Line data Source code
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 : #define CLONE_STOPPED 0x02000000 /* Start in stopped state */
25 : #define CLONE_NEWUTS 0x04000000 /* New utsname group? */
26 : #define CLONE_NEWIPC 0x08000000 /* New ipcs */
27 : #define CLONE_NEWUSER 0x10000000 /* New user namespace */
28 : #define CLONE_NEWPID 0x20000000 /* New pid namespace */
29 : #define CLONE_NEWNET 0x40000000 /* New network namespace */
30 : #define CLONE_IO 0x80000000 /* Clone io context */
31 :
32 : /*
33 : * Scheduling policies
34 : */
35 : #define SCHED_NORMAL 0
36 : #define SCHED_FIFO 1
37 : #define SCHED_RR 2
38 : #define SCHED_BATCH 3
39 : /* SCHED_ISO: reserved but not implemented yet */
40 : #define SCHED_IDLE 5
41 : /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */
42 : #define SCHED_RESET_ON_FORK 0x40000000
43 :
44 : #ifdef __KERNEL__
45 :
46 : struct sched_param {
47 : int sched_priority;
48 : };
49 :
50 : #include <asm/param.h> /* for HZ */
51 :
52 : #include <linux/capability.h>
53 : #include <linux/threads.h>
54 : #include <linux/kernel.h>
55 : #include <linux/types.h>
56 : #include <linux/timex.h>
57 : #include <linux/jiffies.h>
58 : #include <linux/rbtree.h>
59 : #include <linux/thread_info.h>
60 : #include <linux/cpumask.h>
61 : #include <linux/errno.h>
62 : #include <linux/nodemask.h>
63 : #include <linux/mm_types.h>
64 :
65 : #include <asm/system.h>
66 : #include <asm/page.h>
67 : #include <asm/ptrace.h>
68 : #include <asm/cputime.h>
69 :
70 : #include <linux/smp.h>
71 : #include <linux/sem.h>
72 : #include <linux/signal.h>
73 : #include <linux/path.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/kobject.h>
92 : #include <linux/latencytop.h>
93 : #include <linux/cred.h>
94 :
95 : #include <asm/processor.h>
96 :
97 : struct exec_domain;
98 : struct futex_pi_state;
99 : struct robust_list_head;
100 : struct bio;
101 : struct fs_struct;
102 : struct bts_context;
103 : struct perf_event_context;
104 :
105 : /*
106 : * List of flags we want to share for kernel threads,
107 : * if only because they are not used by them anyway.
108 : */
109 : #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND)
110 :
111 : /*
112 : * These are the constant used to fake the fixed-point load-average
113 : * counting. Some notes:
114 : * - 11 bit fractions expand to 22 bits by the multiplies: this gives
115 : * a load-average precision of 10 bits integer + 11 bits fractional
116 : * - if you want to count load-averages more often, you need more
117 : * precision, or rounding will get you. With 2-second counting freq,
118 : * the EXP_n values would be 1981, 2034 and 2043 if still using only
119 : * 11 bit fractions.
120 : */
121 : extern unsigned long avenrun[]; /* Load averages */
122 : extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
123 :
124 : #define FSHIFT 11 /* nr of bits of precision */
125 : #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */
126 : #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */
127 : #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */
128 : #define EXP_5 2014 /* 1/exp(5sec/5min) */
129 : #define EXP_15 2037 /* 1/exp(5sec/15min) */
130 :
131 : #define CALC_LOAD(load,exp,n) \
132 : load *= exp; \
133 : load += n*(FIXED_1-exp); \
134 : load >>= FSHIFT;
135 :
136 : extern unsigned long total_forks;
137 : extern int nr_threads;
138 : DECLARE_PER_CPU(unsigned long, process_counts);
139 : extern int nr_processes(void);
140 : extern unsigned long nr_running(void);
141 : extern unsigned long nr_uninterruptible(void);
142 : extern unsigned long nr_iowait(void);
143 : extern unsigned long nr_iowait_cpu(void);
144 : extern unsigned long this_cpu_load(void);
145 :
146 :
147 : extern void calc_global_load(void);
148 :
149 : extern unsigned long get_parent_ip(unsigned long addr);
150 :
151 : struct seq_file;
152 : struct cfs_rq;
153 : struct task_group;
154 : #ifdef CONFIG_SCHED_DEBUG
155 : extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
156 : extern void proc_sched_set_task(struct task_struct *p);
157 : extern void
158 : print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq);
159 : #else
160 : static inline void
161 : proc_sched_show_task(struct task_struct *p, struct seq_file *m)
162 : {
163 : }
164 : static inline void proc_sched_set_task(struct task_struct *p)
165 : {
166 : }
167 : static inline void
168 : print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq)
169 : {
170 : }
171 : #endif
172 :
173 : /*
174 : * Task state bitmask. NOTE! These bits are also
175 : * encoded in fs/proc/array.c: get_task_state().
176 : *
177 : * We have two separate sets of flags: task->state
178 : * is about runnability, while task->exit_state are
179 : * about the task exiting. Confusing, but this way
180 : * modifying one set can't modify the other one by
181 : * mistake.
182 : */
183 : #define TASK_RUNNING 0
184 : #define TASK_INTERRUPTIBLE 1
185 : #define TASK_UNINTERRUPTIBLE 2
186 : #define __TASK_STOPPED 4
187 : #define __TASK_TRACED 8
188 : /* in tsk->exit_state */
189 : #define EXIT_ZOMBIE 16
190 : #define EXIT_DEAD 32
191 : /* in tsk->state again */
192 : #define TASK_DEAD 64
193 : #define TASK_WAKEKILL 128
194 : #define TASK_WAKING 256
195 : #define TASK_STATE_MAX 512
196 :
197 : #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW"
198 :
199 : extern char ___assert_task_state[1 - 2*!!(
200 : sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
201 :
202 : /* Convenience macros for the sake of set_task_state */
203 : #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
204 : #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED)
205 : #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED)
206 :
207 : /* Convenience macros for the sake of wake_up */
208 : #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
209 : #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
210 :
211 : /* get_task_state() */
212 : #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \
213 : TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
214 : __TASK_TRACED)
215 :
216 : #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0)
217 : #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 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 : extern void sched_init(void);
262 : extern void sched_init_smp(void);
263 : extern asmlinkage void schedule_tail(struct task_struct *prev);
264 : extern void init_idle(struct task_struct *idle, int cpu);
265 : extern void init_idle_bootup_task(struct task_struct *idle);
266 :
267 : extern int runqueue_is_locked(int cpu);
268 : extern void task_rq_unlock_wait(struct task_struct *p);
269 :
270 : extern cpumask_var_t nohz_cpu_mask;
271 : #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ)
272 : extern int select_nohz_load_balancer(int cpu);
273 : extern int get_nohz_load_balancer(void);
274 : #else
275 : static inline int select_nohz_load_balancer(int cpu)
276 : {
277 : return 0;
278 : }
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_DETECT_SOFTLOCKUP
311 : extern void softlockup_tick(void);
312 : extern void touch_softlockup_watchdog(void);
313 : extern void touch_softlockup_watchdog_sync(void);
314 : extern void touch_all_softlockup_watchdogs(void);
315 : extern int proc_dosoftlockup_thresh(struct ctl_table *table, int write,
316 : void __user *buffer,
317 : size_t *lenp, loff_t *ppos);
318 : extern unsigned int softlockup_panic;
319 : extern int softlockup_thresh;
320 : #else
321 : static inline void softlockup_tick(void)
322 : {
323 : }
324 : static inline void touch_softlockup_watchdog(void)
325 : {
326 : }
327 : static inline void touch_softlockup_watchdog_sync(void)
328 : {
329 : }
330 : static inline void touch_all_softlockup_watchdogs(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 : #endif
344 :
345 : /* Attach to any functions which should be ignored in wchan output. */
346 : #define __sched __attribute__((__section__(".sched.text")))
347 :
348 : /* Linker adds these: start and end of __sched functions */
349 : extern char __sched_text_start[], __sched_text_end[];
350 :
351 : /* Is this address in the __sched functions? */
352 : extern int in_sched_functions(unsigned long addr);
353 :
354 : #define MAX_SCHEDULE_TIMEOUT LONG_MAX
355 : extern signed long schedule_timeout(signed long timeout);
356 : extern signed long schedule_timeout_interruptible(signed long timeout);
357 : extern signed long schedule_timeout_killable(signed long timeout);
358 : extern signed long schedule_timeout_uninterruptible(signed long timeout);
359 : asmlinkage void schedule(void);
360 : extern int mutex_spin_on_owner(struct mutex *lock, struct thread_info *owner);
361 1 :
362 : struct nsproxy;
363 : struct user_namespace;
364 :
365 : /*
366 : * Default maximum number of active map areas, this limits the number of vmas
367 : * per mm struct. Users can overwrite this number by sysctl but there is a
368 : * problem.
369 : *
370 : * When a program's coredump is generated as ELF format, a section is created
371 : * per a vma. In ELF, the number of sections is represented in unsigned short.
372 : * This means the number of sections should be smaller than 65535 at coredump.
373 : * Because the kernel adds some informative sections to a image of program at
374 : * generating coredump, we need some margin. The number of extra sections is
375 : * 1-3 now and depends on arch. We use "5" as safe margin, here.
376 : */
377 : #define MAPCOUNT_ELF_CORE_MARGIN (5)
378 : #define DEFAULT_MAX_MAP_COUNT (USHORT_MAX - MAPCOUNT_ELF_CORE_MARGIN)
379 :
380 : extern int sysctl_max_map_count;
381 :
382 : #include <linux/aio.h>
383 :
384 : #ifdef CONFIG_MMU
385 : extern void arch_pick_mmap_layout(struct mm_struct *mm);
386 : extern unsigned long
387 : arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
388 : unsigned long, unsigned long);
389 : extern unsigned long
390 : arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
391 : unsigned long len, unsigned long pgoff,
392 : unsigned long flags);
393 : extern void arch_unmap_area(struct mm_struct *, unsigned long);
394 : extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long);
395 : #else
396 : static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
397 : #endif
398 :
399 : #if USE_SPLIT_PTLOCKS
400 : /*
401 : * The mm counters are not protected by its page_table_lock,
402 : * so must be incremented atomically.
403 : */
404 : #define set_mm_counter(mm, member, value) atomic_long_set(&(mm)->_##member, value)
405 : #define get_mm_counter(mm, member) ((unsigned long)atomic_long_read(&(mm)->_##member))
406 : #define add_mm_counter(mm, member, value) atomic_long_add(value, &(mm)->_##member)
407 : #define inc_mm_counter(mm, member) atomic_long_inc(&(mm)->_##member)
408 : #define dec_mm_counter(mm, member) atomic_long_dec(&(mm)->_##member)
409 :
410 : #else /* !USE_SPLIT_PTLOCKS */
411 : /*
412 : * The mm counters are protected by its page_table_lock,
413 : * so can be incremented directly.
414 : */
415 : #define set_mm_counter(mm, member, value) (mm)->_##member = (value)
416 : #define get_mm_counter(mm, member) ((mm)->_##member)
417 : #define add_mm_counter(mm, member, value) (mm)->_##member += (value)
418 : #define inc_mm_counter(mm, member) (mm)->_##member++
419 : #define dec_mm_counter(mm, member) (mm)->_##member--
420 :
421 : #endif /* !USE_SPLIT_PTLOCKS */
422 :
423 : #define get_mm_rss(mm) \
424 : (get_mm_counter(mm, file_rss) + get_mm_counter(mm, anon_rss))
425 : #define update_hiwater_rss(mm) do { \
426 : unsigned long _rss = get_mm_rss(mm); \
427 : if ((mm)->hiwater_rss < _rss) \
428 : (mm)->hiwater_rss = _rss; \
429 : } while (0)
430 : #define update_hiwater_vm(mm) do { \
431 : if ((mm)->hiwater_vm < (mm)->total_vm) \
432 : (mm)->hiwater_vm = (mm)->total_vm; \
433 : } while (0)
434 :
435 : static inline unsigned long get_mm_hiwater_rss(struct mm_struct *mm)
436 : {
437 : return max(mm->hiwater_rss, get_mm_rss(mm));
438 : }
439 :
440 : static inline void setmax_mm_hiwater_rss(unsigned long *maxrss,
441 : struct mm_struct *mm)
442 : {
443 : unsigned long hiwater_rss = get_mm_hiwater_rss(mm);
444 :
445 : if (*maxrss < hiwater_rss)
446 : *maxrss = hiwater_rss;
447 : }
448 :
449 : static inline unsigned long get_mm_hiwater_vm(struct mm_struct *mm)
450 : {
451 : return max(mm->hiwater_vm, mm->total_vm);
452 : }
453 :
454 : extern void set_dumpable(struct mm_struct *mm, int value);
455 : extern int get_dumpable(struct mm_struct *mm);
456 1 :
457 : /* mm flags */
458 : /* dumpable bits */
459 : #define MMF_DUMPABLE 0 /* core dump is permitted */
460 : #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */
461 :
462 : #define MMF_DUMPABLE_BITS 2
463 : #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
464 :
465 : /* coredump filter bits */
466 : #define MMF_DUMP_ANON_PRIVATE 2
467 : #define MMF_DUMP_ANON_SHARED 3
468 : #define MMF_DUMP_MAPPED_PRIVATE 4
469 : #define MMF_DUMP_MAPPED_SHARED 5
470 : #define MMF_DUMP_ELF_HEADERS 6
471 : #define MMF_DUMP_HUGETLB_PRIVATE 7
472 : #define MMF_DUMP_HUGETLB_SHARED 8
473 :
474 : #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS
475 : #define MMF_DUMP_FILTER_BITS 7
476 : #define MMF_DUMP_FILTER_MASK \
477 : (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
478 : #define MMF_DUMP_FILTER_DEFAULT \
479 : ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
480 : (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
481 :
482 : #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
483 : # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS)
484 : #else
485 : # define MMF_DUMP_MASK_DEFAULT_ELF 0
486 : #endif
487 : /* leave room for more dump flags */
488 : #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */
489 :
490 : #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
491 :
492 : struct sighand_struct {
493 : atomic_t count;
494 : struct k_sigaction action[_NSIG];
495 : spinlock_t siglock;
496 : wait_queue_head_t signalfd_wqh;
497 : };
498 :
499 : struct pacct_struct {
500 : int ac_flag;
501 : long ac_exitcode;
502 : unsigned long ac_mem;
503 : cputime_t ac_utime, ac_stime;
504 : unsigned long ac_minflt, ac_majflt;
505 : };
506 1 :
507 : struct cpu_itimer {
508 : cputime_t expires;
509 : cputime_t incr;
510 : u32 error;
511 : u32 incr_error;
512 : };
513 1 :
514 : /**
515 : * struct task_cputime - collected CPU time counts
516 : * @utime: time spent in user mode, in &cputime_t units
517 : * @stime: time spent in kernel mode, in &cputime_t units
518 : * @sum_exec_runtime: total time spent on the CPU, in nanoseconds
519 : *
520 : * This structure groups together three kinds of CPU time that are
521 : * tracked for threads and thread groups. Most things considering
522 : * CPU time want to group these counts together and treat all three
523 : * of them in parallel.
524 : */
525 : struct task_cputime {
526 : cputime_t utime;
527 : cputime_t stime;
528 : unsigned long long sum_exec_runtime;
529 : };
530 1 : /* Alternate field names when used to cache expirations. */
531 : #define prof_exp stime
532 : #define virt_exp utime
533 : #define sched_exp sum_exec_runtime
534 :
535 : #define INIT_CPUTIME \
536 : (struct task_cputime) { \
537 : .utime = cputime_zero, \
538 : .stime = cputime_zero, \
539 : .sum_exec_runtime = 0, \
540 : }
541 :
542 : /*
543 : * Disable preemption until the scheduler is running.
544 : * Reset by start_kernel()->sched_init()->init_idle().
545 : *
546 : * We include PREEMPT_ACTIVE to avoid cond_resched() from working
547 : * before the scheduler is active -- see should_resched().
548 : */
549 : #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE)
550 :
551 : /**
552 : * struct thread_group_cputimer - thread group interval timer counts
553 : * @cputime: thread group interval timers.
554 : * @running: non-zero when there are timers running and
555 : * @cputime receives updates.
556 : * @lock: lock for fields in this struct.
557 : *
558 : * This structure contains the version of task_cputime, above, that is
559 : * used for thread group CPU timer calculations.
560 : */
561 : struct thread_group_cputimer {
562 : struct task_cputime cputime;
563 : int running;
564 : spinlock_t lock;
565 : };
566 2 :
567 : /*
568 : * NOTE! "signal_struct" does not have it's own
569 : * locking, because a shared signal_struct always
570 : * implies a shared sighand_struct, so locking
571 : * sighand_struct is always a proper superset of
572 : * the locking of signal_struct.
573 : */
574 : struct signal_struct {
575 : atomic_t count;
576 : atomic_t live;
577 :
578 : wait_queue_head_t wait_chldexit; /* for wait4() */
579 :
580 : /* current thread group signal load-balancing target: */
581 : struct task_struct *curr_target;
582 :
583 : /* shared signal handling: */
584 : struct sigpending shared_pending;
585 :
586 : /* thread group exit support */
587 : int group_exit_code;
588 : /* overloaded:
589 : * - notify group_exit_task when ->count is equal to notify_count
590 : * - everyone except group_exit_task is stopped during signal delivery
591 : * of fatal signals, group_exit_task processes the signal.
592 : */
593 : int notify_count;
594 : struct task_struct *group_exit_task;
595 :
596 : /* thread group stop support, overloads group_exit_code too */
597 : int group_stop_count;
598 : unsigned int flags; /* see SIGNAL_* flags below */
599 :
600 : /* POSIX.1b Interval Timers */
601 : struct list_head posix_timers;
602 :
603 : /* ITIMER_REAL timer for the process */
604 : struct hrtimer real_timer;
605 : struct pid *leader_pid;
606 : ktime_t it_real_incr;
607 :
608 : /*
609 : * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
610 : * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
611 : * values are defined to 0 and 1 respectively
612 : */
613 : struct cpu_itimer it[2];
614 :
615 : /*
616 : * Thread group totals for process CPU timers.
617 : * See thread_group_cputimer(), et al, for details.
618 : */
619 : struct thread_group_cputimer cputimer;
620 :
621 : /* Earliest-expiration cache. */
622 : struct task_cputime cputime_expires;
623 :
624 : struct list_head cpu_timers[3];
625 :
626 : struct pid *tty_old_pgrp;
627 :
628 : /* boolean value for session group leader */
629 : int leader;
630 :
631 : struct tty_struct *tty; /* NULL if no tty */
632 :
633 : /*
634 : * Cumulative resource counters for dead threads in the group,
635 : * and for reaped dead child processes forked by this group.
636 : * Live threads maintain their own counters and add to these
637 : * in __exit_signal, except for the group leader.
638 : */
639 : cputime_t utime, stime, cutime, cstime;
640 : cputime_t gtime;
641 : cputime_t cgtime;
642 : #ifndef CONFIG_VIRT_CPU_ACCOUNTING
643 : cputime_t prev_utime, prev_stime;
644 : #endif
645 : unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
646 : unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
647 : unsigned long inblock, oublock, cinblock, coublock;
648 : unsigned long maxrss, cmaxrss;
649 : struct task_io_accounting ioac;
650 :
651 : /*
652 : * Cumulative ns of schedule CPU time fo dead threads in the
653 : * group, not including a zombie group leader, (This only differs
654 : * from jiffies_to_ns(utime + stime) if sched_clock uses something
655 : * other than jiffies.)
656 : */
657 : unsigned long long sum_sched_runtime;
658 :
659 : /*
660 : * We don't bother to synchronize most readers of this at all,
661 : * because there is no reader checking a limit that actually needs
662 : * to get both rlim_cur and rlim_max atomically, and either one
663 : * alone is a single word that can safely be read normally.
664 : * getrlimit/setrlimit use task_lock(current->group_leader) to
665 : * protect this instead of the siglock, because they really
666 : * have no need to disable irqs.
667 : */
668 : struct rlimit rlim[RLIM_NLIMITS];
669 :
670 : #ifdef CONFIG_BSD_PROCESS_ACCT
671 : struct pacct_struct pacct; /* per-process accounting information */
672 : #endif
673 : #ifdef CONFIG_TASKSTATS
674 : struct taskstats *stats;
675 : #endif
676 : #ifdef CONFIG_AUDIT
677 : unsigned audit_tty;
678 : struct tty_audit_buf *tty_audit_buf;
679 : #endif
680 :
681 : int oom_adj; /* OOM kill score adjustment (bit shift) */
682 : };
683 :
684 : /* Context switch must be unlocked if interrupts are to be enabled */
685 : #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW
686 : # define __ARCH_WANT_UNLOCKED_CTXSW
687 : #endif
688 :
689 : /*
690 : * Bits in flags field of signal_struct.
691 : */
692 : #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */
693 : #define SIGNAL_STOP_DEQUEUED 0x00000002 /* stop signal dequeued */
694 : #define SIGNAL_STOP_CONTINUED 0x00000004 /* SIGCONT since WCONTINUED reap */
695 : #define SIGNAL_GROUP_EXIT 0x00000008 /* group exit in progress */
696 : /*
697 : * Pending notifications to parent.
698 : */
699 : #define SIGNAL_CLD_STOPPED 0x00000010
700 : #define SIGNAL_CLD_CONTINUED 0x00000020
701 : #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
702 :
703 : #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */
704 :
705 : /* If true, all threads except ->group_exit_task have pending SIGKILL */
706 : static inline int signal_group_exit(const struct signal_struct *sig)
707 : {
708 : return (sig->flags & SIGNAL_GROUP_EXIT) ||
709 : (sig->group_exit_task != NULL);
710 : }
711 1 :
712 : /*
713 : * Some day this will be a full-fledged user tracking system..
714 : */
715 : struct user_struct {
716 : atomic_t __count; /* reference count */
717 : atomic_t processes; /* How many processes does this user have? */
718 : atomic_t files; /* How many open files does this user have? */
719 : atomic_t sigpending; /* How many pending signals does this user have? */
720 : #ifdef CONFIG_INOTIFY_USER
721 : atomic_t inotify_watches; /* How many inotify watches does this user have? */
722 : atomic_t inotify_devs; /* How many inotify devs does this user have opened? */
723 : #endif
724 : #ifdef CONFIG_EPOLL
725 : atomic_t epoll_watches; /* The number of file descriptors currently watched */
726 : #endif
727 : #ifdef CONFIG_POSIX_MQUEUE
728 : /* protected by mq_lock */
729 : unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
730 : #endif
731 : unsigned long locked_shm; /* How many pages of mlocked shm ? */
732 :
733 : #ifdef CONFIG_KEYS
734 : struct key *uid_keyring; /* UID specific keyring */
735 : struct key *session_keyring; /* UID's default session keyring */
736 : #endif
737 :
738 : /* Hash table maintenance information */
739 : struct hlist_node uidhash_node;
740 : uid_t uid;
741 : struct user_namespace *user_ns;
742 :
743 : #ifdef CONFIG_USER_SCHED
744 : struct task_group *tg;
745 : #ifdef CONFIG_SYSFS
746 : struct kobject kobj;
747 : struct delayed_work work;
748 : #endif
749 : #endif
750 :
751 : #ifdef CONFIG_PERF_EVENTS
752 : atomic_long_t locked_vm;
753 : #endif
754 : };
755 :
756 : extern int uids_sysfs_init(void);
757 :
758 : extern struct user_struct *find_user(uid_t);
759 :
760 : extern struct user_struct root_user;
761 1 : #define INIT_USER (&root_user)
762 :
763 :
764 : struct backing_dev_info;
765 : struct reclaim_state;
766 :
767 : #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
768 : struct sched_info {
769 : /* cumulative counters */
770 : unsigned long pcount; /* # of times run on this cpu */
771 : unsigned long long run_delay; /* time spent waiting on a runqueue */
772 :
773 : /* timestamps */
774 : unsigned long long last_arrival,/* when we last ran on a cpu */
775 : last_queued; /* when we were last queued to run */
776 : #ifdef CONFIG_SCHEDSTATS
777 : /* BKL stats */
778 : unsigned int bkl_count;
779 : #endif
780 : };
781 : #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */
782 :
783 : #ifdef CONFIG_TASK_DELAY_ACCT
784 : struct task_delay_info {
785 : spinlock_t lock;
786 : unsigned int flags; /* Private per-task flags */
787 :
788 : /* For each stat XXX, add following, aligned appropriately
789 : *
790 : * struct timespec XXX_start, XXX_end;
791 : * u64 XXX_delay;
792 : * u32 XXX_count;
793 : *
794 : * Atomicity of updates to XXX_delay, XXX_count protected by
795 : * single lock above (split into XXX_lock if contention is an issue).
796 : */
797 :
798 : /*
799 : * XXX_count is incremented on every XXX operation, the delay
800 : * associated with the operation is added to XXX_delay.
801 : * XXX_delay contains the accumulated delay time in nanoseconds.
802 : */
803 : struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */
804 : u64 blkio_delay; /* wait for sync block io completion */
805 : u64 swapin_delay; /* wait for swapin block io completion */
806 : u32 blkio_count; /* total count of the number of sync block */
807 : /* io operations performed */
808 : u32 swapin_count; /* total count of the number of swapin block */
809 : /* io operations performed */
810 :
811 : struct timespec freepages_start, freepages_end;
812 : u64 freepages_delay; /* wait for memory reclaim */
813 : u32 freepages_count; /* total count of memory reclaim */
814 : };
815 : #endif /* CONFIG_TASK_DELAY_ACCT */
816 :
817 : static inline int sched_info_on(void)
818 : {
819 : #ifdef CONFIG_SCHEDSTATS
820 : return 1;
821 : #elif defined(CONFIG_TASK_DELAY_ACCT)
822 : extern int delayacct_on;
823 : return delayacct_on;
824 : #else
825 : return 0;
826 : #endif
827 1 : }
828 :
829 : enum cpu_idle_type {
830 : CPU_IDLE,
831 : CPU_NOT_IDLE,
832 : CPU_NEWLY_IDLE,
833 : CPU_MAX_IDLE_TYPES
834 : };
835 :
836 : /*
837 : * sched-domains (multiprocessor balancing) declarations:
838 : */
839 :
840 : /*
841 : * Increase resolution of nice-level calculations:
842 : */
843 : #define SCHED_LOAD_SHIFT 10
844 : #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT)
845 :
846 : #define SCHED_LOAD_SCALE_FUZZ SCHED_LOAD_SCALE
847 :
848 : #ifdef CONFIG_SMP
849 : #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */
850 : #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */
851 : #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */
852 : #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */
853 : #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */
854 : #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */
855 : #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */
856 : #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */
857 : #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */
858 : #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */
859 : #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */
860 :
861 : #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */
862 :
863 : enum powersavings_balance_level {
864 : POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */
865 : POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package
866 : * first for long running threads
867 : */
868 : POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle
869 : * cpu package for power savings
870 : */
871 : MAX_POWERSAVINGS_BALANCE_LEVELS
872 : };
873 :
874 : extern int sched_mc_power_savings, sched_smt_power_savings;
875 :
876 : static inline int sd_balance_for_mc_power(void)
877 : {
878 : if (sched_smt_power_savings)
879 : return SD_POWERSAVINGS_BALANCE;
880 :
881 : if (!sched_mc_power_savings)
882 : return SD_PREFER_SIBLING;
883 :
884 : return 0;
885 : }
886 :
887 : static inline int sd_balance_for_package_power(void)
888 : {
889 : if (sched_mc_power_savings | sched_smt_power_savings)
890 : return SD_POWERSAVINGS_BALANCE;
891 :
892 : return SD_PREFER_SIBLING;
893 : }
894 :
895 : /*
896 : * Optimise SD flags for power savings:
897 : * SD_BALANCE_NEWIDLE helps agressive task consolidation and power savings.
898 : * Keep default SD flags if sched_{smt,mc}_power_saving=0
899 : */
900 :
901 : static inline int sd_power_saving_flags(void)
902 : {
903 : if (sched_mc_power_savings | sched_smt_power_savings)
904 : return SD_BALANCE_NEWIDLE;
905 :
906 : return 0;
907 : }
908 1 :
909 : struct sched_group {
910 : struct sched_group *next; /* Must be a circular list */
911 :
912 : /*
913 : * CPU power of this group, SCHED_LOAD_SCALE being max power for a
914 : * single CPU.
915 : */
916 : unsigned int cpu_power;
917 :
918 : /*
919 : * The CPUs this group covers.
920 : *
921 : * NOTE: this field is variable length. (Allocated dynamically
922 : * by attaching extra space to the end of the structure,
923 : * depending on how many CPUs the kernel has booted up with)
924 : *
925 : * It is also be embedded into static data structures at build
926 : * time. (See 'struct static_sched_group' in kernel/sched.c)
927 : */
928 : unsigned long cpumask[0];
929 : };
930 :
931 : static inline struct cpumask *sched_group_cpus(struct sched_group *sg)
932 : {
933 : return to_cpumask(sg->cpumask);
934 : }
935 1 :
936 : enum sched_domain_level {
937 : SD_LV_NONE = 0,
938 : SD_LV_SIBLING,
939 : SD_LV_MC,
940 : SD_LV_CPU,
941 : SD_LV_NODE,
942 : SD_LV_ALLNODES,
943 : SD_LV_MAX
944 : };
945 :
946 : struct sched_domain_attr {
947 : int relax_domain_level;
948 : };
949 1 :
950 : #define SD_ATTR_INIT (struct sched_domain_attr) { \
951 : .relax_domain_level = -1, \
952 : }
953 :
954 : struct sched_domain {
955 : /* These fields must be setup */
956 : struct sched_domain *parent; /* top domain must be null terminated */
957 : struct sched_domain *child; /* bottom domain must be null terminated */
958 : struct sched_group *groups; /* the balancing groups of the domain */
959 : unsigned long min_interval; /* Minimum balance interval ms */
960 : unsigned long max_interval; /* Maximum balance interval ms */
961 : unsigned int busy_factor; /* less balancing by factor if busy */
962 : unsigned int imbalance_pct; /* No balance until over watermark */
963 : unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */
964 : unsigned int busy_idx;
965 : unsigned int idle_idx;
966 : unsigned int newidle_idx;
967 : unsigned int wake_idx;
968 : unsigned int forkexec_idx;
969 : unsigned int smt_gain;
970 : int flags; /* See SD_* */
971 : enum sched_domain_level level;
972 :
973 : /* Runtime fields. */
974 : unsigned long last_balance; /* init to jiffies. units in jiffies */
975 : unsigned int balance_interval; /* initialise to 1. units in ms. */
976 : unsigned int nr_balance_failed; /* initialise to 0 */
977 :
978 : u64 last_update;
979 :
980 : #ifdef CONFIG_SCHEDSTATS
981 : /* load_balance() stats */
982 : unsigned int lb_count[CPU_MAX_IDLE_TYPES];
983 : unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
984 : unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
985 : unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
986 : unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
987 : unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
988 : unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
989 : unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
990 :
991 : /* Active load balancing */
992 : unsigned int alb_count;
993 : unsigned int alb_failed;
994 : unsigned int alb_pushed;
995 :
996 : /* SD_BALANCE_EXEC stats */
997 : unsigned int sbe_count;
998 : unsigned int sbe_balanced;
999 : unsigned int sbe_pushed;
1000 :
1001 : /* SD_BALANCE_FORK stats */
1002 : unsigned int sbf_count;
1003 : unsigned int sbf_balanced;
1004 : unsigned int sbf_pushed;
1005 :
1006 : /* try_to_wake_up() stats */
1007 : unsigned int ttwu_wake_remote;
1008 : unsigned int ttwu_move_affine;
1009 : unsigned int ttwu_move_balance;
1010 : #endif
1011 : #ifdef CONFIG_SCHED_DEBUG
1012 : char *name;
1013 : #endif
1014 :
1015 : /*
1016 : * Span of all CPUs in this domain.
1017 : *
1018 : * NOTE: this field is variable length. (Allocated dynamically
1019 : * by attaching extra space to the end of the structure,
1020 : * depending on how many CPUs the kernel has booted up with)
1021 : *
1022 : * It is also be embedded into static data structures at build
1023 : * time. (See 'struct static_sched_domain' in kernel/sched.c)
1024 : */
1025 : unsigned long span[0];
1026 : };
1027 :
1028 : static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1029 : {
1030 : return to_cpumask(sd->span);
1031 : }
1032 :
1033 : extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1034 : struct sched_domain_attr *dattr_new);
1035 :
1036 : /* Allocate an array of sched domains, for partition_sched_domains(). */
1037 : cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1038 : void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1039 :
1040 : /* Test a flag in parent sched domain */
1041 : static inline int test_sd_parent(struct sched_domain *sd, int flag)
1042 : {
1043 : if (sd->parent && (sd->parent->flags & flag))
1044 : return 1;
1045 :
1046 : return 0;
1047 : }
1048 :
1049 : unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu);
1050 : unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu);
1051 1 :
1052 : #else /* CONFIG_SMP */
1053 :
1054 : struct sched_domain_attr;
1055 :
1056 : static inline void
1057 : partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1058 : struct sched_domain_attr *dattr_new)
1059 : {
1060 : }
1061 : #endif /* !CONFIG_SMP */
1062 :
1063 :
1064 : struct io_context; /* See blkdev.h */
1065 :
1066 :
1067 : #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1068 : extern void prefetch_stack(struct task_struct *t);
1069 : #else
1070 : static inline void prefetch_stack(struct task_struct *t) { }
1071 : #endif
1072 :
1073 : struct audit_context; /* See audit.c */
1074 1 : struct mempolicy;
1075 1 : struct pipe_inode_info;
1076 : struct uts_namespace;
1077 :
1078 : struct rq;
1079 : struct sched_domain;
1080 :
1081 : /*
1082 : * wake flags
1083 : */
1084 : #define WF_SYNC 0x01 /* waker goes to sleep after wakup */
1085 : #define WF_FORK 0x02 /* child wakeup after fork */
1086 :
1087 : struct sched_class {
1088 : const struct sched_class *next;
1089 :
1090 : void (*enqueue_task) (struct rq *rq, struct task_struct *p, int wakeup);
1091 : void (*dequeue_task) (struct rq *rq, struct task_struct *p, int sleep);
1092 : void (*yield_task) (struct rq *rq);
1093 :
1094 : void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags);
1095 :
1096 : struct task_struct * (*pick_next_task) (struct rq *rq);
1097 : void (*put_prev_task) (struct rq *rq, struct task_struct *p);
1098 :
1099 : #ifdef CONFIG_SMP
1100 : int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags);
1101 :
1102 : unsigned long (*load_balance) (struct rq *this_rq, int this_cpu,
1103 : struct rq *busiest, unsigned long max_load_move,
1104 : struct sched_domain *sd, enum cpu_idle_type idle,
1105 : int *all_pinned, int *this_best_prio);
1106 :
1107 : int (*move_one_task) (struct rq *this_rq, int this_cpu,
1108 : struct rq *busiest, struct sched_domain *sd,
1109 : enum cpu_idle_type idle);
1110 : void (*pre_schedule) (struct rq *this_rq, struct task_struct *task);
1111 : void (*post_schedule) (struct rq *this_rq);
1112 : void (*task_waking) (struct rq *this_rq, struct task_struct *task);
1113 : void (*task_woken) (struct rq *this_rq, struct task_struct *task);
1114 :
1115 : void (*set_cpus_allowed)(struct task_struct *p,
1116 : const struct cpumask *newmask);
1117 :
1118 : void (*rq_online)(struct rq *rq);
1119 : void (*rq_offline)(struct rq *rq);
1120 : #endif
1121 :
1122 : void (*set_curr_task) (struct rq *rq);
1123 : void (*task_tick) (struct rq *rq, struct task_struct *p, int queued);
1124 : void (*task_fork) (struct task_struct *p);
1125 :
1126 : void (*switched_from) (struct rq *this_rq, struct task_struct *task,
1127 : int running);
1128 : void (*switched_to) (struct rq *this_rq, struct task_struct *task,
1129 : int running);
1130 : void (*prio_changed) (struct rq *this_rq, struct task_struct *task,
1131 : int oldprio, int running);
1132 :
1133 : unsigned int (*get_rr_interval) (struct rq *rq,
1134 : struct task_struct *task);
1135 1 :
1136 : #ifdef CONFIG_FAIR_GROUP_SCHED
1137 : void (*moved_group) (struct task_struct *p, int on_rq);
1138 : #endif
1139 : };
1140 :
1141 : struct load_weight {
1142 : unsigned long weight, inv_weight;
1143 : };
1144 1 :
1145 : /*
1146 : * CFS stats for a schedulable entity (task, task-group etc)
1147 : *
1148 : * Current field usage histogram:
1149 : *
1150 : * 4 se->block_start
1151 : * 4 se->run_node
1152 : * 4 se->sleep_start
1153 : * 6 se->load.weight
1154 : */
1155 : struct sched_entity {
1156 : struct load_weight load; /* for load-balancing */
1157 : struct rb_node run_node;
1158 : struct list_head group_node;
1159 : unsigned int on_rq;
1160 :
1161 : u64 exec_start;
1162 : u64 sum_exec_runtime;
1163 : u64 vruntime;
1164 : u64 prev_sum_exec_runtime;
1165 :
1166 : u64 last_wakeup;
1167 : u64 avg_overlap;
1168 :
1169 : u64 nr_migrations;
1170 :
1171 : u64 start_runtime;
1172 : u64 avg_wakeup;
1173 :
1174 1 : #ifdef CONFIG_SCHEDSTATS
1175 : u64 wait_start;
1176 : u64 wait_max;
1177 : u64 wait_count;
1178 : u64 wait_sum;
1179 : u64 iowait_count;
1180 : u64 iowait_sum;
1181 :
1182 : u64 sleep_start;
1183 : u64 sleep_max;
1184 : s64 sum_sleep_runtime;
1185 :
1186 : u64 block_start;
1187 : u64 block_max;
1188 : u64 exec_max;
1189 : u64 slice_max;
1190 :
1191 : u64 nr_migrations_cold;
1192 : u64 nr_failed_migrations_affine;
1193 : u64 nr_failed_migrations_running;
1194 : u64 nr_failed_migrations_hot;
1195 : u64 nr_forced_migrations;
1196 :
1197 : u64 nr_wakeups;
1198 : u64 nr_wakeups_sync;
1199 : u64 nr_wakeups_migrate;
1200 : u64 nr_wakeups_local;
1201 : u64 nr_wakeups_remote;
1202 : u64 nr_wakeups_affine;
1203 : u64 nr_wakeups_affine_attempts;
1204 : u64 nr_wakeups_passive;
1205 : u64 nr_wakeups_idle;
1206 : #endif
1207 :
1208 : #ifdef CONFIG_FAIR_GROUP_SCHED
1209 : struct sched_entity *parent;
1210 : /* rq on which this entity is (to be) queued: */
1211 : struct cfs_rq *cfs_rq;
1212 : /* rq "owned" by this entity/group: */
1213 : struct cfs_rq *my_q;
1214 : #endif
1215 : };
1216 :
1217 : struct sched_rt_entity {
1218 : struct list_head run_list;
1219 : unsigned long timeout;
1220 : unsigned int time_slice;
1221 : int nr_cpus_allowed;
1222 :
1223 : struct sched_rt_entity *back;
1224 : #ifdef CONFIG_RT_GROUP_SCHED
1225 : struct sched_rt_entity *parent;
1226 2 : /* rq on which this entity is (to be) queued: */
1227 : struct rt_rq *rt_rq;
1228 : /* rq "owned" by this entity/group: */
1229 : struct rt_rq *my_q;
1230 : #endif
1231 : };
1232 :
1233 : struct rcu_node;
1234 :
1235 : struct task_struct {
1236 : volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */
1237 : void *stack;
1238 : atomic_t usage;
1239 : unsigned int flags; /* per process flags, defined below */
1240 : unsigned int ptrace;
1241 :
1242 : int lock_depth; /* BKL lock depth */
1243 :
1244 : #ifdef CONFIG_SMP
1245 : #ifdef __ARCH_WANT_UNLOCKED_CTXSW
1246 : int oncpu;
1247 : #endif
1248 : #endif
1249 :
1250 : int prio, static_prio, normal_prio;
1251 : unsigned int rt_priority;
1252 : const struct sched_class *sched_class;
1253 : struct sched_entity se;
1254 : struct sched_rt_entity rt;
1255 :
1256 : #ifdef CONFIG_PREEMPT_NOTIFIERS
1257 : /* list of struct preempt_notifier: */
1258 : struct hlist_head preempt_notifiers;
1259 : #endif
1260 :
1261 : /*
1262 : * fpu_counter contains the number of consecutive context switches
1263 : * that the FPU is used. If this is over a threshold, the lazy fpu
1264 : * saving becomes unlazy to save the trap. This is an unsigned char
1265 : * so that after 256 times the counter wraps and the behavior turns
1266 : * lazy again; this to deal with bursty apps that only use FPU for
1267 : * a short time
1268 : */
1269 : unsigned char fpu_counter;
1270 : #ifdef CONFIG_BLK_DEV_IO_TRACE
1271 : unsigned int btrace_seq;
1272 : #endif
1273 :
1274 : unsigned int policy;
1275 : cpumask_t cpus_allowed;
1276 :
1277 : #ifdef CONFIG_TREE_PREEMPT_RCU
1278 : int rcu_read_lock_nesting;
1279 : char rcu_read_unlock_special;
1280 : struct rcu_node *rcu_blocked_node;
1281 : struct list_head rcu_node_entry;
1282 : #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */
1283 :
1284 : #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT)
1285 : struct sched_info sched_info;
1286 : #endif
1287 :
1288 : struct list_head tasks;
1289 : struct plist_node pushable_tasks;
1290 :
1291 : struct mm_struct *mm, *active_mm;
1292 :
1293 : /* task state */
1294 : int exit_state;
1295 : int exit_code, exit_signal;
1296 : int pdeath_signal; /* The signal sent when the parent dies */
1297 : /* ??? */
1298 : unsigned int personality;
1299 : unsigned did_exec:1;
1300 : unsigned in_execve:1; /* Tell the LSMs that the process is doing an
1301 : * execve */
1302 : unsigned in_iowait:1;
1303 :
1304 :
1305 : /* Revert to default priority/policy when forking */
1306 : unsigned sched_reset_on_fork:1;
1307 :
1308 : pid_t pid;
1309 : pid_t tgid;
1310 :
1311 : #ifdef CONFIG_CC_STACKPROTECTOR
1312 : /* Canary value for the -fstack-protector gcc feature */
1313 : unsigned long stack_canary;
1314 : #endif
1315 :
1316 : /*
1317 : * pointers to (original) parent process, youngest child, younger sibling,
1318 : * older sibling, respectively. (p->father can be replaced with
1319 : * p->real_parent->pid)
1320 : */
1321 : struct task_struct *real_parent; /* real parent process */
1322 : struct task_struct *parent; /* recipient of SIGCHLD, wait4() reports */
1323 : /*
1324 : * children/sibling forms the list of my natural children
1325 : */
1326 : struct list_head children; /* list of my children */
1327 : struct list_head sibling; /* linkage in my parent's children list */
1328 : struct task_struct *group_leader; /* threadgroup leader */
1329 :
1330 : /*
1331 : * ptraced is the list of tasks this task is using ptrace on.
1332 : * This includes both natural children and PTRACE_ATTACH targets.
1333 : * p->ptrace_entry is p's link on the p->parent->ptraced list.
1334 : */
1335 : struct list_head ptraced;
1336 : struct list_head ptrace_entry;
1337 :
1338 : /*
1339 : * This is the tracer handle for the ptrace BTS extension.
1340 : * This field actually belongs to the ptracer task.
1341 : */
1342 : struct bts_context *bts;
1343 :
1344 : /* PID/PID hash table linkage. */
1345 : struct pid_link pids[PIDTYPE_MAX];
1346 : struct list_head thread_group;
1347 :
1348 : struct completion *vfork_done; /* for vfork() */
1349 : int __user *set_child_tid; /* CLONE_CHILD_SETTID */
1350 : int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */
1351 :
1352 : cputime_t utime, stime, utimescaled, stimescaled;
1353 : cputime_t gtime;
1354 : #ifndef CONFIG_VIRT_CPU_ACCOUNTING
1355 : cputime_t prev_utime, prev_stime;
1356 : #endif
1357 : unsigned long nvcsw, nivcsw; /* context switch counts */
1358 : struct timespec start_time; /* monotonic time */
1359 : struct timespec real_start_time; /* boot based time */
1360 : /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1361 : unsigned long min_flt, maj_flt;
1362 :
1363 : struct task_cputime cputime_expires;
1364 : struct list_head cpu_timers[3];
1365 :
1366 : /* process credentials */
1367 : const struct cred *real_cred; /* objective and real subjective task
1368 : * credentials (COW) */
1369 : const struct cred *cred; /* effective (overridable) subjective task
1370 : * credentials (COW) */
1371 : struct mutex cred_guard_mutex; /* guard against foreign influences on
1372 : * credential calculations
1373 : * (notably. ptrace) */
1374 : struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */
1375 :
1376 : char comm[TASK_COMM_LEN]; /* executable name excluding path
1377 : - access with [gs]et_task_comm (which lock
1378 : it with task_lock())
1379 : - initialized normally by setup_new_exec */
1380 : /* file system info */
1381 : int link_count, total_link_count;
1382 : #ifdef CONFIG_SYSVIPC
1383 : /* ipc stuff */
1384 : struct sysv_sem sysvsem;
1385 : #endif
1386 : #ifdef CONFIG_DETECT_HUNG_TASK
1387 : /* hung task detection */
1388 : unsigned long last_switch_count;
1389 : #endif
1390 : /* CPU-specific state of this task */
1391 : struct thread_struct thread;
1392 : /* filesystem information */
1393 : struct fs_struct *fs;
1394 : /* open file information */
1395 : struct files_struct *files;
1396 : /* namespaces */
1397 : struct nsproxy *nsproxy;
1398 : /* signal handlers */
1399 : struct signal_struct *signal;
1400 : struct sighand_struct *sighand;
1401 :
1402 : sigset_t blocked, real_blocked;
1403 : sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1404 : struct sigpending pending;
1405 :
1406 : unsigned long sas_ss_sp;
1407 : size_t sas_ss_size;
1408 : int (*notifier)(void *priv);
1409 : void *notifier_data;
1410 : sigset_t *notifier_mask;
1411 : struct audit_context *audit_context;
1412 : #ifdef CONFIG_AUDITSYSCALL
1413 : uid_t loginuid;
1414 : unsigned int sessionid;
1415 : #endif
1416 : seccomp_t seccomp;
1417 :
1418 : /* Thread group tracking */
1419 : u32 parent_exec_id;
1420 : u32 self_exec_id;
1421 : /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1422 : * mempolicy */
1423 : spinlock_t alloc_lock;
1424 :
1425 : #ifdef CONFIG_GENERIC_HARDIRQS
1426 : /* IRQ handler threads */
1427 : struct irqaction *irqaction;
1428 : #endif
1429 :
1430 : /* Protection of the PI data structures: */
1431 : raw_spinlock_t pi_lock;
1432 :
1433 : #ifdef CONFIG_RT_MUTEXES
1434 : /* PI waiters blocked on a rt_mutex held by this task */
1435 : struct plist_head pi_waiters;
1436 : /* Deadlock detection and priority inheritance handling */
1437 : struct rt_mutex_waiter *pi_blocked_on;
1438 : #endif
1439 :
1440 : #ifdef CONFIG_DEBUG_MUTEXES
1441 : /* mutex deadlock detection */
1442 : struct mutex_waiter *blocked_on;
1443 : #endif
1444 : #ifdef CONFIG_TRACE_IRQFLAGS
1445 : unsigned int irq_events;
1446 : unsigned long hardirq_enable_ip;
1447 : unsigned long hardirq_disable_ip;
1448 : unsigned int hardirq_enable_event;
1449 : unsigned int hardirq_disable_event;
1450 : int hardirqs_enabled;
1451 : int hardirq_context;
1452 : unsigned long softirq_disable_ip;
1453 : unsigned long softirq_enable_ip;
1454 : unsigned int softirq_disable_event;
1455 : unsigned int softirq_enable_event;
1456 : int softirqs_enabled;
1457 : int softirq_context;
1458 : #endif
1459 : #ifdef CONFIG_LOCKDEP
1460 : # define MAX_LOCK_DEPTH 48UL
1461 : u64 curr_chain_key;
1462 : int lockdep_depth;
1463 : unsigned int lockdep_recursion;
1464 : struct held_lock held_locks[MAX_LOCK_DEPTH];
1465 : gfp_t lockdep_reclaim_gfp;
1466 : #endif
1467 :
1468 : /* journalling filesystem info */
1469 : void *journal_info;
1470 :
1471 : /* stacked block device info */
1472 : struct bio *bio_list, **bio_tail;
1473 :
1474 : /* VM state */
1475 : struct reclaim_state *reclaim_state;
1476 :
1477 : struct backing_dev_info *backing_dev_info;
1478 :
1479 : struct io_context *io_context;
1480 :
1481 : unsigned long ptrace_message;
1482 : siginfo_t *last_siginfo; /* For ptrace use. */
1483 : struct task_io_accounting ioac;
1484 : #if defined(CONFIG_TASK_XACCT)
1485 : u64 acct_rss_mem1; /* accumulated rss usage */
1486 : u64 acct_vm_mem1; /* accumulated virtual memory usage */
1487 : cputime_t acct_timexpd; /* stime + utime since last update */
1488 : #endif
1489 : #ifdef CONFIG_CPUSETS
1490 : nodemask_t mems_allowed; /* Protected by alloc_lock */
1491 : int cpuset_mem_spread_rotor;
1492 : #endif
1493 : #ifdef CONFIG_CGROUPS
1494 : /* Control Group info protected by css_set_lock */
1495 : struct css_set *cgroups;
1496 : /* cg_list protected by css_set_lock and tsk->alloc_lock */
1497 : struct list_head cg_list;
1498 : #endif
1499 : #ifdef CONFIG_FUTEX
1500 : struct robust_list_head __user *robust_list;
1501 : #ifdef CONFIG_COMPAT
1502 : struct compat_robust_list_head __user *compat_robust_list;
1503 : #endif
1504 : struct list_head pi_state_list;
1505 : struct futex_pi_state *pi_state_cache;
1506 : #endif
1507 : #ifdef CONFIG_PERF_EVENTS
1508 : struct perf_event_context *perf_event_ctxp;
1509 : struct mutex perf_event_mutex;
1510 : struct list_head perf_event_list;
1511 : #endif
1512 : #ifdef CONFIG_NUMA
1513 : struct mempolicy *mempolicy; /* Protected by alloc_lock */
1514 : short il_next;
1515 : #endif
1516 : atomic_t fs_excl; /* holding fs exclusive resources */
1517 : struct rcu_head rcu;
1518 :
1519 : /*
1520 : * cache last used pipe for splice
1521 : */
1522 : struct pipe_inode_info *splice_pipe;
1523 : #ifdef CONFIG_TASK_DELAY_ACCT
1524 : struct task_delay_info *delays;
1525 : #endif
1526 : #ifdef CONFIG_FAULT_INJECTION
1527 : int make_it_fail;
1528 : #endif
1529 : struct prop_local_single dirties;
1530 : #ifdef CONFIG_LATENCYTOP
1531 : int latency_record_count;
1532 : struct latency_record latency_record[LT_SAVECOUNT];
1533 : #endif
1534 : /*
1535 : * time slack values; these are used to round up poll() and
1536 : * select() etc timeout values. These are in nanoseconds.
1537 : */
1538 : unsigned long timer_slack_ns;
1539 : unsigned long default_timer_slack_ns;
1540 :
1541 : struct list_head *scm_work_list;
1542 : #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1543 : /* Index of current stored adress in ret_stack */
1544 : int curr_ret_stack;
1545 : /* Stack of return addresses for return function tracing */
1546 : struct ftrace_ret_stack *ret_stack;
1547 : /* time stamp for last schedule */
1548 : unsigned long long ftrace_timestamp;
1549 : /*
1550 : * Number of functions that haven't been traced
1551 : * because of depth overrun.
1552 : */
1553 : atomic_t trace_overrun;
1554 : /* Pause for the tracing */
1555 : atomic_t tracing_graph_pause;
1556 : #endif
1557 : #ifdef CONFIG_TRACING
1558 : /* state flags for use by tracers */
1559 : unsigned long trace;
1560 : /* bitmask of trace recursion */
1561 : unsigned long trace_recursion;
1562 : #endif /* CONFIG_TRACING */
1563 : #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */
1564 : struct memcg_batch_info {
1565 : int do_batch; /* incremented when batch uncharge started */
1566 : struct mem_cgroup *memcg; /* target memcg of uncharge */
1567 : unsigned long bytes; /* uncharged usage */
1568 : unsigned long memsw_bytes; /* uncharged mem+swap usage */
1569 : } memcg_batch;
1570 : #endif
1571 : #ifdef CONFIG_HAVE_HW_BREAKPOINT
1572 : atomic_t ptrace_bp_refcnt;
1573 : #endif
1574 : };
1575 :
1576 : /* Future-safe accessor for struct task_struct's cpus_allowed. */
1577 : #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1578 :
1579 : /*
1580 : * Priority of a process goes from 0..MAX_PRIO-1, valid RT
1581 : * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH
1582 : * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority
1583 : * values are inverted: lower p->prio value means higher priority.
1584 : *
1585 : * The MAX_USER_RT_PRIO value allows the actual maximum
1586 : * RT priority to be separate from the value exported to
1587 : * user-space. This allows kernel threads to set their
1588 : * priority to a value higher than any user task. Note:
1589 : * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO.
1590 : */
1591 :
1592 : #define MAX_USER_RT_PRIO 100
1593 : #define MAX_RT_PRIO MAX_USER_RT_PRIO
1594 :
1595 : #define MAX_PRIO (MAX_RT_PRIO + 40)
1596 : #define DEFAULT_PRIO (MAX_RT_PRIO + 20)
1597 :
1598 : static inline int rt_prio(int prio)
1599 : {
1600 : if (unlikely(prio < MAX_RT_PRIO))
1601 : return 1;
1602 : return 0;
1603 : }
1604 :
1605 : static inline int rt_task(struct task_struct *p)
1606 : {
1607 : return rt_prio(p->prio);
1608 : }
1609 :
1610 : static inline struct pid *task_pid(struct task_struct *task)
1611 : {
1612 : return task->pids[PIDTYPE_PID].pid;
1613 : }
1614 :
1615 : static inline struct pid *task_tgid(struct task_struct *task)
1616 : {
1617 : return task->group_leader->pids[PIDTYPE_PID].pid;
1618 : }
1619 :
1620 : /*
1621 : * Without tasklist or rcu lock it is not safe to dereference
1622 : * the result of task_pgrp/task_session even if task == current,
1623 : * we can race with another thread doing sys_setsid/sys_setpgid.
1624 : */
1625 : static inline struct pid *task_pgrp(struct task_struct *task)
1626 : {
1627 : return task->group_leader->pids[PIDTYPE_PGID].pid;
1628 : }
1629 :
1630 : static inline struct pid *task_session(struct task_struct *task)
1631 : {
1632 : return task->group_leader->pids[PIDTYPE_SID].pid;
1633 : }
1634 :
1635 : struct pid_namespace;
1636 :
1637 : /*
1638 : * the helpers to get the task's different pids as they are seen
1639 : * from various namespaces
1640 : *
1641 : * task_xid_nr() : global id, i.e. the id seen from the init namespace;
1642 : * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of
1643 : * current.
1644 : * task_xid_nr_ns() : id seen from the ns specified;
1645 : *
1646 : * set_task_vxid() : assigns a virtual id to a task;
1647 : *
1648 : * see also pid_nr() etc in include/linux/pid.h
1649 : */
1650 : pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
1651 : struct pid_namespace *ns);
1652 :
1653 : static inline pid_t task_pid_nr(struct task_struct *tsk)
1654 : {
1655 : return tsk->pid;
1656 : }
1657 :
1658 : static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
1659 : struct pid_namespace *ns)
1660 : {
1661 : return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
1662 : }
1663 :
1664 : static inline pid_t task_pid_vnr(struct task_struct *tsk)
1665 : {
1666 : return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
1667 : }
1668 :
1669 :
1670 : static inline pid_t task_tgid_nr(struct task_struct *tsk)
1671 : {
1672 : return tsk->tgid;
1673 : }
1674 :
1675 : pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
1676 :
1677 : static inline pid_t task_tgid_vnr(struct task_struct *tsk)
1678 : {
1679 : return pid_vnr(task_tgid(tsk));
1680 : }
1681 :
1682 :
1683 : static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
1684 : struct pid_namespace *ns)
1685 : {
1686 : return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
1687 : }
1688 :
1689 : static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
1690 : {
1691 : return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
1692 : }
1693 :
1694 :
1695 : static inline pid_t task_session_nr_ns(struct task_struct *tsk,
1696 : struct pid_namespace *ns)
1697 : {
1698 : return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
1699 : }
1700 :
1701 : static inline pid_t task_session_vnr(struct task_struct *tsk)
1702 : {
1703 : return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
1704 : }
1705 :
1706 : /* obsolete, do not use */
1707 : static inline pid_t task_pgrp_nr(struct task_struct *tsk)
1708 : {
1709 : return task_pgrp_nr_ns(tsk, &init_pid_ns);
1710 : }
1711 :
1712 : /**
1713 : * pid_alive - check that a task structure is not stale
1714 : * @p: Task structure to be checked.
1715 : *
1716 : * Test if a process is not yet dead (at most zombie state)
1717 : * If pid_alive fails, then pointers within the task structure
1718 : * can be stale and must not be dereferenced.
1719 : */
1720 : static inline int pid_alive(struct task_struct *p)
1721 : {
1722 : return p->pids[PIDTYPE_PID].pid != NULL;
1723 : }
1724 :
1725 : /**
1726 : * is_global_init - check if a task structure is init
1727 : * @tsk: Task structure to be checked.
1728 : *
1729 : * Check if a task structure is the first user space task the kernel created.
1730 : */
1731 : static inline int is_global_init(struct task_struct *tsk)
1732 : {
1733 : return tsk->pid == 1;
1734 : }
1735 :
1736 : /*
1737 : * is_container_init:
1738 : * check whether in the task is init in its own pid namespace.
1739 : */
1740 : extern int is_container_init(struct task_struct *tsk);
1741 :
1742 : extern struct pid *cad_pid;
1743 :
1744 : extern void free_task(struct task_struct *tsk);
1745 : #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
1746 :
1747 : extern void __put_task_struct(struct task_struct *t);
1748 :
1749 : static inline void put_task_struct(struct task_struct *t)
1750 : {
1751 : if (atomic_dec_and_test(&t->usage))
1752 : __put_task_struct(t);
1753 : }
1754 :
1755 : extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1756 : extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st);
1757 :
1758 : /*
1759 : * Per process flags
1760 : */
1761 : #define PF_ALIGNWARN 0x00000001 /* Print alignment warning msgs */
1762 : /* Not implemented yet, only for 486*/
1763 : #define PF_STARTING 0x00000002 /* being created */
1764 : #define PF_EXITING 0x00000004 /* getting shut down */
1765 : #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */
1766 : #define PF_VCPU 0x00000010 /* I'm a virtual CPU */
1767 : #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */
1768 : #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */
1769 : #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */
1770 : #define PF_DUMPCORE 0x00000200 /* dumped core */
1771 : #define PF_SIGNALED 0x00000400 /* killed by a signal */
1772 : #define PF_MEMALLOC 0x00000800 /* Allocating memory */
1773 : #define PF_FLUSHER 0x00001000 /* responsible for disk writeback */
1774 : #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */
1775 : #define PF_FREEZING 0x00004000 /* freeze in progress. do not account to load */
1776 : #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */
1777 : #define PF_FROZEN 0x00010000 /* frozen for system suspend */
1778 : #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */
1779 : #define PF_KSWAPD 0x00040000 /* I am kswapd */
1780 : #define PF_OOM_ORIGIN 0x00080000 /* Allocating much memory to others */
1781 : #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */
1782 : #define PF_KTHREAD 0x00200000 /* I am a kernel thread */
1783 : #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */
1784 : #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */
1785 : #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */
1786 : #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */
1787 : #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */
1788 : #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */
1789 : #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */
1790 : #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */
1791 : #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezeable */
1792 : #define PF_FREEZER_NOSIG 0x80000000 /* Freezer won't send signals to it */
1793 :
1794 : /*
1795 : * Only the _current_ task can read/write to tsk->flags, but other
1796 : * tasks can access tsk->flags in readonly mode for example
1797 : * with tsk_used_math (like during threaded core dumping).
1798 : * There is however an exception to this rule during ptrace
1799 : * or during fork: the ptracer task is allowed to write to the
1800 : * child->flags of its traced child (same goes for fork, the parent
1801 : * can write to the child->flags), because we're guaranteed the
1802 : * child is not running and in turn not changing child->flags
1803 : * at the same time the parent does it.
1804 : */
1805 : #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
1806 : #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
1807 : #define clear_used_math() clear_stopped_child_used_math(current)
1808 : #define set_used_math() set_stopped_child_used_math(current)
1809 : #define conditional_stopped_child_used_math(condition, child) \
1810 : do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
1811 : #define conditional_used_math(condition) \
1812 : conditional_stopped_child_used_math(condition, current)
1813 : #define copy_to_stopped_child_used_math(child) \
1814 : do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
1815 : /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
1816 : #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
1817 : #define used_math() tsk_used_math(current)
1818 :
1819 : #ifdef CONFIG_TREE_PREEMPT_RCU
1820 :
1821 : #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */
1822 : #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */
1823 :
1824 : static inline void rcu_copy_process(struct task_struct *p)
1825 : {
1826 : p->rcu_read_lock_nesting = 0;
1827 : p->rcu_read_unlock_special = 0;
1828 : p->rcu_blocked_node = NULL;
1829 : INIT_LIST_HEAD(&p->rcu_node_entry);
1830 : }
1831 :
1832 : #else
1833 :
1834 : static inline void rcu_copy_process(struct task_struct *p)
1835 : {
1836 : }
1837 :
1838 : #endif
1839 :
1840 : #ifdef CONFIG_SMP
1841 : extern int set_cpus_allowed_ptr(struct task_struct *p,
1842 : const struct cpumask *new_mask);
1843 : #else
1844 : static inline int set_cpus_allowed_ptr(struct task_struct *p,
1845 : const struct cpumask *new_mask)
1846 : {
1847 : if (!cpumask_test_cpu(0, new_mask))
1848 : return -EINVAL;
1849 : return 0;
1850 : }
1851 : #endif
1852 :
1853 : #ifndef CONFIG_CPUMASK_OFFSTACK
1854 : static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask)
1855 : {
1856 : return set_cpus_allowed_ptr(p, &new_mask);
1857 : }
1858 : #endif
1859 :
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 : #ifdef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1867 : extern int sched_clock_stable;
1868 : #endif
1869 :
1870 : /* ftrace calls sched_clock() directly */
1871 : extern unsigned long long notrace sched_clock(void);
1872 :
1873 : extern void sched_clock_init(void);
1874 : extern u64 sched_clock_cpu(int cpu);
1875 :
1876 : #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
1877 : static inline void sched_clock_tick(void)
1878 : {
1879 : }
1880 :
1881 : static inline void sched_clock_idle_sleep_event(void)
1882 : {
1883 : }
1884 :
1885 : static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
1886 : {
1887 : }
1888 : #else
1889 : extern void sched_clock_tick(void);
1890 : extern void sched_clock_idle_sleep_event(void);
1891 : extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1892 : #endif
1893 :
1894 : /*
1895 : * For kernel-internal use: high-speed (but slightly incorrect) per-cpu
1896 : * clock constructed from sched_clock():
1897 : */
1898 : extern unsigned long long cpu_clock(int cpu);
1899 :
1900 : extern unsigned long long
1901 : task_sched_runtime(struct task_struct *task);
1902 : extern unsigned long long thread_group_sched_runtime(struct task_struct *task);
1903 :
1904 : /* sched_exec is called by processes performing an exec */
1905 : #ifdef CONFIG_SMP
1906 : extern void sched_exec(void);
1907 : #else
1908 : #define sched_exec() {}
1909 : #endif
1910 :
1911 : extern void sched_clock_idle_sleep_event(void);
1912 : extern void sched_clock_idle_wakeup_event(u64 delta_ns);
1913 :
1914 : #ifdef CONFIG_HOTPLUG_CPU
1915 : extern void idle_task_exit(void);
1916 : #else
1917 : static inline void idle_task_exit(void) {}
1918 : #endif
1919 :
1920 : extern void sched_idle_next(void);
1921 :
1922 : #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP)
1923 : extern void wake_up_idle_cpu(int cpu);
1924 : #else
1925 : static inline void wake_up_idle_cpu(int cpu) { }
1926 : #endif
1927 :
1928 : extern unsigned int sysctl_sched_latency;
1929 : extern unsigned int sysctl_sched_min_granularity;
1930 : extern unsigned int sysctl_sched_wakeup_granularity;
1931 : extern unsigned int sysctl_sched_shares_ratelimit;
1932 : extern unsigned int sysctl_sched_shares_thresh;
1933 : extern unsigned int sysctl_sched_child_runs_first;
1934 :
1935 : enum sched_tunable_scaling {
1936 : SCHED_TUNABLESCALING_NONE,
1937 : SCHED_TUNABLESCALING_LOG,
1938 : SCHED_TUNABLESCALING_LINEAR,
1939 : SCHED_TUNABLESCALING_END,
1940 : };
1941 : extern enum sched_tunable_scaling sysctl_sched_tunable_scaling;
1942 :
1943 : #ifdef CONFIG_SCHED_DEBUG
1944 : extern unsigned int sysctl_sched_migration_cost;
1945 : extern unsigned int sysctl_sched_nr_migrate;
1946 : extern unsigned int sysctl_sched_time_avg;
1947 : extern unsigned int sysctl_timer_migration;
1948 :
1949 : int sched_proc_update_handler(struct ctl_table *table, int write,
1950 : void __user *buffer, size_t *length,
1951 : loff_t *ppos);
1952 : #endif
1953 : #ifdef CONFIG_SCHED_DEBUG
1954 : static inline unsigned int get_sysctl_timer_migration(void)
1955 : {
1956 : return sysctl_timer_migration;
1957 : }
1958 : #else
1959 : static inline unsigned int get_sysctl_timer_migration(void)
1960 : {
1961 : return 1;
1962 : }
1963 : #endif
1964 : extern unsigned int sysctl_sched_rt_period;
1965 : extern int sysctl_sched_rt_runtime;
1966 :
1967 : int sched_rt_handler(struct ctl_table *table, int write,
1968 : void __user *buffer, size_t *lenp,
1969 : loff_t *ppos);
1970 :
1971 : extern unsigned int sysctl_sched_compat_yield;
1972 :
1973 : #ifdef CONFIG_RT_MUTEXES
1974 : extern int rt_mutex_getprio(struct task_struct *p);
1975 : extern void rt_mutex_setprio(struct task_struct *p, int prio);
1976 : extern void rt_mutex_adjust_pi(struct task_struct *p);
1977 : #else
1978 : static inline int rt_mutex_getprio(struct task_struct *p)
1979 : {
1980 : return p->normal_prio;
1981 : }
1982 : # define rt_mutex_adjust_pi(p) do { } while (0)
1983 : #endif
1984 :
1985 : extern void set_user_nice(struct task_struct *p, long nice);
1986 : extern int task_prio(const struct task_struct *p);
1987 : extern int task_nice(const struct task_struct *p);
1988 : extern int can_nice(const struct task_struct *p, const int nice);
1989 : extern int task_curr(const struct task_struct *p);
1990 : extern int idle_cpu(int cpu);
1991 : extern int sched_setscheduler(struct task_struct *, int, struct sched_param *);
1992 : extern int sched_setscheduler_nocheck(struct task_struct *, int,
1993 : struct sched_param *);
1994 : extern struct task_struct *idle_task(int cpu);
1995 : extern struct task_struct *curr_task(int cpu);
1996 : extern void set_curr_task(int cpu, struct task_struct *p);
1997 :
1998 : void yield(void);
1999 :
2000 : /*
2001 : * The default (Linux) execution domain.
2002 : */
2003 : extern struct exec_domain default_exec_domain;
2004 :
2005 : union thread_union {
2006 : struct thread_info thread_info;
2007 : unsigned long stack[THREAD_SIZE/sizeof(long)];
2008 : };
2009 :
2010 : #ifndef __HAVE_ARCH_KSTACK_END
2011 : static inline int kstack_end(void *addr)
2012 : {
2013 : /* Reliable end of stack detection:
2014 : * Some APM bios versions misalign the stack
2015 : */
2016 : return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2017 : }
2018 : #endif
2019 :
2020 : extern union thread_union init_thread_union;
2021 : extern struct task_struct init_task;
2022 :
2023 : extern struct mm_struct init_mm;
2024 :
2025 : extern struct pid_namespace init_pid_ns;
2026 :
2027 : /*
2028 : * find a task by one of its numerical ids
2029 : *
2030 : * find_task_by_pid_ns():
2031 : * finds a task by its pid in the specified namespace
2032 : * find_task_by_vpid():
2033 : * finds a task by its virtual pid
2034 : *
2035 : * see also find_vpid() etc in include/linux/pid.h
2036 : */
2037 :
2038 : extern struct task_struct *find_task_by_vpid(pid_t nr);
2039 : extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2040 : struct pid_namespace *ns);
2041 :
2042 : extern void __set_special_pids(struct pid *pid);
2043 :
2044 : /* per-UID process charging. */
2045 : extern struct user_struct * alloc_uid(struct user_namespace *, uid_t);
2046 : static inline struct user_struct *get_uid(struct user_struct *u)
2047 : {
2048 : atomic_inc(&u->__count);
2049 : return u;
2050 : }
2051 : extern void free_uid(struct user_struct *);
2052 : extern void release_uids(struct user_namespace *ns);
2053 :
2054 : #include <asm/current.h>
2055 :
2056 : extern void do_timer(unsigned long ticks);
2057 :
2058 : extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2059 : extern int wake_up_process(struct task_struct *tsk);
2060 : extern void wake_up_new_task(struct task_struct *tsk,
2061 : unsigned long clone_flags);
2062 : #ifdef CONFIG_SMP
2063 : extern void kick_process(struct task_struct *tsk);
2064 : #else
2065 : static inline void kick_process(struct task_struct *tsk) { }
2066 : #endif
2067 : extern void sched_fork(struct task_struct *p, int clone_flags);
2068 : extern void sched_dead(struct task_struct *p);
2069 :
2070 : extern void proc_caches_init(void);
2071 : extern void flush_signals(struct task_struct *);
2072 : extern void __flush_signals(struct task_struct *);
2073 : extern void ignore_signals(struct task_struct *);
2074 : extern void flush_signal_handlers(struct task_struct *, int force_default);
2075 : extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2076 :
2077 : static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info)
2078 : {
2079 : unsigned long flags;
2080 : int ret;
2081 :
2082 : spin_lock_irqsave(&tsk->sighand->siglock, flags);
2083 : ret = dequeue_signal(tsk, mask, info);
2084 : spin_unlock_irqrestore(&tsk->sighand->siglock, flags);
2085 :
2086 : return ret;
2087 : }
2088 :
2089 : extern void block_all_signals(int (*notifier)(void *priv), void *priv,
2090 : sigset_t *mask);
2091 : extern void unblock_all_signals(void);
2092 : extern void release_task(struct task_struct * p);
2093 : extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2094 : extern int force_sigsegv(int, struct task_struct *);
2095 : extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2096 : extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2097 : extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2098 : extern int kill_pid_info_as_uid(int, struct siginfo *, struct pid *, uid_t, uid_t, u32);
2099 : extern int kill_pgrp(struct pid *pid, int sig, int priv);
2100 : extern int kill_pid(struct pid *pid, int sig, int priv);
2101 : extern int kill_proc_info(int, struct siginfo *, pid_t);
2102 : extern int do_notify_parent(struct task_struct *, int);
2103 : extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2104 : extern void force_sig(int, struct task_struct *);
2105 : extern int send_sig(int, struct task_struct *, int);
2106 : extern void zap_other_threads(struct task_struct *p);
2107 : extern struct sigqueue *sigqueue_alloc(void);
2108 : extern void sigqueue_free(struct sigqueue *);
2109 : extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group);
2110 : extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2111 : extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long);
2112 :
2113 : static inline int kill_cad_pid(int sig, int priv)
2114 : {
2115 : return kill_pid(cad_pid, sig, priv);
2116 : }
2117 :
2118 : /* These can be the second arg to send_sig_info/send_group_sig_info. */
2119 : #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2120 : #define SEND_SIG_PRIV ((struct siginfo *) 1)
2121 : #define SEND_SIG_FORCED ((struct siginfo *) 2)
2122 :
2123 : /*
2124 : * True if we are on the alternate signal stack.
2125 : */
2126 : static inline int on_sig_stack(unsigned long sp)
2127 : {
2128 : #ifdef CONFIG_STACK_GROWSUP
2129 : return sp >= current->sas_ss_sp &&
2130 : sp - current->sas_ss_sp < current->sas_ss_size;
2131 : #else
2132 : return sp > current->sas_ss_sp &&
2133 : sp - current->sas_ss_sp <= current->sas_ss_size;
2134 : #endif
2135 : }
2136 :
2137 : static inline int sas_ss_flags(unsigned long sp)
2138 : {
2139 : return (current->sas_ss_size == 0 ? SS_DISABLE
2140 : : on_sig_stack(sp) ? SS_ONSTACK : 0);
2141 : }
2142 :
2143 : /*
2144 : * Routines for handling mm_structs
2145 : */
2146 : extern struct mm_struct * mm_alloc(void);
2147 :
2148 : /* mmdrop drops the mm and the page tables */
2149 : extern void __mmdrop(struct mm_struct *);
2150 : static inline void mmdrop(struct mm_struct * mm)
2151 : {
2152 : if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2153 : __mmdrop(mm);
2154 : }
2155 :
2156 : /* mmput gets rid of the mappings and all user-space */
2157 : extern void mmput(struct mm_struct *);
2158 : /* Grab a reference to a task's mm, if it is not already going away */
2159 : extern struct mm_struct *get_task_mm(struct task_struct *task);
2160 : /* Remove the current tasks stale references to the old mm_struct */
2161 : extern void mm_release(struct task_struct *, struct mm_struct *);
2162 : /* Allocate a new mm structure and copy contents from tsk->mm */
2163 : extern struct mm_struct *dup_mm(struct task_struct *tsk);
2164 :
2165 : extern int copy_thread(unsigned long, unsigned long, unsigned long,
2166 : struct task_struct *, struct pt_regs *);
2167 : extern void flush_thread(void);
2168 : extern void exit_thread(void);
2169 :
2170 : extern void exit_files(struct task_struct *);
2171 : extern void __cleanup_signal(struct signal_struct *);
2172 : extern void __cleanup_sighand(struct sighand_struct *);
2173 :
2174 : extern void exit_itimers(struct signal_struct *);
2175 : extern void flush_itimer_signals(void);
2176 :
2177 : extern NORET_TYPE void do_group_exit(int);
2178 :
2179 : extern void daemonize(const char *, ...);
2180 : extern int allow_signal(int);
2181 : extern int disallow_signal(int);
2182 :
2183 : extern int do_execve(char *, char __user * __user *, char __user * __user *, struct pt_regs *);
2184 : extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *);
2185 : struct task_struct *fork_idle(int);
2186 :
2187 : extern void set_task_comm(struct task_struct *tsk, char *from);
2188 : extern char *get_task_comm(char *to, struct task_struct *tsk);
2189 :
2190 : #ifdef CONFIG_SMP
2191 : extern void wait_task_context_switch(struct task_struct *p);
2192 : extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2193 : #else
2194 : static inline void wait_task_context_switch(struct task_struct *p) {}
2195 : static inline unsigned long wait_task_inactive(struct task_struct *p,
2196 : long match_state)
2197 : {
2198 : return 1;
2199 : }
2200 : #endif
2201 :
2202 : #define next_task(p) \
2203 : list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2204 :
2205 : #define for_each_process(p) \
2206 : for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2207 :
2208 : extern bool current_is_single_threaded(void);
2209 :
2210 : /*
2211 : * Careful: do_each_thread/while_each_thread is a double loop so
2212 : * 'break' will not work as expected - use goto instead.
2213 : */
2214 : #define do_each_thread(g, t) \
2215 : for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2216 :
2217 : #define while_each_thread(g, t) \
2218 : while ((t = next_thread(t)) != g)
2219 :
2220 : /* de_thread depends on thread_group_leader not being a pid based check */
2221 : #define thread_group_leader(p) (p == p->group_leader)
2222 :
2223 : /* Do to the insanities of de_thread it is possible for a process
2224 : * to have the pid of the thread group leader without actually being
2225 : * the thread group leader. For iteration through the pids in proc
2226 : * all we care about is that we have a task with the appropriate
2227 : * pid, we don't actually care if we have the right task.
2228 : */
2229 : static inline int has_group_leader_pid(struct task_struct *p)
2230 : {
2231 : return p->pid == p->tgid;
2232 : }
2233 :
2234 : static inline
2235 : int same_thread_group(struct task_struct *p1, struct task_struct *p2)
2236 : {
2237 : return p1->tgid == p2->tgid;
2238 : }
2239 :
2240 : static inline struct task_struct *next_thread(const struct task_struct *p)
2241 : {
2242 : return list_entry_rcu(p->thread_group.next,
2243 : struct task_struct, thread_group);
2244 : }
2245 :
2246 : static inline int thread_group_empty(struct task_struct *p)
2247 : {
2248 : return list_empty(&p->thread_group);
2249 : }
2250 :
2251 : #define delay_group_leader(p) \
2252 : (thread_group_leader(p) && !thread_group_empty(p))
2253 :
2254 : static inline int task_detached(struct task_struct *p)
2255 : {
2256 : return p->exit_signal == -1;
2257 : }
2258 :
2259 : /*
2260 : * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
2261 : * subscriptions and synchronises with wait4(). Also used in procfs. Also
2262 : * pins the final release of task.io_context. Also protects ->cpuset and
2263 : * ->cgroup.subsys[].
2264 : *
2265 : * Nests both inside and outside of read_lock(&tasklist_lock).
2266 : * It must not be nested with write_lock_irq(&tasklist_lock),
2267 : * neither inside nor outside.
2268 : */
2269 : static inline void task_lock(struct task_struct *p)
2270 : {
2271 : spin_lock(&p->alloc_lock);
2272 : }
2273 :
2274 : static inline void task_unlock(struct task_struct *p)
2275 : {
2276 : spin_unlock(&p->alloc_lock);
2277 : }
2278 :
2279 : extern struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
2280 : unsigned long *flags);
2281 :
2282 : static inline void unlock_task_sighand(struct task_struct *tsk,
2283 : unsigned long *flags)
2284 : {
2285 : spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
2286 : }
2287 :
2288 : #ifndef __HAVE_THREAD_FUNCTIONS
2289 :
2290 : #define task_thread_info(task) ((struct thread_info *)(task)->stack)
2291 : #define task_stack_page(task) ((task)->stack)
2292 :
2293 : static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
2294 : {
2295 : *task_thread_info(p) = *task_thread_info(org);
2296 : task_thread_info(p)->task = p;
2297 : }
2298 :
2299 : static inline unsigned long *end_of_stack(struct task_struct *p)
2300 : {
2301 : return (unsigned long *)(task_thread_info(p) + 1);
2302 : }
2303 :
2304 : #endif
2305 :
2306 : static inline int object_is_on_stack(void *obj)
2307 : {
2308 : void *stack = task_stack_page(current);
2309 :
2310 : return (obj >= stack) && (obj < (stack + THREAD_SIZE));
2311 : }
2312 :
2313 : extern void thread_info_cache_init(void);
2314 :
2315 : #ifdef CONFIG_DEBUG_STACK_USAGE
2316 : static inline unsigned long stack_not_used(struct task_struct *p)
2317 : {
2318 : unsigned long *n = end_of_stack(p);
2319 :
2320 : do { /* Skip over canary */
2321 : n++;
2322 : } while (!*n);
2323 :
2324 : return (unsigned long)n - (unsigned long)end_of_stack(p);
2325 : }
2326 : #endif
2327 :
2328 : /* set thread flags in other task's structures
2329 : * - see asm/thread_info.h for TIF_xxxx flags available
2330 : */
2331 : static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
2332 : {
2333 : set_ti_thread_flag(task_thread_info(tsk), flag);
2334 : }
2335 :
2336 : static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2337 : {
2338 : clear_ti_thread_flag(task_thread_info(tsk), flag);
2339 : }
2340 :
2341 : static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
2342 : {
2343 : return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
2344 : }
2345 :
2346 : static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
2347 : {
2348 : return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
2349 : }
2350 :
2351 : static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
2352 : {
2353 : return test_ti_thread_flag(task_thread_info(tsk), flag);
2354 : }
2355 :
2356 : static inline void set_tsk_need_resched(struct task_struct *tsk)
2357 : {
2358 : set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2359 : }
2360 :
2361 : static inline void clear_tsk_need_resched(struct task_struct *tsk)
2362 : {
2363 : clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
2364 : }
2365 :
2366 : static inline int test_tsk_need_resched(struct task_struct *tsk)
2367 : {
2368 : return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
2369 : }
2370 :
2371 : static inline int restart_syscall(void)
2372 : {
2373 : set_tsk_thread_flag(current, TIF_SIGPENDING);
2374 : return -ERESTARTNOINTR;
2375 : }
2376 :
2377 : static inline int signal_pending(struct task_struct *p)
2378 : {
2379 : return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
2380 : }
2381 :
2382 : static inline int __fatal_signal_pending(struct task_struct *p)
2383 : {
2384 : return unlikely(sigismember(&p->pending.signal, SIGKILL));
2385 : }
2386 :
2387 : static inline int fatal_signal_pending(struct task_struct *p)
2388 : {
2389 : return signal_pending(p) && __fatal_signal_pending(p);
2390 : }
2391 :
2392 : static inline int signal_pending_state(long state, struct task_struct *p)
2393 : {
2394 : if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
2395 : return 0;
2396 : if (!signal_pending(p))
2397 : return 0;
2398 :
2399 : return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
2400 : }
2401 :
2402 : static inline int need_resched(void)
2403 : {
2404 0 : return unlikely(test_thread_flag(TIF_NEED_RESCHED));
2405 0 : }
2406 0 :
2407 : /*
2408 : * cond_resched() and cond_resched_lock(): latency reduction via
2409 : * explicit rescheduling in places that are safe. The return
2410 : * value indicates whether a reschedule was done in fact.
2411 : * cond_resched_lock() will drop the spinlock before scheduling,
2412 : * cond_resched_softirq() will enable bhs before scheduling.
2413 : */
2414 : extern int _cond_resched(void);
2415 :
2416 : #define cond_resched() ({ \
2417 : __might_sleep(__FILE__, __LINE__, 0); \
2418 : _cond_resched(); \
2419 : })
2420 :
2421 : extern int __cond_resched_lock(spinlock_t *lock);
2422 :
2423 : #ifdef CONFIG_PREEMPT
2424 : #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET
2425 : #else
2426 : #define PREEMPT_LOCK_OFFSET 0
2427 : #endif
2428 :
2429 : #define cond_resched_lock(lock) ({ \
2430 : __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \
2431 : __cond_resched_lock(lock); \
2432 : })
2433 :
2434 : extern int __cond_resched_softirq(void);
2435 :
2436 : #define cond_resched_softirq() ({ \
2437 : __might_sleep(__FILE__, __LINE__, SOFTIRQ_OFFSET); \
2438 : __cond_resched_softirq(); \
2439 : })
2440 :
2441 : /*
2442 : * Does a critical section need to be broken due to another
2443 : * task waiting?: (technically does not depend on CONFIG_PREEMPT,
2444 : * but a general need for low latency)
2445 : */
2446 : static inline int spin_needbreak(spinlock_t *lock)
2447 : {
2448 : #ifdef CONFIG_PREEMPT
2449 : return spin_is_contended(lock);
2450 : #else
2451 0 : return 0;
2452 : #endif
2453 : }
2454 :
2455 : /*
2456 : * Thread group CPU time accounting.
2457 : */
2458 : void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
2459 : void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
2460 :
2461 : static inline void thread_group_cputime_init(struct signal_struct *sig)
2462 : {
2463 : sig->cputimer.cputime = INIT_CPUTIME;
2464 : spin_lock_init(&sig->cputimer.lock);
2465 : sig->cputimer.running = 0;
2466 : }
2467 :
2468 : static inline void thread_group_cputime_free(struct signal_struct *sig)
2469 : {
2470 : }
2471 :
2472 : /*
2473 : * Reevaluate whether the task has signals pending delivery.
2474 : * Wake the task if so.
2475 : * This is required every time the blocked sigset_t changes.
2476 : * callers must hold sighand->siglock.
2477 : */
2478 : extern void recalc_sigpending_and_wake(struct task_struct *t);
2479 : extern void recalc_sigpending(void);
2480 :
2481 : extern void signal_wake_up(struct task_struct *t, int resume_stopped);
2482 :
2483 : /*
2484 : * Wrappers for p->thread_info->cpu access. No-op on UP.
2485 : */
2486 : #ifdef CONFIG_SMP
2487 :
2488 : static inline unsigned int task_cpu(const struct task_struct *p)
2489 : {
2490 : return task_thread_info(p)->cpu;
2491 : }
2492 :
2493 : extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
2494 :
2495 : #else
2496 :
2497 : static inline unsigned int task_cpu(const struct task_struct *p)
2498 : {
2499 : return 0;
2500 : }
2501 :
2502 : static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
2503 : {
2504 : }
2505 :
2506 : #endif /* CONFIG_SMP */
2507 :
2508 : #ifdef CONFIG_TRACING
2509 : extern void
2510 : __trace_special(void *__tr, void *__data,
2511 : unsigned long arg1, unsigned long arg2, unsigned long arg3);
2512 : #else
2513 : static inline void
2514 : __trace_special(void *__tr, void *__data,
2515 : unsigned long arg1, unsigned long arg2, unsigned long arg3)
2516 : {
2517 : }
2518 : #endif
2519 :
2520 : extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
2521 : extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
2522 :
2523 : extern void normalize_rt_tasks(void);
2524 :
2525 : #ifdef CONFIG_GROUP_SCHED
2526 :
2527 : extern struct task_group init_task_group;
2528 : #ifdef CONFIG_USER_SCHED
2529 : extern struct task_group root_task_group;
2530 : extern void set_tg_uid(struct user_struct *user);
2531 : #endif
2532 :
2533 : extern struct task_group *sched_create_group(struct task_group *parent);
2534 : extern void sched_destroy_group(struct task_group *tg);
2535 : extern void sched_move_task(struct task_struct *tsk);
2536 : #ifdef CONFIG_FAIR_GROUP_SCHED
2537 : extern int sched_group_set_shares(struct task_group *tg, unsigned long shares);
2538 : extern unsigned long sched_group_shares(struct task_group *tg);
2539 : #endif
2540 : #ifdef CONFIG_RT_GROUP_SCHED
2541 : extern int sched_group_set_rt_runtime(struct task_group *tg,
2542 : long rt_runtime_us);
2543 : extern long sched_group_rt_runtime(struct task_group *tg);
2544 : extern int sched_group_set_rt_period(struct task_group *tg,
2545 : long rt_period_us);
2546 : extern long sched_group_rt_period(struct task_group *tg);
2547 : extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk);
2548 : #endif
2549 : #endif
2550 :
2551 : extern int task_can_switch_user(struct user_struct *up,
2552 : struct task_struct *tsk);
2553 :
2554 : #ifdef CONFIG_TASK_XACCT
2555 : static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2556 : {
2557 : tsk->ioac.rchar += amt;
2558 : }
2559 :
2560 : static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2561 : {
2562 : tsk->ioac.wchar += amt;
2563 : }
2564 :
2565 : static inline void inc_syscr(struct task_struct *tsk)
2566 : {
2567 : tsk->ioac.syscr++;
2568 : }
2569 :
2570 : static inline void inc_syscw(struct task_struct *tsk)
2571 : {
2572 : tsk->ioac.syscw++;
2573 : }
2574 : #else
2575 : static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
2576 : {
2577 : }
2578 :
2579 : static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
2580 : {
2581 : }
2582 :
2583 : static inline void inc_syscr(struct task_struct *tsk)
2584 : {
2585 : }
2586 :
2587 : static inline void inc_syscw(struct task_struct *tsk)
2588 : {
2589 : }
2590 : #endif
2591 :
2592 : #ifndef TASK_SIZE_OF
2593 : #define TASK_SIZE_OF(tsk) TASK_SIZE
2594 : #endif
2595 :
2596 : /*
2597 : * Call the function if the target task is executing on a CPU right now:
2598 : */
2599 : extern void task_oncpu_function_call(struct task_struct *p,
2600 : void (*func) (void *info), void *info);
2601 :
2602 :
2603 : #ifdef CONFIG_MM_OWNER
2604 : extern void mm_update_next_owner(struct mm_struct *mm);
2605 : extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p);
2606 : #else
2607 : static inline void mm_update_next_owner(struct mm_struct *mm)
2608 : {
2609 : }
2610 :
2611 : static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p)
2612 : {
2613 : }
2614 : #endif /* CONFIG_MM_OWNER */
2615 :
2616 : static inline unsigned long task_rlimit(const struct task_struct *tsk,
2617 : unsigned int limit)
2618 : {
2619 : return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur);
2620 : }
2621 :
2622 : static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
2623 : unsigned int limit)
2624 : {
2625 : return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max);
2626 : }
2627 :
2628 : static inline unsigned long rlimit(unsigned int limit)
2629 : {
2630 : return task_rlimit(current, limit);
2631 : }
2632 :
2633 : static inline unsigned long rlimit_max(unsigned int limit)
2634 : {
2635 : return task_rlimit_max(current, limit);
2636 : }
2637 :
2638 : #endif /* __KERNEL__ */
2639 :
2640 : #endif
|