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