 /* file system info */
1351 int link_count, total_link_count;
1352 #ifdef CONFIG_SYSVIPC
1353 /* ipc stuff */
1354 struct sysv_sem sysvsem;
1355 #endif
1356 #ifdef CONFIG_DETECT_HUNG_TASK
1357 /* hung task detection */
1358 unsigned long last_switch_count;
1359 #endif
1360 /* CPU-specific state of this task */
1361 struct thread_struct thread;
1362 /* filesystem information */
1363 struct fs_struct *fs;
1364 /* open file information */
1365 struct files_struct *files;
1366 /* namespaces */
1367 struct nsproxy *nsproxy;
1368 /* signal handlers */
1369 struct signal_struct *signal;
1370 struct sighand_struct *sighand;
1371
1372 sigset_t blocked, real_blocked;
1373 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1374 struct sigpending pending;
1375
1376 unsigned long sas_ss_sp;
1377 size_t sas_ss_size;
1378 int (*notifier)(void *priv);
1379 void *notifier_data;
1380 sigset_t *notifier_mask;
1381 struct audit_context *audit_context;
1382 #ifdef CONFIG_AUDITSYSCALL
1383 uid_t loginuid;
1384 unsigned int sessionid;
1385 #endif
1386 seccomp_t seccomp;
1387
1388 /* Thread group tracking */
1389 u32 parent_exec_id;
1390 u32 self_exec_id;
1391 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1392 * mempolicy */
1393 spinlock_t alloc_lock;
1394
1395 #ifdef CONFIG_GENERIC_HARDIRQS
1396 /* IRQ handler threads */
1397 struct irqaction *irqaction;
1398 #endif
1399
1400 /* Protection of the PI data structures: */
1401 raw_spinlock_t pi_lock;
1402
1403 #ifdef CONFIG_RT_MUTEXES
1404 /* PI waiters blocked on a rt_mutex held by this task */
1405 struct plist_head pi_waiters;
1406 /* Deadlock detection and priority inheritance handling */
1407 struct rt_mutex_waiter *pi_blocked_on;
1408 #endif
1409
1410 #ifdef CONFIG_DEBUG_MUTEXES
1411 /* mutex deadlock detection */
1412 struct mutex_waiter *blocked_on;
1413 #endif
1414 #ifdef CONFIG_TRACE_IRQFLAGS
1415 unsigned int irq_events;
1416 unsigned long hardirq_enable_ip;
1417 unsigned long hardirq_disable_ip;
1418 unsigned int hardirq_enable_event;
1419 unsigned int hardirq_disable_event;
1420 int hardirqs_enabled;
1421 int hardirq_context;
1422 unsigned long softirq_disable_ip;
1423 unsigned long softirq_enable_ip;
1424 unsigned int softirq_disable_event;
1425 unsigned int softirq_enable_event;
1426 int softirqs_enabled;
1427 int softirq_context;
1428 #endif
1429 #ifdef CONFIG_LOCKDEP
1430 # define MAX_LOCK_DEPTH 48UL
1431 u64 curr_chain_key;
1432 int lockdep_depth;
1433 unsigned int lockdep_recursion;
1434 struct held_lock held_locks[MAX_LOCK_DEPTH];
1435 gfp_t lockdep_reclaim_gfp;
1436 #endif
1437
1438 /* journalling filesystem info */
1439 void *journal_info;
1440
1441 /* stacked block device info */
1442 struct bio_list *bio_list;
1443
1444 #ifdef CONFIG_BLOCK
1445 /* stack plugging */
1446 struct blk_plug *plug;
1447 #endif
1448
1449 /* VM state */
1450 struct reclaim_state *reclaim_state;
1451
1452 struct backing_dev_info *backing_dev_info;
1453
1454 struct io_context *io_context;
1455
1456 unsigned long ptrace_message;
1457 siginfo_t *last_siginfo; /* For ptrace use. */
1458 struct task_io_accounting ioac;
1459 #if defined(CONFIG_TASK_XACCT)
1460 u64 acct_rss_mem1; /* accumulated rss usage */
1461 u64 acct_vm_mem1; /* accumulated virtual memory usage */
1462 cputime_t acct_timexpd; /* stime + utime since last update */
1463 #endif
1464 #ifdef CONFIG_CPUSETS
1465 nodemask_t mems_allowed; /* Protected by alloc_lock */
1466 int mems_allowed_change_disable;
1467 int cpuset_mem_spread_rotor;
1468 int cpuset_slab_spread_rotor;
1469 #endif
1470 #ifdef CONFIG_CGROUPS
1471 /* Control Group info protected by css_set_lock */
1472 struct css_set __rcu *cgroups;
1473 /* cg_list protected by css_set_lock and tsk->alloc_lock */
1474 struct list_head cg_list;
1475 #endif
1476 #ifdef CONFIG_FUTEX
1477 struct robust_list_head __user *robust_list;
1478 #ifdef CONFIG_COMPAT
1479 struct compat_robust_list_head __user *compat_robust_list;
1480 #endif
1481 struct list_head pi_state_list;
1482 struct futex_pi_state *pi_state_cache;
1483 #endif
1484 #ifdef CONFIG_PERF_EVENTS
1485 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1486 struct mutex perf_event_mutex;
1487 struct list_head perf_event_list;
1488 #endif
1489 #ifdef CONFIG_NUMA
1490 struct mempolicy *mempolicy; /* Protected by alloc_lock */
1491 short il_next;
1492 short pref_node_fork;
1493 #endif
1494 atomic_t fs_excl; /* holding fs exclusive resources */
1495 struct rcu_head rcu;
1496
1497 /*
1498 * cache last used pipe for splice
1499 */
1500 struct pipe_inode_info *splice_pipe;
1501 #ifdef CONFIG_TASK_DELAY_ACCT
1502 struct task_delay_info *delays;
1503 #endif
1504 #ifdef CONFIG_FAULT_INJECTION
1505 int make_it_fail;
1506 #endif
1507 struct prop_local_single dirties;
1508 #ifdef CONFIG_LATENCYTOP
1509 int latency_record_count;
1510 struct latency_record latency_record[LT_SAVECOUNT];
1511 #endif
1512 /*
1513 * time slack values; these are used to round up poll() and
1514 * select() etc timeout values. These are in nanoseconds.
1515 */
1516 unsigned long timer_slack_ns;
1517 unsigned long default_timer_slack_ns;
1518
1519 struct list_head *scm_work_list;
1520 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1521 /* Index of current stored address in ret_stack */
1522 int curr_ret_stack;
1523 /* Stack of return addresses for return function tracing */
1524 struct ftrace_ret_stack *ret_stack;
1525 /* time stamp for last schedule */
1526 unsigned long long ftrace_timestamp;
1527 /*
1528 * Number of functions that haven't been traced
1529 * because of depth overrun.
1530 */
1531 atomic_t trace_overrun;
1532 /* Pause for the tracing */
1533 atomic_t tracing_graph_pause;
1534 #endif
1535 #ifdef CONFIG_TRACING
1536 /* state flags for use by tracers */
1537 unsigned long trace;
1538 /* bitmask of trace recursion */
1539 unsigned long trace_recursion;
1540 #endif /* CONFIG_TRACING */
1541 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1542 struct memcg_batch_info {
1543 int do_batch; /* incremented when batch uncharge started */
1544 struct mem_cgroup *memcg; /* target memcg of uncharge */
1545 unsigned long nr_pages; /* uncharged usage */
1546 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */
1547 } memcg_batch;
1548 #endif
1549 #ifdef CONFIG_HAVE_HW_BREAKPOINT
1550 atomic_t ptrace_bp_refcnt;
1551 #endif
1552 };
1553
1554 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1555 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1556
1557 /*
1558 * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1559 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1560 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1561 * values are inverted: lower p->prio value means higher priority.
1562 *
1563 * The MAX_USER_RT_PRIO value allows the actual maximum
1564 * RT priority to be separate from the value exported to
1565 * user-space. This allows kernel threads to set their
1566 * priority to a value higher than any user task. Note:
1567 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1568 */
1569
1570 #define MAX_USER_RT_PRIO 100
1571 #define MAX_RT_PRIO MAX_USER_RT_PRIO
1572
1573 #define MAX_PRIO (MAX_RT_PRIO + 40)
1574 #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1575
1576 static inline int rt_prio(int prio)
1577 {
1578 if (unlikely(prio < MAX_RT_PRIO))
1579 return 1;
1580 return 0;
1581 }
1582
1583 static inline int rt_task(struct task_struct *p)
1584 {
1585 return rt_prio(p->prio);
1586 }
1587
1588 static inline struct pid *task_pid(struct task_struct *task)
1589 {
1590 return task->pids[PIDTYPE_PID].pid;
1591 }
1592
1593 static inline struct pid *task_tgid(struct task_struct *task)
1594 {
1595 return task->group_leader->pids[PIDTYPE_PID].pid;
1596 }
1597
1598 /*
1599 * Without tasklist or rcu lock it is not safe to dereference
1600 * the result of task_pgrp/task_session even if task == current,
1601 * we can race with another thread doing sys_setsid/sys_setpgid.
1602 */
1603 static inline struct pid *task_pgrp(struct task_struct *task)
1604 {
1605 return task->group_leader->pids[PIDTYPE_PGID].pid;
1606 }
1607
1608 static inline struct pid *task_session(struct task_struct *task)
1609 {
1610 return task->group_leader->pids[PIDTYPE_SID].pid;
1611 }
1612
1613 struct pid_namespace;
1614
1615 /*
1616 * the helpers to get the task's different pids as they are seen
1617 * from various namespaces
1618 *
1619 * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1620 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1621 * current.
1622 * task_xid_nr_ns() : id seen from the ns specified;
1623 *
1624 * set_task_vxid() : assigns a virtual id to a task;
1625 *
1626 * see also pid_nr() etc in include/linux/pid.h
1627 */
1628 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1629 struct pid_namespace *ns);
1630
1631 static inline pid_t task_pid_nr(struct task_struct *tsk)
1632 {
1633 return tsk->pid;
1634 }
1635
1636 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1637 struct pid_namespace *ns)
1638 {
1639 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1640 }
1641
1642 static inline pid_t task_pid_vnr(struct task_struct *tsk)
1643 {
1644 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1645 }
1646
1647
1648 static inline pid_t task_tgid_nr(struct task_struct *tsk)
1649 {
1650 return tsk->tgid;
1651 }
1652
1653 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1654
1655 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1656 {
1657 return pid_vnr(task_tgid(tsk));
1658 }
1659
1660
1661 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1662 struct pid_namespace *ns)
1663 {
1664 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1665 }
1666
1667 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1668 {
1669 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1670 }
1671
1672
1673 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1674 struct pid_namespace *ns)
1675 {
1676 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1677 }
1678
1679 static inline pid_t task_session_vnr(struct task_struct *tsk)
1680 {
1681 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1682 }
1683
1684 /* obsolete, do not use */
1685 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1686 {
1687 return task_pgrp_nr_ns(tsk, &init_pid_ns);
1688 }
1689
1690 /**
1691 * pid_alive - check that a task structure is not stale
1692 * @p: Task structure to be checked.
1693 *
1694 * Test if a process is not yet dead (at most zombie state)
1695 * If pid_alive fails, then pointers within the task structure
1696 * can be stale and must not be dereferenced.
1697 */
1698 static inline int pid_alive(struct task_struct *p)
1699 {
1700 return p->pids[PIDTYPE_PID].pid != NULL;
1701 }
1702
1703 /**
1704 * is_global_init - check if a task structure is init
1705 * @tsk: Task structure to be checked.
1706 *
1707 * Check if a task structure is the first user space task the kernel created.
1708 */
1709 static inline int is_global_init(struct task_struct *tsk)
1710 {
1711 return tsk->pid == 1;
1712 }
1713
1714 /*
1715 * is_container_init:
1716 * check whether in the task is init in its own pid namespace.
1717 */
1718 extern int is_container_init(struct task_struct *tsk);
1719
1720 extern struct pid *cad_pid;
1721
1722 extern void free_task(struct task_struct *tsk);
1723 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1724
1725 extern void __put_task_struct(struct task_struct *t);
1726
1727 static inline void put_task_struct(struct task_struct *t)
1728 {
1729 if (atomic_dec_and_test(&t->usage))
1730 __put_task_struct(t);
1731 }
1732
1733 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1734 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1735
1736 /*
1737 * Per process flags
1738 */
1739 #define PF_STARTING 0x00000002 /* being created */
1740 #define PF_EXITING 0x00000004 /* getting shut down */
1741 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1742 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1743 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */
1744 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1745 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1746 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1747 #define PF_DUMPCORE 0x00000200 /* dumped core */
1748 #define PF_SIGNALED 0x00000400 /* killed by a signal */
1749 #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1750 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1751 #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1752 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1753 #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1754 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1755 #define PF_KSWAPD 0x00040000 /* I am kswapd */
1756 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1757 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1758 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1759 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1760 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1761 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1762 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1763 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1764 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1765 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1766 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */
1767 #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1768
1769 /*
1770 * Only the _current_ task can read/write to tsk->flags, but other
1771 * tasks can access tsk->flags in readonly mode for example
1772 * with tsk_used_math (like during threaded core dumping).
1773 * There is however an exception to this rule during ptrace
1774 * or during fork: the ptracer task is allowed to write to the
1775 * child->flags of its traced child (same goes for fork, the parent
1776 * can write to the child->flags), because we're guaranteed the
1777 * child is not running and in turn not changing child->flags
1778 * at the same time the parent does it.
1779 */
1780 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1781 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1782 #define clear_used_math() clear_stopped_child_used_math(current)
1783 #define set_used_math() set_stopped_child_used_math(current)
1784 #define conditional_stopped_child_used_math(condition, child) \
1785 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1786 #define conditional_used_math(condition) \
1787 conditional_stopped_child_used_math(condition, current)
1788 #define copy_to_stopped_child_used_math(child) \
1789 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1790 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1791 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1792 #define used_math() tsk_used_math(current)
1793
1794 /*
1795 * task->group_stop flags
1796 */
1797 #define GROUP_STOP_SIGMASK 0xffff /* signr of the last group stop */
1798 #define GROUP_STOP_PENDING (1 << 16) /* task should stop for group stop */
1799 #define GROUP_STOP_CONSUME (1 << 17) /* consume group stop count */
1800 #define GROUP_STOP_TRAPPING (1 << 18) /* switching from STOPPED to TRACED */
1801 #define GROUP_STOP_DEQUEUED (1 << 19) /* stop signal dequeued */
1802
1803 extern void task_clear_group_stop_pending(struct task_struct *task);
1804
1805 #ifdef CONFIG_PREEMPT_RCU
1806
1807 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1808 #define RCU_READ_UNLOCK_BOOSTED (1 << 1) /* boosted while in RCU read-side. */
1809 #define RCU_READ_UNLOCK_NEED_QS (1 << 2) /* RCU core needs CPU response. */
1810
1811 static inline void rcu_copy_process(struct task_struct *p)
1812 {
1813 p->rcu_read_lock_nesting = 0;
1814 p->rcu_read_unlock_special = 0;
1815 #ifdef CONFIG_TREE_PREEMPT_RCU
1816 p->rcu_blocked_node = NULL;
1817 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1818 #ifdef CONFIG_RCU_BOOST
1819 p->rcu_boost_mutex = NULL;
1820 #endif /* #ifdef CONFIG_RCU_BOOST */
1821 INIT_LIST_HEAD(&p->rcu_node_entry);
1822 }
1823
1824 #else
1825
1826 static inline void rcu_copy_process(struct task_struct *p)
1827 {
1828 }
1829
1830 #endif
1831
1832 #ifdef CONFIG_SMP
1833 extern int set_cpus_allowed_ptr(struct task_struct *p,
1834 const struct cpumask *new_mask);
1835 #else
1836 static inline int set_cpus_allowed_ptr(struct task_struct *p,
1837 const struct cpumask *new_mask)
1838 {
1839 if (!cpumask_test_cpu(0, new_mask))
1840 return -EINVAL;
1841 return 0;
1842 }
1843 #endif
1844
1845 #ifndef CONFIG_CPUMASK_OFFSTACK
1846 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1847 {
1848 return set_cpus_allowed_ptr(p, &new_mask);
1849 }
1850 #endif
1851
1852 /*
1853 * Do not use outside of architecture code which knows its limitations.
1854 *
1855 * sched_clock() has no promise of monotonicity or bounded drift between
1856 * CPUs, use (which you should not) requires disabling IRQs.
1857 *
1858 * Please use one of the three interfaces below.
1859 */
1860 extern unsigned long long notrace sched_clock(void);
1861 /*
1862 * See the comment in kernel/sched_clock.c
1863 */
1864 extern u64 cpu_clock(int cpu);
1865 extern u64 local_clock(void);
1866 extern u64 sched_clock_cpu(int cpu);
1867
1868
1869 extern void sched_clock_init(void);
1870
1871 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1872 static inline void sched_clock_tick(void)
1873 {
1874 }
1875
1876 static inline void sched_clock_idle_sleep_event(void)
1877 {
1878 }
1879
1880 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1881 {
1882 }
1883 #else
1884 /*
1885 * Architectures can set this to 1 if they have specified
1886 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
1887 * but then during bootup it turns out that sched_clock()
1888 * is reliable after all:
1889 */
1890 extern int sched_clock_stable;
1891
1892 extern void sched_clock_tick(void);
1893 extern void sched_clock_idle_sleep_event(void);
1894 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1895 #endif
1896
1897 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
1898 /*
1899 * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
1900 * The reason for this explicit opt-in is not to have perf penalty with
1901 * slow sched_clocks.
1902 */
1903 extern void enable_sched_clock_irqtime(void);
1904 extern void disable_sched_clock_irqtime(void);
1905 #else
1906 static inline void enable_sched_clock_irqtime(void) {}
1907 static inline void disable_sched_clock_irqtime(void) {}
1908 #endif
1909
1910 extern unsigned long long
1911 task_sched_runtime(struct task_struct *task);
1912 extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1913
1914 /* sched_exec is called by processes performing an exec */
1915 #ifdef CONFIG_SMP
1916 extern void sched_exec(void);
1917 #else
1918 #define sched_exec() {}
1919 #endif
1920
1921 extern void sched_clock_idle_sleep_event(void);
1922 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1923
1924 #ifdef CONFIG_HOTPLUG_CPU
1925 extern void idle_task_exit(void);
1926 #else
1927 static inline void idle_task_exit(void) {}
1928 #endif
1929
1930 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1931 extern void wake_up_idle_cpu(int cpu);
1932 #else
1933 static inline void wake_up_idle_cpu(int cpu) { }
1934 #endif
1935
1936 extern unsigned int sysctl_sched_latency;
1937 extern unsigned int sysctl_sched_min_granularity;
1938 extern unsigned int sysctl_sched_wakeup_granularity;
1939 extern unsigned int sysctl_sched_child_runs_first;
1940
1941 enum sched_tunable_scaling {
1942 SCHED_TUNABLESCALING_NONE,
1943 SCHED_TUNABLESCALING_LOG,
1944 SCHED_TUNABLESCALING_LINEAR,
1945 SCHED_TUNABLESCALING_END,
1946 };
1947 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1948
1949 #ifdef CONFIG_SCHED_DEBUG
1950 extern unsigned int sysctl_sched_migration_cost;
1951 extern unsigned int sysctl_sched_nr_migrate;
1952 extern unsigned int sysctl_sched_time_avg;
1953 extern unsigned int sysctl_timer_migration;
1954 extern unsigned int sysctl_sched_shares_window;
1955
1956 int sched_proc_update_handler(struct ctl_table *table, int write,
1957 void __user *buffer, size_t *length,
1958 loff_t *ppos);
1959 #endif
1960 #ifdef CONFIG_SCHED_DEBUG
1961 static inline unsigned int get_sysctl_timer_migration(void)
1962 {
1963 return sysctl_timer_migration;
1964 }
1965 #else
1966 static inline unsigned int get_sysctl_timer_migration(void)
1967 {
1968 return 1;
1969 }
1970 #endif
1971 extern unsigned int sysctl_sched_rt_period;
1972 extern int sysctl_sched_rt_runtime;
1973
1974 int sched_rt_handler(struct ctl_table *table, int write,
1975 void __user *buffer, size_t *lenp,
1976 loff_t *ppos);
1977
1978 #ifdef CONFIG_SCHED_AUTOGROUP
1979 extern unsigned int sysctl_sched_autogroup_enabled;
1980
1981 extern void sched_autogroup_create_attach(struct task_struct *p);
1982 extern void sched_autogroup_detach(struct task_struct *p);
1983 extern void sched_autogroup_fork(struct signal_struct *sig);
1984 extern void sched_autogroup_exit(struct signal_struct *sig);
1985 #ifdef CONFIG_PROC_FS
1986 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
1987 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int *nice);
1988 #endif
1989 #else
1990 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
1991 static inline void sched_autogroup_detach(struct task_struct *p) { }
1992 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
1993 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
1994 #endif
1995
1996 #ifdef CONFIG_RT_MUTEXES
1997 extern int rt_mutex_getprio(struct task_struct *p);
1998 extern void rt_mutex_setprio(struct task_struct *p, int prio);
1999 extern void rt_mutex_adjust_pi(struct task_struct *p);
2000 #else
2001 static inline int rt_mutex_getprio(struct task_struct *p)
2002 {
2003 return p->normal_prio;
2004 }
2005 # define rt_mutex_adjust_pi(p) do { } while (0)
2006 #endif
2007
2008 extern bool yield_to(struct task_struct *p, bool preempt);
2009 extern void set_user_nice(struct task_struct *p, long nice);
2010 extern int task_prio(const struct task_struct *p);
2011 extern int task_nice(const struct task_struct *p);
2012 extern int can_nice(const struct task_struct *p, const int nice);
2013 extern int task_curr(const struct task_struct *p);
2014 extern int idle_cpu(int cpu);
2015 extern int sched_setscheduler(struct task_struct *, int,
2016 const struct sched_param *);
2017 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2018 const struct sched_param *);
2019 extern struct task_struct *idle_task(int cpu);
2020 extern struct task_struct *curr_task(int cpu);
2021 extern void set_curr_task(int cpu, struct task_struct *p);
2022
2023 void yield(void);
2024
2025 /*
2026 * The default (Linux) execution domain.
2027 */
2028 extern struct exec_domain default_exec_domain;
2029
2030 union thread_union {
2031 struct thread_info thread_info;
2032 unsigned long stack[THREAD_SIZE/sizeof(long)];
2033 };
2034
2035 #ifndef __HAVE_ARCH_KSTACK_END
2036 static inline int kstack_end(void *addr)
2037 {
2038 /* Reliable end of stack detection:
2039 * Some APM bios versions misalign the stack
2040 */
2041 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2042 }
2043 #endif
2044
2045 extern union thread_union init_thread_union;
2046 extern struct task_struct init_task;
2047
2048 extern struct mm_struct init_mm;
2049
2050 extern struct pid_namespace init_pid_ns;
2051
2052 /*
2053 * find a task by one of its numerical ids
2054 *
2055 * find_task_by_pid_ns():
2056 * finds a task by its pid in the specified namespace
2057 * find_task_by_vpid():
2058 * finds a task by its virtual pid
2059 *
2060 * see also find_vpid() etc in include/linux/pid.h
2061 */
2062
2063 extern struct task_struct *find_task_by_vpid(pid_t nr);
2064 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2065 struct pid_namespace *ns);
2066
2067 extern void __set_special_pids(struct pid *pid);
2068
2069 /* per-UID process charging. */
2070 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2071 static inline struct user_struct *get_uid(struct user_struct *u)
2072 {
2073 atomic_inc(&u->__count);
2074 return u;
2075 }
2076 extern void free_uid(struct user_struct *);
2077 extern void release_uids(struct user_namespace *ns);
2078
2079 #include <asm/current.h>
2080
2081 extern void xtime_update(unsigned long ticks);
2082
2083 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2084 extern int wake_up_process(struct task_struct *tsk);
2085 extern void wake_up_new_task(struct task_struct *tsk);
2086 #ifdef CONFIG_SMP
2087 extern void kick_process(struct task_struct *tsk);
2088 #else
2089 static inline void kick_process(struct task_struct *tsk) { }
2090 #endif
2091 extern void sched_fork(struct task_struct *p);
2092 extern void sched_dead(struct task_struct *p);
2093
2094 extern void proc_caches_init(void);
2095 extern void flush_signals(struct task_struct *);
2096 extern void __flush_signals(struct task_struct *);
2097 extern void ignore_signals(struct task_struct *);
2098 extern void flush_signal_handlers(struct task_struct *, int force_default);
2099 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2100
2101 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2102 {
2103 unsigned long flags;
2104 int ret;
2105
2106 spin_lock_irqsave(&tsk->sighand->siglock, flags);
2107 ret = dequeue_signal(tsk, mask, info);
2108 spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2109
2110 return ret;
2111 }
2112
2113 extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2114 sigset_t *mask);
2115 extern void unblock_all_signals(void);
2116 extern void release_task(struct task_struct * p);
2117 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2118 extern int force_sigsegv(int, struct task_struct *);
2119 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2120 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2121 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2122 extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2123 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2124 extern int kill_pid(struct pid *pid, int sig, int priv);
2125 extern int kill_proc_info(int, struct siginfo *, pid_t);
2126 extern int do_notify_parent(struct task_struct *, int);
2127 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2128 extern void force_sig(int, struct task_struct *);
2129 extern int send_sig(int, struct task_struct *, int);
2130 extern int zap_other_threads(struct task_struct *p);
2131 extern struct sigqueue *sigqueue_alloc(void);
2132 extern void sigqueue_free(struct sigqueue *);
2133 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2134 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2135 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2136
2137 static inline int kill_cad_pid(int sig, int priv)
2138 {
2139 return kill_pid(cad_pid, sig, priv);
2140 }
2141
2142 /* These can be the second arg to send_sig_info/send_group_sig_info. */
2143 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2144 #define SEND_SIG_PRIV ((struct siginfo *) 1)
2145 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2146
2147 /*
2148 * True if we are on the alternate signal stack.
2149 */
2150 static inline int on_sig_stack(unsigned long sp)
2151 {
2152 #ifdef CONFIG_STACK_GROWSUP
2153 return sp >= current->sas_ss_sp &&
2154 sp - current->sas_ss_sp < current->sas_ss_size;
2155 #else
2156 return sp > current->sas_ss_sp &&
2157 sp - current->sas_ss_sp <= current->sas_ss_size;
2158 #endif
2159 }
2160
2161 static inline int sas_ss_flags(unsigned long sp)
2162 {
2163 return (current->sas_ss_size == 0 ? SS_DISABLE
2164 : on_sig_stack(sp) ? SS_ONSTACK : 0);
2165 }
2166
2167 /*
2168 * Routines for handling mm_structs
2169 */
2170 extern struct mm_struct * mm_alloc(void);
2171
2172 /* mmdrop drops the mm and the page tables */
2173 extern void __mmdrop(struct mm_struct *);
2174 static inline void mmdrop(struct mm_struct * mm)
2175 {
2176 if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2177 __mmdrop(mm);
2178 }
2179 extern int mm_init_cpumask(struct mm_struct *mm, struct mm_struct *oldmm);
2180
2181 /* mmput gets rid of the mappings and all user-space */
2182 extern void mmput(struct mm_struct *);
2183 /* Grab a reference to a task's mm, if it is not already going away */
2184 extern struct mm_struct *get_task_mm(struct task_struct *task);
2185 /* Remove the current tasks stale references to the old mm_struct */
2186 extern void mm_release(struct task_struct *, struct mm_struct *);
2187 /* Allocate a new mm structure and copy contents from tsk->mm */
2188 extern struct mm_struct *dup_mm(struct task_struct *tsk);
2189
2190 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2191 struct task_struct *, struct pt_regs *);
2192 extern void flush_thread(void);
2193 extern void exit_thread(void);
2194
2195 extern void exit_files(struct task_struct *);
2196 extern void __cleanup_sighand(struct sighand_struct *);
2197
2198 extern void exit_itimers(struct signal_struct *);
2199 extern void flush_itimer_signals(void);
2200
2201 extern NORET_TYPE void do_group_exit(int);
2202
2203 extern void daemonize(const char *, ...);
2204 extern int allow_signal(int);
2205 extern int disallow_signal(int);
2206
2207 extern int do_execve(const char *,
2208 const char __user * const __user *,
2209 const char __user * const __user *, struct pt_regs *);
2210 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2211 struct task_struct *fork_idle(int);
2212
2213 extern void set_task_comm(struct task_struct *tsk, char *from);
2214 extern char *get_task_comm(char *to, struct task_struct *tsk);
2215
2216 #ifdef CONFIG_SMP
2217 void scheduler_ipi(void);
2218 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2219 #else
2220 static inline void scheduler_ipi(void) { }
2221 static inline unsigned long wait_task_inactive(struct task_struct *p,
2222 long match_state)
2223 {
2224 return 1;
2225 }
2226 #endif
2227
2228 #define next_task(p) \
2229 list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2230
2231 #define for_each_process(p) \
2232 for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2233
2234 extern bool current_is_single_threaded(void);
2235
2236 /*
2237 * Careful: do_each_thread/while_each_thread is a double loop so
2238 * 'break' will not work as expected - use goto instead.
2239 */
2240 #define do_each_thread(g, t) \
2241 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2242
2243 #define while_each_thread(g, t) \
2244 while ((t = next_thread(t)) != g)
2245
2246 static inline int get_nr_threads(struct task_struct *tsk)
2247 {
2248 return tsk->signal->nr_threads;
2249 }
2250
2251 /* de_thread depends on thread_group_leader not being a pid based check */
2252 #define thread_group_leader(p) (p == p->group_leader)
2253
2254 /* Do to the insanities of de_thread it is possible for a process
2255 * to have the pid of the thread group leader without actually being
2256 * the thread group leader. For iteration through the pids in proc
2257 * all we care about is that we have a task with the appropriate
2258 * pid, we don't actually care if we have the right task.
2259 */
2260 static inline int has_group_leader_pid(struct task_struct *p)
2261 {
2262 return p->pid == p->tgid;
2263 }
2264
2265 static inline
2266 int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2267 {
2268 return p1->tgid == p2->tgid;
2269 }
2270
2271 static inline struct task_struct *next_thread(const struct task_struct *p)
2272 {
2273 return list_entry_rcu(p->thread_group.next,
2274 struct task_struct, thread_group);
2275 }
2276
2277 static inline int thread_group_empty(struct task_struct *p)
2278 {
2279 return list_empty(&p->thread_group);
2280 }
2281
2282 #define delay_group_leader(p) \
2283 (thread_group_leader(p) && !thread_group_empty(p))
2284
2285 static inline int task_detached(struct task_struct *p)
2286 {
2287 return p->exit_signal == -1;
2288 }
2289
2290 /*
2291 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2292 * subscriptions and synchronises with wait4(). Also used in procfs. Also
2293 * pins the final release of task.io_context. Also protects ->cpuset and
2294 * ->cgroup.subsys[].
2295 *
2296 * Nests both inside and outside of read_lock(&tasklist_lock).
2297 * It must not be nested with write_lock_irq(&tasklist_lock),
2298 * neither inside nor outside.
2299 */
2300 static inline void task_lock(struct task_struct *p)
2301 {
2302 spin_lock(&p->alloc_lock);
2303 }
2304
2305 static inline void task_unlock(struct task_struct *p)
2306 {
2307 spin_unlock(&p->alloc_lock);
2308 }
2309
2310 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
2311 unsigned long *flags);
2312
2313 #define lock_task_sighand(tsk, flags) \
2314 ({ struct sighand_struct *__ss; \
2315 __cond_lock(&(tsk)->sighand->siglock, \
2316 (__ss = __lock_task_sighand(tsk, flags))); \
2317 __ss; \
2318 }) \
2319
2320 static inline void unlock_task_sighand(struct task_struct *tsk,
2321 unsigned long *flags)
2322 {
2323 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2324 }
2325
2326 #ifndef __HAVE_THREAD_FUNCTIONS
2327
2328 #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2329 #define task_stack_page(task) ((task)->stack)
2330
2331 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2332 {
2333 *task_thread_info(p) = *task_thread_info(org);
2334 task_thread_info(p)->task = p;
2335 }
2336
2337 static inline unsigned long *end_of_stack(struct task_struct *p)
2338 {
2339 return (unsigned long *)(task_thread_info(p) + 1);
2340 }
2341
2342 #endif
2343
2344 static inline int object_is_on_stack(void *obj)
2345 {
2346 void *stack = task_stack_page(current);
2347
2348 return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2349 }
2350
2351 extern void thread_info_cache_init(void);
2352
2353 #ifdef CONFIG_DEBUG_STACK_USAGE
2354 static inline unsigned long stack_not_used(struct task_struct *p)
2355 {
2356 unsigned long *n = end_of_stack(p);
2357
2358 do { /* Skip over canary */
2359 n++;
2360 } while (!*n);
2361
2362 return (unsigned long)n - (unsigned long)end_of_stack(p);
2363 }
2364 #endif
2365
2366 /* set thread flags in other task's structures
2367 * - see asm/thread_info.h for TIF_xxxx flags available
2368 */
2369 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2370 {
2371 set_ti_thread_flag(task_thread_info(tsk), flag);
2372 }
2373
2374 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2375 {
2376 clear_ti_thread_flag(task_thread_info(tsk), flag);
2377 }
2378
2379 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2380 {
2381 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2382 }
2383
2384 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2385 {
2386 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2387 }
2388
2389 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2390 {
2391 return test_ti_thread_flag(task_thread_info(tsk), flag);
2392 }
2393
2394 static inline void set_tsk_need_resched(struct task_struct *tsk)
2395 {
2396 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2397 }
2398
2399 static inline void clear_tsk_need_resched(struct task_struct *tsk)
2400 {
2401 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2402 }
2403
2404 static inline int test_tsk_need_resched(struct task_struct *tsk)
2405 {
2406 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2407 }
2408
2409 static inline int restart_syscall(void)
2410 {
2411 set_tsk_thread_flag(current, TIF_SIGPENDING);
2412 return -ERESTARTNOINTR;
2413 }
2414
2415 static inline int signal_pending(struct task_struct *p)
2416 {
2417 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2418 }
2419
2420 static inline int __fatal_signal_pending(struct task_struct *p)
2421 {
2422 return unlikely(sigismember(&p->pending.signal, SIGKILL));
2423 }
2424
2425 static inline int fatal_signal_pending(struct task_struct *p)
2426 {
2427 return signal_pending(p) && __fatal_signal_pending(p);
2428 }
2429
2430 static inline int signal_pending_state(long state, struct task_struct *p)
2431 {
2432 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2433 return 0;
2434 if (!signal_pending(p))
2435 return 0;
2436
2437 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2438 }
2439
2440 static inline int need_resched(void)
2441 {
2442 return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2443 }
2444
2445 /*
2446 * cond_resched() and cond_resched_lock(): latency reduction via
2447 * explicit rescheduling in places that are safe. The return
2448 * value indicates whether a reschedule was done in fact.
2449 * cond_resched_lock() will drop the spinlock before scheduling,
2450 * cond_resched_softirq() will enable bhs before scheduling.
2451 */
2452 extern int _cond_resched(void);
2453
2454 #define cond_resched() ({ \
2455 __might_sleep(__FILE__, __LINE__, 0); \
2456 _cond_resched(); \
2457 })
2458
2459 extern int __cond_resched_lock(spinlock_t *lock);
2460
2461 #ifdef CONFIG_PREEMPT
2462 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2463 #else
2464 #define PREEMPT_LOCK_OFFSET 0
2465 #endif
2466
2467 #define cond_resched_lock(lock) ({ \
2468 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2469 __cond_resched_lock(lock); \
2470 })
2471
2472 extern int __cond_resched_softirq(void);
2473
2474 #define cond_resched_softirq() ({ \
2475 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \
2476 __cond_resched_softirq(); \
2477 })
2478
2479 /*
2480 * Does a critical section need to be broken due to another
2481 * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2482 * but a general need for low latency)
2483 */
2484 static inline int spin_needbreak(spinlock_t *lock)
2485 {
2486 #ifdef CONFIG_PREEMPT
2487 return spin_is_contended(lock);
2488 #else
2489 return 0;
2490 #endif
2491 }
2492
2493 /*
2494 * Thread group CPU time accounting.
2495 */
2496 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2497 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2498
2499 static inline void thread_group_cputime_init(struct signal_struct *sig)
2500 {
2501 spin_lock_init(&sig->cputimer.lock);
2502 }
2503
2504 /*
2505 * Reevaluate whether the task has signals pending delivery.
2506 * Wake the task if so.
2507 * This is required every time the blocked sigset_t changes.
2508 * callers must hold sighand->siglock.
2509 */
2510 extern void recalc_sigpending_and_wake(struct task_struct *t);
2511 extern void recalc_sigpending(void);
2512
2513 extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2514
2515 /*
2516 * Wrappers for p->thread_info->cpu access. No-op on UP.
2517 */
2518 #ifdef CONFIG_SMP
2519
2520 static inline unsigned int task_cpu(const struct task_struct *p)
2521 {
2522 return task_thread_info(p)->cpu;
2523 }
2524
2525 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2526
2527 #else
2528
2529 static inline unsigned int task_cpu(const struct task_struct *p)
2530 {
2531 return 0;
2532 }
2533
2534 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2535 {
2536 }
2537
2538 #endif /* CONFIG_SMP */
2539
2540 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2541 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2542
2543 extern void normalize_rt_tasks(void);
2544
2545 #ifdef CONFIG_CGROUP_SCHED
2546
2547 extern struct task_group root_task_group;
2548
2549 extern struct task_group *sched_create_group(struct task_group *parent);
2550 extern void sched_destroy_group(struct task_group *tg);
2551 extern void sched_move_task(struct task_struct *tsk);
2552 #ifdef CONFIG_FAIR_GROUP_SCHED
2553 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2554 extern unsigned long sched_group_shares(struct task_group *tg);
2555 #endif
2556 #ifdef CONFIG_RT_GROUP_SCHED
2557 extern int sched_group_set_rt_runtime(struct task_group *tg,
2558 long rt_runtime_us);
2559 extern long sched_group_rt_runtime(struct task_group *tg);
2560 extern int sched_group_set_rt_period(struct task_group *tg,
2561 long rt_period_us);
2562 extern long sched_group_rt_period(struct task_group *tg);
2563 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2564 #endif
2565 #endif
2566
2567 extern int task_can_switch_user(struct user_struct *up,
2568 struct task_struct *tsk);
2569
2570 #ifdef CONFIG_TASK_XACCT
2571 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2572 {
2573 tsk->ioac.rchar += amt;
2574 }
2575
2576 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2577 {
2578 tsk->ioac.wchar += amt;
2579 }
2580
2581 static inline void inc_syscr(struct task_struct *tsk)
2582 {
2583 tsk->ioac.syscr++;
2584 }
2585
2586 static inline void inc_syscw(struct task_struct *tsk)
2587 {
2588 tsk->ioac.syscw++;
2589 }
2590 #else
2591 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2592 {
2593 }
2594
2595 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2596 {
2597 }
2598
2599 static inline void inc_syscr(struct task_struct *tsk)
2600 {
2601 }
2602
2603 static inline void inc_syscw(struct task_struct *tsk)
2604 {
2605 }
2606 #endif
2607
2608 #ifndef TASK_SIZE_OF
2609 #define TASK_SIZE_OF(tsk) TASK_SIZE
2610 #endif
2611
2612 #ifdef CONFIG_MM_OWNER
2613 extern void mm_update_next_owner(struct mm_struct *mm);
2614 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2615 #else
2616 static inline void mm_update_next_owner(struct mm_struct *mm)
2617 {
2618 }
2619
2620 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2621 {
2622 }
2623 #endif /* CONFIG_MM_OWNER */
2624
2625 static inline unsigned long task_rlimit(const struct task_struct *tsk,
2626 unsigned int limit)
2627 {
2628 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2629 }
2630
2631 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2632 unsigned int limit)
2633 {
2634 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2635 }
2636
2637 static inline unsigned long rlimit(unsigned int limit)
2638 {
2639 return task_rlimit(current, limit);
2640 }
2641
2642 static inline unsigned long rlimit_max(unsigned int limit)
2643 {
2644 return task_rlimit_max(current, limit);
2645 }
2646
2647 #endif /* __KERNEL__ */
2648
2649 #endif
Linux kernel - Deliverables
This page took 0.229919 seconds and 6 git commands to generate.