]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_SCHED_H |
2 | #define _LINUX_SCHED_H | |
3 | ||
5eca1c10 IM |
4 | /* |
5 | * Define 'struct task_struct' and provide the main scheduler | |
6 | * APIs (schedule(), wakeup variants, etc.) | |
7 | */ | |
b7b3c76a | 8 | |
5eca1c10 | 9 | #include <uapi/linux/sched.h> |
5c228079 | 10 | |
5eca1c10 | 11 | #include <asm/current.h> |
1da177e4 | 12 | |
5eca1c10 | 13 | #include <linux/pid.h> |
1da177e4 | 14 | #include <linux/sem.h> |
ab602f79 | 15 | #include <linux/shm.h> |
5eca1c10 IM |
16 | #include <linux/kcov.h> |
17 | #include <linux/mutex.h> | |
18 | #include <linux/plist.h> | |
19 | #include <linux/hrtimer.h> | |
1da177e4 | 20 | #include <linux/seccomp.h> |
5eca1c10 | 21 | #include <linux/nodemask.h> |
b68070e1 | 22 | #include <linux/rcupdate.h> |
a3b6714e | 23 | #include <linux/resource.h> |
9745512c | 24 | #include <linux/latencytop.h> |
5eca1c10 IM |
25 | #include <linux/sched/prio.h> |
26 | #include <linux/signal_types.h> | |
27 | #include <linux/mm_types_task.h> | |
28 | #include <linux/task_io_accounting.h> | |
a3b6714e | 29 | |
5eca1c10 | 30 | /* task_struct member predeclarations (sorted alphabetically): */ |
c7af7877 | 31 | struct audit_context; |
c7af7877 | 32 | struct backing_dev_info; |
bddd87c7 | 33 | struct bio_list; |
73c10101 | 34 | struct blk_plug; |
c7af7877 | 35 | struct cfs_rq; |
c7af7877 IM |
36 | struct fs_struct; |
37 | struct futex_pi_state; | |
38 | struct io_context; | |
39 | struct mempolicy; | |
89076bc3 | 40 | struct nameidata; |
c7af7877 IM |
41 | struct nsproxy; |
42 | struct perf_event_context; | |
43 | struct pid_namespace; | |
44 | struct pipe_inode_info; | |
45 | struct rcu_node; | |
46 | struct reclaim_state; | |
47 | struct robust_list_head; | |
48 | struct sched_attr; | |
49 | struct sched_param; | |
43ae34cb | 50 | struct seq_file; |
c7af7877 IM |
51 | struct sighand_struct; |
52 | struct signal_struct; | |
53 | struct task_delay_info; | |
4cf86d77 | 54 | struct task_group; |
1da177e4 | 55 | |
4a8342d2 LT |
56 | /* |
57 | * Task state bitmask. NOTE! These bits are also | |
58 | * encoded in fs/proc/array.c: get_task_state(). | |
59 | * | |
60 | * We have two separate sets of flags: task->state | |
61 | * is about runnability, while task->exit_state are | |
62 | * about the task exiting. Confusing, but this way | |
63 | * modifying one set can't modify the other one by | |
64 | * mistake. | |
65 | */ | |
5eca1c10 IM |
66 | |
67 | /* Used in tsk->state: */ | |
68 | #define TASK_RUNNING 0 | |
69 | #define TASK_INTERRUPTIBLE 1 | |
70 | #define TASK_UNINTERRUPTIBLE 2 | |
71 | #define __TASK_STOPPED 4 | |
72 | #define __TASK_TRACED 8 | |
73 | /* Used in tsk->exit_state: */ | |
74 | #define EXIT_DEAD 16 | |
75 | #define EXIT_ZOMBIE 32 | |
76 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) | |
77 | /* Used in tsk->state again: */ | |
78 | #define TASK_DEAD 64 | |
79 | #define TASK_WAKEKILL 128 | |
80 | #define TASK_WAKING 256 | |
81 | #define TASK_PARKED 512 | |
82 | #define TASK_NOLOAD 1024 | |
83 | #define TASK_NEW 2048 | |
84 | #define TASK_STATE_MAX 4096 | |
85 | ||
86 | #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn" | |
87 | ||
88 | /* Convenience macros for the sake of set_current_state: */ | |
89 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) | |
90 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
91 | #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) | |
92 | ||
93 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) | |
94 | ||
95 | /* Convenience macros for the sake of wake_up(): */ | |
96 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
97 | #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) | |
98 | ||
99 | /* get_task_state(): */ | |
100 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
101 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ | |
102 | __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD) | |
103 | ||
104 | #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) | |
105 | ||
106 | #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) | |
107 | ||
108 | #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) | |
109 | ||
110 | #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ | |
111 | (task->flags & PF_FROZEN) == 0 && \ | |
112 | (task->state & TASK_NOLOAD) == 0) | |
1da177e4 | 113 | |
8eb23b9f PZ |
114 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
115 | ||
8eb23b9f PZ |
116 | #define __set_current_state(state_value) \ |
117 | do { \ | |
118 | current->task_state_change = _THIS_IP_; \ | |
119 | current->state = (state_value); \ | |
120 | } while (0) | |
121 | #define set_current_state(state_value) \ | |
122 | do { \ | |
123 | current->task_state_change = _THIS_IP_; \ | |
a2250238 | 124 | smp_store_mb(current->state, (state_value)); \ |
8eb23b9f PZ |
125 | } while (0) |
126 | ||
127 | #else | |
498d0c57 AM |
128 | /* |
129 | * set_current_state() includes a barrier so that the write of current->state | |
130 | * is correctly serialised wrt the caller's subsequent test of whether to | |
131 | * actually sleep: | |
132 | * | |
a2250238 | 133 | * for (;;) { |
498d0c57 | 134 | * set_current_state(TASK_UNINTERRUPTIBLE); |
a2250238 PZ |
135 | * if (!need_sleep) |
136 | * break; | |
137 | * | |
138 | * schedule(); | |
139 | * } | |
140 | * __set_current_state(TASK_RUNNING); | |
141 | * | |
142 | * If the caller does not need such serialisation (because, for instance, the | |
143 | * condition test and condition change and wakeup are under the same lock) then | |
144 | * use __set_current_state(). | |
145 | * | |
146 | * The above is typically ordered against the wakeup, which does: | |
147 | * | |
148 | * need_sleep = false; | |
149 | * wake_up_state(p, TASK_UNINTERRUPTIBLE); | |
150 | * | |
151 | * Where wake_up_state() (and all other wakeup primitives) imply enough | |
152 | * barriers to order the store of the variable against wakeup. | |
153 | * | |
154 | * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, | |
155 | * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a | |
156 | * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). | |
498d0c57 | 157 | * |
a2250238 | 158 | * This is obviously fine, since they both store the exact same value. |
498d0c57 | 159 | * |
a2250238 | 160 | * Also see the comments of try_to_wake_up(). |
498d0c57 | 161 | */ |
5eca1c10 IM |
162 | #define __set_current_state(state_value) do { current->state = (state_value); } while (0) |
163 | #define set_current_state(state_value) smp_store_mb(current->state, (state_value)) | |
8eb23b9f PZ |
164 | #endif |
165 | ||
5eca1c10 IM |
166 | /* Task command name length: */ |
167 | #define TASK_COMM_LEN 16 | |
1da177e4 | 168 | |
5eca1c10 | 169 | extern cpumask_var_t cpu_isolated_map; |
3fa0818b | 170 | |
1da177e4 LT |
171 | extern void scheduler_tick(void); |
172 | ||
5eca1c10 IM |
173 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX |
174 | ||
175 | extern long schedule_timeout(long timeout); | |
176 | extern long schedule_timeout_interruptible(long timeout); | |
177 | extern long schedule_timeout_killable(long timeout); | |
178 | extern long schedule_timeout_uninterruptible(long timeout); | |
179 | extern long schedule_timeout_idle(long timeout); | |
1da177e4 | 180 | asmlinkage void schedule(void); |
c5491ea7 | 181 | extern void schedule_preempt_disabled(void); |
1da177e4 | 182 | |
10ab5643 TH |
183 | extern int __must_check io_schedule_prepare(void); |
184 | extern void io_schedule_finish(int token); | |
9cff8ade | 185 | extern long io_schedule_timeout(long timeout); |
10ab5643 | 186 | extern void io_schedule(void); |
9cff8ade | 187 | |
d37f761d | 188 | /** |
0ba42a59 | 189 | * struct prev_cputime - snapshot of system and user cputime |
d37f761d FW |
190 | * @utime: time spent in user mode |
191 | * @stime: time spent in system mode | |
9d7fb042 | 192 | * @lock: protects the above two fields |
d37f761d | 193 | * |
9d7fb042 PZ |
194 | * Stores previous user/system time values such that we can guarantee |
195 | * monotonicity. | |
d37f761d | 196 | */ |
9d7fb042 PZ |
197 | struct prev_cputime { |
198 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
5eca1c10 IM |
199 | u64 utime; |
200 | u64 stime; | |
201 | raw_spinlock_t lock; | |
9d7fb042 | 202 | #endif |
d37f761d FW |
203 | }; |
204 | ||
f06febc9 FM |
205 | /** |
206 | * struct task_cputime - collected CPU time counts | |
5613fda9 FW |
207 | * @utime: time spent in user mode, in nanoseconds |
208 | * @stime: time spent in kernel mode, in nanoseconds | |
f06febc9 | 209 | * @sum_exec_runtime: total time spent on the CPU, in nanoseconds |
5ce73a4a | 210 | * |
9d7fb042 PZ |
211 | * This structure groups together three kinds of CPU time that are tracked for |
212 | * threads and thread groups. Most things considering CPU time want to group | |
213 | * these counts together and treat all three of them in parallel. | |
f06febc9 FM |
214 | */ |
215 | struct task_cputime { | |
5eca1c10 IM |
216 | u64 utime; |
217 | u64 stime; | |
218 | unsigned long long sum_exec_runtime; | |
f06febc9 | 219 | }; |
9d7fb042 | 220 | |
5eca1c10 IM |
221 | /* Alternate field names when used on cache expirations: */ |
222 | #define virt_exp utime | |
223 | #define prof_exp stime | |
224 | #define sched_exp sum_exec_runtime | |
f06febc9 | 225 | |
bac5b6b6 FW |
226 | enum vtime_state { |
227 | /* Task is sleeping or running in a CPU with VTIME inactive: */ | |
228 | VTIME_INACTIVE = 0, | |
229 | /* Task runs in userspace in a CPU with VTIME active: */ | |
230 | VTIME_USER, | |
231 | /* Task runs in kernelspace in a CPU with VTIME active: */ | |
232 | VTIME_SYS, | |
233 | }; | |
234 | ||
235 | struct vtime { | |
236 | seqcount_t seqcount; | |
237 | unsigned long long starttime; | |
238 | enum vtime_state state; | |
2a42eb95 WL |
239 | u64 utime; |
240 | u64 stime; | |
241 | u64 gtime; | |
bac5b6b6 FW |
242 | }; |
243 | ||
1da177e4 | 244 | struct sched_info { |
7f5f8e8d | 245 | #ifdef CONFIG_SCHED_INFO |
5eca1c10 IM |
246 | /* Cumulative counters: */ |
247 | ||
248 | /* # of times we have run on this CPU: */ | |
249 | unsigned long pcount; | |
250 | ||
251 | /* Time spent waiting on a runqueue: */ | |
252 | unsigned long long run_delay; | |
253 | ||
254 | /* Timestamps: */ | |
255 | ||
256 | /* When did we last run on a CPU? */ | |
257 | unsigned long long last_arrival; | |
258 | ||
259 | /* When were we last queued to run? */ | |
260 | unsigned long long last_queued; | |
1da177e4 | 261 | |
f6db8347 | 262 | #endif /* CONFIG_SCHED_INFO */ |
7f5f8e8d | 263 | }; |
1da177e4 | 264 | |
6ecdd749 YD |
265 | /* |
266 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
267 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
268 | * | |
269 | * We define a basic fixed point arithmetic range, and then formalize | |
270 | * all these metrics based on that basic range. | |
271 | */ | |
5eca1c10 IM |
272 | # define SCHED_FIXEDPOINT_SHIFT 10 |
273 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
6ecdd749 | 274 | |
20b8a59f | 275 | struct load_weight { |
5eca1c10 IM |
276 | unsigned long weight; |
277 | u32 inv_weight; | |
20b8a59f IM |
278 | }; |
279 | ||
9d89c257 | 280 | /* |
7b595334 YD |
281 | * The load_avg/util_avg accumulates an infinite geometric series |
282 | * (see __update_load_avg() in kernel/sched/fair.c). | |
283 | * | |
284 | * [load_avg definition] | |
285 | * | |
286 | * load_avg = runnable% * scale_load_down(load) | |
287 | * | |
288 | * where runnable% is the time ratio that a sched_entity is runnable. | |
289 | * For cfs_rq, it is the aggregated load_avg of all runnable and | |
9d89c257 | 290 | * blocked sched_entities. |
7b595334 YD |
291 | * |
292 | * load_avg may also take frequency scaling into account: | |
293 | * | |
294 | * load_avg = runnable% * scale_load_down(load) * freq% | |
295 | * | |
296 | * where freq% is the CPU frequency normalized to the highest frequency. | |
297 | * | |
298 | * [util_avg definition] | |
299 | * | |
300 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
301 | * | |
302 | * where running% is the time ratio that a sched_entity is running on | |
303 | * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable | |
304 | * and blocked sched_entities. | |
305 | * | |
306 | * util_avg may also factor frequency scaling and CPU capacity scaling: | |
307 | * | |
308 | * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity% | |
309 | * | |
310 | * where freq% is the same as above, and capacity% is the CPU capacity | |
311 | * normalized to the greatest capacity (due to uarch differences, etc). | |
312 | * | |
313 | * N.B., the above ratios (runnable%, running%, freq%, and capacity%) | |
314 | * themselves are in the range of [0, 1]. To do fixed point arithmetics, | |
315 | * we therefore scale them to as large a range as necessary. This is for | |
316 | * example reflected by util_avg's SCHED_CAPACITY_SCALE. | |
317 | * | |
318 | * [Overflow issue] | |
319 | * | |
320 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
321 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
322 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
323 | * | |
324 | * For all other cases (including 32-bit kernels), struct load_weight's | |
325 | * weight will overflow first before we do, because: | |
326 | * | |
327 | * Max(load_avg) <= Max(load.weight) | |
328 | * | |
329 | * Then it is the load_weight's responsibility to consider overflow | |
330 | * issues. | |
9d89c257 | 331 | */ |
9d85f21c | 332 | struct sched_avg { |
5eca1c10 IM |
333 | u64 last_update_time; |
334 | u64 load_sum; | |
335 | u32 util_sum; | |
336 | u32 period_contrib; | |
337 | unsigned long load_avg; | |
338 | unsigned long util_avg; | |
9d85f21c PT |
339 | }; |
340 | ||
41acab88 | 341 | struct sched_statistics { |
7f5f8e8d | 342 | #ifdef CONFIG_SCHEDSTATS |
5eca1c10 IM |
343 | u64 wait_start; |
344 | u64 wait_max; | |
345 | u64 wait_count; | |
346 | u64 wait_sum; | |
347 | u64 iowait_count; | |
348 | u64 iowait_sum; | |
349 | ||
350 | u64 sleep_start; | |
351 | u64 sleep_max; | |
352 | s64 sum_sleep_runtime; | |
353 | ||
354 | u64 block_start; | |
355 | u64 block_max; | |
356 | u64 exec_max; | |
357 | u64 slice_max; | |
358 | ||
359 | u64 nr_migrations_cold; | |
360 | u64 nr_failed_migrations_affine; | |
361 | u64 nr_failed_migrations_running; | |
362 | u64 nr_failed_migrations_hot; | |
363 | u64 nr_forced_migrations; | |
364 | ||
365 | u64 nr_wakeups; | |
366 | u64 nr_wakeups_sync; | |
367 | u64 nr_wakeups_migrate; | |
368 | u64 nr_wakeups_local; | |
369 | u64 nr_wakeups_remote; | |
370 | u64 nr_wakeups_affine; | |
371 | u64 nr_wakeups_affine_attempts; | |
372 | u64 nr_wakeups_passive; | |
373 | u64 nr_wakeups_idle; | |
41acab88 | 374 | #endif |
7f5f8e8d | 375 | }; |
41acab88 LDM |
376 | |
377 | struct sched_entity { | |
5eca1c10 IM |
378 | /* For load-balancing: */ |
379 | struct load_weight load; | |
380 | struct rb_node run_node; | |
381 | struct list_head group_node; | |
382 | unsigned int on_rq; | |
41acab88 | 383 | |
5eca1c10 IM |
384 | u64 exec_start; |
385 | u64 sum_exec_runtime; | |
386 | u64 vruntime; | |
387 | u64 prev_sum_exec_runtime; | |
41acab88 | 388 | |
5eca1c10 | 389 | u64 nr_migrations; |
41acab88 | 390 | |
5eca1c10 | 391 | struct sched_statistics statistics; |
94c18227 | 392 | |
20b8a59f | 393 | #ifdef CONFIG_FAIR_GROUP_SCHED |
5eca1c10 IM |
394 | int depth; |
395 | struct sched_entity *parent; | |
20b8a59f | 396 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 397 | struct cfs_rq *cfs_rq; |
20b8a59f | 398 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 399 | struct cfs_rq *my_q; |
20b8a59f | 400 | #endif |
8bd75c77 | 401 | |
141965c7 | 402 | #ifdef CONFIG_SMP |
5a107804 JO |
403 | /* |
404 | * Per entity load average tracking. | |
405 | * | |
406 | * Put into separate cache line so it does not | |
407 | * collide with read-mostly values above. | |
408 | */ | |
5eca1c10 | 409 | struct sched_avg avg ____cacheline_aligned_in_smp; |
9d85f21c | 410 | #endif |
20b8a59f | 411 | }; |
70b97a7f | 412 | |
fa717060 | 413 | struct sched_rt_entity { |
5eca1c10 IM |
414 | struct list_head run_list; |
415 | unsigned long timeout; | |
416 | unsigned long watchdog_stamp; | |
417 | unsigned int time_slice; | |
418 | unsigned short on_rq; | |
419 | unsigned short on_list; | |
420 | ||
421 | struct sched_rt_entity *back; | |
052f1dc7 | 422 | #ifdef CONFIG_RT_GROUP_SCHED |
5eca1c10 | 423 | struct sched_rt_entity *parent; |
6f505b16 | 424 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 425 | struct rt_rq *rt_rq; |
6f505b16 | 426 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 427 | struct rt_rq *my_q; |
6f505b16 | 428 | #endif |
fa717060 PZ |
429 | }; |
430 | ||
aab03e05 | 431 | struct sched_dl_entity { |
5eca1c10 | 432 | struct rb_node rb_node; |
aab03e05 DF |
433 | |
434 | /* | |
435 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 436 | * during sched_setattr(), they will remain the same until |
437 | * the next sched_setattr(). | |
aab03e05 | 438 | */ |
5eca1c10 IM |
439 | u64 dl_runtime; /* Maximum runtime for each instance */ |
440 | u64 dl_deadline; /* Relative deadline of each instance */ | |
441 | u64 dl_period; /* Separation of two instances (period) */ | |
54d6d303 | 442 | u64 dl_bw; /* dl_runtime / dl_period */ |
3effcb42 | 443 | u64 dl_density; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
444 | |
445 | /* | |
446 | * Actual scheduling parameters. Initialized with the values above, | |
447 | * they are continously updated during task execution. Note that | |
448 | * the remaining runtime could be < 0 in case we are in overrun. | |
449 | */ | |
5eca1c10 IM |
450 | s64 runtime; /* Remaining runtime for this instance */ |
451 | u64 deadline; /* Absolute deadline for this instance */ | |
452 | unsigned int flags; /* Specifying the scheduler behaviour */ | |
aab03e05 DF |
453 | |
454 | /* | |
455 | * Some bool flags: | |
456 | * | |
457 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
458 | * task has to wait for a replenishment to be performed at the | |
459 | * next firing of dl_timer. | |
460 | * | |
2d3d891d DF |
461 | * @dl_boosted tells if we are boosted due to DI. If so we are |
462 | * outside bandwidth enforcement mechanism (but only until we | |
5bfd126e JL |
463 | * exit the critical section); |
464 | * | |
5eca1c10 | 465 | * @dl_yielded tells if task gave up the CPU before consuming |
5bfd126e | 466 | * all its available runtime during the last job. |
209a0cbd LA |
467 | * |
468 | * @dl_non_contending tells if the task is inactive while still | |
469 | * contributing to the active utilization. In other words, it | |
470 | * indicates if the inactive timer has been armed and its handler | |
471 | * has not been executed yet. This flag is useful to avoid race | |
472 | * conditions between the inactive timer handler and the wakeup | |
473 | * code. | |
aab03e05 | 474 | */ |
5eca1c10 IM |
475 | int dl_throttled; |
476 | int dl_boosted; | |
477 | int dl_yielded; | |
209a0cbd | 478 | int dl_non_contending; |
aab03e05 DF |
479 | |
480 | /* | |
481 | * Bandwidth enforcement timer. Each -deadline task has its | |
482 | * own bandwidth to be enforced, thus we need one timer per task. | |
483 | */ | |
5eca1c10 | 484 | struct hrtimer dl_timer; |
209a0cbd LA |
485 | |
486 | /* | |
487 | * Inactive timer, responsible for decreasing the active utilization | |
488 | * at the "0-lag time". When a -deadline task blocks, it contributes | |
489 | * to GRUB's active utilization until the "0-lag time", hence a | |
490 | * timer is needed to decrease the active utilization at the correct | |
491 | * time. | |
492 | */ | |
493 | struct hrtimer inactive_timer; | |
aab03e05 | 494 | }; |
8bd75c77 | 495 | |
1d082fd0 PM |
496 | union rcu_special { |
497 | struct { | |
5eca1c10 IM |
498 | u8 blocked; |
499 | u8 need_qs; | |
500 | u8 exp_need_qs; | |
501 | ||
502 | /* Otherwise the compiler can store garbage here: */ | |
503 | u8 pad; | |
8203d6d0 PM |
504 | } b; /* Bits. */ |
505 | u32 s; /* Set of bits. */ | |
1d082fd0 | 506 | }; |
86848966 | 507 | |
8dc85d54 PZ |
508 | enum perf_event_task_context { |
509 | perf_invalid_context = -1, | |
510 | perf_hw_context = 0, | |
89a1e187 | 511 | perf_sw_context, |
8dc85d54 PZ |
512 | perf_nr_task_contexts, |
513 | }; | |
514 | ||
eb61baf6 IM |
515 | struct wake_q_node { |
516 | struct wake_q_node *next; | |
517 | }; | |
518 | ||
1da177e4 | 519 | struct task_struct { |
c65eacbe AL |
520 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
521 | /* | |
522 | * For reasons of header soup (see current_thread_info()), this | |
523 | * must be the first element of task_struct. | |
524 | */ | |
5eca1c10 | 525 | struct thread_info thread_info; |
c65eacbe | 526 | #endif |
5eca1c10 IM |
527 | /* -1 unrunnable, 0 runnable, >0 stopped: */ |
528 | volatile long state; | |
529 | void *stack; | |
530 | atomic_t usage; | |
531 | /* Per task flags (PF_*), defined further below: */ | |
532 | unsigned int flags; | |
533 | unsigned int ptrace; | |
1da177e4 | 534 | |
2dd73a4f | 535 | #ifdef CONFIG_SMP |
5eca1c10 IM |
536 | struct llist_node wake_entry; |
537 | int on_cpu; | |
c65eacbe | 538 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 IM |
539 | /* Current CPU: */ |
540 | unsigned int cpu; | |
c65eacbe | 541 | #endif |
5eca1c10 IM |
542 | unsigned int wakee_flips; |
543 | unsigned long wakee_flip_decay_ts; | |
544 | struct task_struct *last_wakee; | |
ac66f547 | 545 | |
5eca1c10 | 546 | int wake_cpu; |
2dd73a4f | 547 | #endif |
5eca1c10 IM |
548 | int on_rq; |
549 | ||
550 | int prio; | |
551 | int static_prio; | |
552 | int normal_prio; | |
553 | unsigned int rt_priority; | |
50e645a8 | 554 | |
5eca1c10 IM |
555 | const struct sched_class *sched_class; |
556 | struct sched_entity se; | |
557 | struct sched_rt_entity rt; | |
8323f26c | 558 | #ifdef CONFIG_CGROUP_SCHED |
5eca1c10 | 559 | struct task_group *sched_task_group; |
8323f26c | 560 | #endif |
5eca1c10 | 561 | struct sched_dl_entity dl; |
1da177e4 | 562 | |
e107be36 | 563 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
5eca1c10 IM |
564 | /* List of struct preempt_notifier: */ |
565 | struct hlist_head preempt_notifiers; | |
e107be36 AK |
566 | #endif |
567 | ||
6c5c9341 | 568 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
5eca1c10 | 569 | unsigned int btrace_seq; |
6c5c9341 | 570 | #endif |
1da177e4 | 571 | |
5eca1c10 IM |
572 | unsigned int policy; |
573 | int nr_cpus_allowed; | |
574 | cpumask_t cpus_allowed; | |
1da177e4 | 575 | |
a57eb940 | 576 | #ifdef CONFIG_PREEMPT_RCU |
5eca1c10 IM |
577 | int rcu_read_lock_nesting; |
578 | union rcu_special rcu_read_unlock_special; | |
579 | struct list_head rcu_node_entry; | |
580 | struct rcu_node *rcu_blocked_node; | |
28f6569a | 581 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
5eca1c10 | 582 | |
8315f422 | 583 | #ifdef CONFIG_TASKS_RCU |
5eca1c10 IM |
584 | unsigned long rcu_tasks_nvcsw; |
585 | bool rcu_tasks_holdout; | |
586 | struct list_head rcu_tasks_holdout_list; | |
587 | int rcu_tasks_idle_cpu; | |
8315f422 | 588 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 589 | |
5eca1c10 | 590 | struct sched_info sched_info; |
1da177e4 | 591 | |
5eca1c10 | 592 | struct list_head tasks; |
806c09a7 | 593 | #ifdef CONFIG_SMP |
5eca1c10 IM |
594 | struct plist_node pushable_tasks; |
595 | struct rb_node pushable_dl_tasks; | |
806c09a7 | 596 | #endif |
1da177e4 | 597 | |
5eca1c10 IM |
598 | struct mm_struct *mm; |
599 | struct mm_struct *active_mm; | |
314ff785 IM |
600 | |
601 | /* Per-thread vma caching: */ | |
5eca1c10 | 602 | struct vmacache vmacache; |
314ff785 | 603 | |
5eca1c10 IM |
604 | #ifdef SPLIT_RSS_COUNTING |
605 | struct task_rss_stat rss_stat; | |
34e55232 | 606 | #endif |
5eca1c10 IM |
607 | int exit_state; |
608 | int exit_code; | |
609 | int exit_signal; | |
610 | /* The signal sent when the parent dies: */ | |
611 | int pdeath_signal; | |
612 | /* JOBCTL_*, siglock protected: */ | |
613 | unsigned long jobctl; | |
614 | ||
615 | /* Used for emulating ABI behavior of previous Linux versions: */ | |
616 | unsigned int personality; | |
617 | ||
618 | /* Scheduler bits, serialized by scheduler locks: */ | |
619 | unsigned sched_reset_on_fork:1; | |
620 | unsigned sched_contributes_to_load:1; | |
621 | unsigned sched_migrated:1; | |
622 | unsigned sched_remote_wakeup:1; | |
623 | /* Force alignment to the next boundary: */ | |
624 | unsigned :0; | |
625 | ||
626 | /* Unserialized, strictly 'current' */ | |
627 | ||
628 | /* Bit to tell LSMs we're in execve(): */ | |
629 | unsigned in_execve:1; | |
630 | unsigned in_iowait:1; | |
631 | #ifndef TIF_RESTORE_SIGMASK | |
632 | unsigned restore_sigmask:1; | |
7e781418 | 633 | #endif |
626ebc41 | 634 | #ifdef CONFIG_MEMCG |
5eca1c10 | 635 | unsigned memcg_may_oom:1; |
127424c8 | 636 | #ifndef CONFIG_SLOB |
5eca1c10 | 637 | unsigned memcg_kmem_skip_account:1; |
6f185c29 | 638 | #endif |
127424c8 | 639 | #endif |
ff303e66 | 640 | #ifdef CONFIG_COMPAT_BRK |
5eca1c10 | 641 | unsigned brk_randomized:1; |
ff303e66 | 642 | #endif |
77f88796 TH |
643 | #ifdef CONFIG_CGROUPS |
644 | /* disallow userland-initiated cgroup migration */ | |
645 | unsigned no_cgroup_migration:1; | |
646 | #endif | |
6f185c29 | 647 | |
5eca1c10 | 648 | unsigned long atomic_flags; /* Flags requiring atomic access. */ |
1d4457f9 | 649 | |
5eca1c10 | 650 | struct restart_block restart_block; |
f56141e3 | 651 | |
5eca1c10 IM |
652 | pid_t pid; |
653 | pid_t tgid; | |
0a425405 | 654 | |
1314562a | 655 | #ifdef CONFIG_CC_STACKPROTECTOR |
5eca1c10 IM |
656 | /* Canary value for the -fstack-protector GCC feature: */ |
657 | unsigned long stack_canary; | |
1314562a | 658 | #endif |
4d1d61a6 | 659 | /* |
5eca1c10 | 660 | * Pointers to the (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 661 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 662 | * p->real_parent->pid) |
1da177e4 | 663 | */ |
5eca1c10 IM |
664 | |
665 | /* Real parent process: */ | |
666 | struct task_struct __rcu *real_parent; | |
667 | ||
668 | /* Recipient of SIGCHLD, wait4() reports: */ | |
669 | struct task_struct __rcu *parent; | |
670 | ||
1da177e4 | 671 | /* |
5eca1c10 | 672 | * Children/sibling form the list of natural children: |
1da177e4 | 673 | */ |
5eca1c10 IM |
674 | struct list_head children; |
675 | struct list_head sibling; | |
676 | struct task_struct *group_leader; | |
1da177e4 | 677 | |
f470021a | 678 | /* |
5eca1c10 IM |
679 | * 'ptraced' is the list of tasks this task is using ptrace() on. |
680 | * | |
f470021a | 681 | * This includes both natural children and PTRACE_ATTACH targets. |
5eca1c10 | 682 | * 'ptrace_entry' is this task's link on the p->parent->ptraced list. |
f470021a | 683 | */ |
5eca1c10 IM |
684 | struct list_head ptraced; |
685 | struct list_head ptrace_entry; | |
f470021a | 686 | |
1da177e4 | 687 | /* PID/PID hash table linkage. */ |
5eca1c10 IM |
688 | struct pid_link pids[PIDTYPE_MAX]; |
689 | struct list_head thread_group; | |
690 | struct list_head thread_node; | |
691 | ||
692 | struct completion *vfork_done; | |
1da177e4 | 693 | |
5eca1c10 IM |
694 | /* CLONE_CHILD_SETTID: */ |
695 | int __user *set_child_tid; | |
1da177e4 | 696 | |
5eca1c10 IM |
697 | /* CLONE_CHILD_CLEARTID: */ |
698 | int __user *clear_child_tid; | |
699 | ||
700 | u64 utime; | |
701 | u64 stime; | |
40565b5a | 702 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME |
5eca1c10 IM |
703 | u64 utimescaled; |
704 | u64 stimescaled; | |
40565b5a | 705 | #endif |
5eca1c10 IM |
706 | u64 gtime; |
707 | struct prev_cputime prev_cputime; | |
6a61671b | 708 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
bac5b6b6 | 709 | struct vtime vtime; |
d99ca3b9 | 710 | #endif |
d027d45d FW |
711 | |
712 | #ifdef CONFIG_NO_HZ_FULL | |
5eca1c10 | 713 | atomic_t tick_dep_mask; |
d027d45d | 714 | #endif |
5eca1c10 IM |
715 | /* Context switch counts: */ |
716 | unsigned long nvcsw; | |
717 | unsigned long nivcsw; | |
718 | ||
719 | /* Monotonic time in nsecs: */ | |
720 | u64 start_time; | |
721 | ||
722 | /* Boot based time in nsecs: */ | |
723 | u64 real_start_time; | |
724 | ||
725 | /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */ | |
726 | unsigned long min_flt; | |
727 | unsigned long maj_flt; | |
1da177e4 | 728 | |
b18b6a9c | 729 | #ifdef CONFIG_POSIX_TIMERS |
5eca1c10 IM |
730 | struct task_cputime cputime_expires; |
731 | struct list_head cpu_timers[3]; | |
b18b6a9c | 732 | #endif |
1da177e4 | 733 | |
5eca1c10 IM |
734 | /* Process credentials: */ |
735 | ||
736 | /* Tracer's credentials at attach: */ | |
737 | const struct cred __rcu *ptracer_cred; | |
738 | ||
739 | /* Objective and real subjective task credentials (COW): */ | |
740 | const struct cred __rcu *real_cred; | |
741 | ||
742 | /* Effective (overridable) subjective task credentials (COW): */ | |
743 | const struct cred __rcu *cred; | |
744 | ||
745 | /* | |
746 | * executable name, excluding path. | |
747 | * | |
748 | * - normally initialized setup_new_exec() | |
749 | * - access it with [gs]et_task_comm() | |
750 | * - lock it with task_lock() | |
751 | */ | |
752 | char comm[TASK_COMM_LEN]; | |
753 | ||
754 | struct nameidata *nameidata; | |
755 | ||
3d5b6fcc | 756 | #ifdef CONFIG_SYSVIPC |
5eca1c10 IM |
757 | struct sysv_sem sysvsem; |
758 | struct sysv_shm sysvshm; | |
3d5b6fcc | 759 | #endif |
e162b39a | 760 | #ifdef CONFIG_DETECT_HUNG_TASK |
5eca1c10 | 761 | unsigned long last_switch_count; |
82a1fcb9 | 762 | #endif |
5eca1c10 IM |
763 | /* Filesystem information: */ |
764 | struct fs_struct *fs; | |
765 | ||
766 | /* Open file information: */ | |
767 | struct files_struct *files; | |
768 | ||
769 | /* Namespaces: */ | |
770 | struct nsproxy *nsproxy; | |
771 | ||
772 | /* Signal handlers: */ | |
773 | struct signal_struct *signal; | |
774 | struct sighand_struct *sighand; | |
775 | sigset_t blocked; | |
776 | sigset_t real_blocked; | |
777 | /* Restored if set_restore_sigmask() was used: */ | |
778 | sigset_t saved_sigmask; | |
779 | struct sigpending pending; | |
780 | unsigned long sas_ss_sp; | |
781 | size_t sas_ss_size; | |
782 | unsigned int sas_ss_flags; | |
783 | ||
784 | struct callback_head *task_works; | |
785 | ||
786 | struct audit_context *audit_context; | |
bfef93a5 | 787 | #ifdef CONFIG_AUDITSYSCALL |
5eca1c10 IM |
788 | kuid_t loginuid; |
789 | unsigned int sessionid; | |
bfef93a5 | 790 | #endif |
5eca1c10 IM |
791 | struct seccomp seccomp; |
792 | ||
793 | /* Thread group tracking: */ | |
794 | u32 parent_exec_id; | |
795 | u32 self_exec_id; | |
1da177e4 | 796 | |
5eca1c10 IM |
797 | /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */ |
798 | spinlock_t alloc_lock; | |
1da177e4 | 799 | |
b29739f9 | 800 | /* Protection of the PI data structures: */ |
5eca1c10 | 801 | raw_spinlock_t pi_lock; |
b29739f9 | 802 | |
5eca1c10 | 803 | struct wake_q_node wake_q; |
76751049 | 804 | |
23f78d4a | 805 | #ifdef CONFIG_RT_MUTEXES |
5eca1c10 IM |
806 | /* PI waiters blocked on a rt_mutex held by this task: */ |
807 | struct rb_root pi_waiters; | |
808 | struct rb_node *pi_waiters_leftmost; | |
e96a7705 XP |
809 | /* Updated under owner's pi_lock and rq lock */ |
810 | struct task_struct *pi_top_task; | |
5eca1c10 IM |
811 | /* Deadlock detection and priority inheritance handling: */ |
812 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
813 | #endif |
814 | ||
408894ee | 815 | #ifdef CONFIG_DEBUG_MUTEXES |
5eca1c10 IM |
816 | /* Mutex deadlock detection: */ |
817 | struct mutex_waiter *blocked_on; | |
408894ee | 818 | #endif |
5eca1c10 | 819 | |
de30a2b3 | 820 | #ifdef CONFIG_TRACE_IRQFLAGS |
5eca1c10 IM |
821 | unsigned int irq_events; |
822 | unsigned long hardirq_enable_ip; | |
823 | unsigned long hardirq_disable_ip; | |
824 | unsigned int hardirq_enable_event; | |
825 | unsigned int hardirq_disable_event; | |
826 | int hardirqs_enabled; | |
827 | int hardirq_context; | |
828 | unsigned long softirq_disable_ip; | |
829 | unsigned long softirq_enable_ip; | |
830 | unsigned int softirq_disable_event; | |
831 | unsigned int softirq_enable_event; | |
832 | int softirqs_enabled; | |
833 | int softirq_context; | |
de30a2b3 | 834 | #endif |
5eca1c10 | 835 | |
fbb9ce95 | 836 | #ifdef CONFIG_LOCKDEP |
5eca1c10 IM |
837 | # define MAX_LOCK_DEPTH 48UL |
838 | u64 curr_chain_key; | |
839 | int lockdep_depth; | |
840 | unsigned int lockdep_recursion; | |
841 | struct held_lock held_locks[MAX_LOCK_DEPTH]; | |
842 | gfp_t lockdep_reclaim_gfp; | |
fbb9ce95 | 843 | #endif |
5eca1c10 | 844 | |
c6d30853 | 845 | #ifdef CONFIG_UBSAN |
5eca1c10 | 846 | unsigned int in_ubsan; |
c6d30853 | 847 | #endif |
408894ee | 848 | |
5eca1c10 IM |
849 | /* Journalling filesystem info: */ |
850 | void *journal_info; | |
1da177e4 | 851 | |
5eca1c10 IM |
852 | /* Stacked block device info: */ |
853 | struct bio_list *bio_list; | |
d89d8796 | 854 | |
73c10101 | 855 | #ifdef CONFIG_BLOCK |
5eca1c10 IM |
856 | /* Stack plugging: */ |
857 | struct blk_plug *plug; | |
73c10101 JA |
858 | #endif |
859 | ||
5eca1c10 IM |
860 | /* VM state: */ |
861 | struct reclaim_state *reclaim_state; | |
862 | ||
863 | struct backing_dev_info *backing_dev_info; | |
1da177e4 | 864 | |
5eca1c10 | 865 | struct io_context *io_context; |
1da177e4 | 866 | |
5eca1c10 IM |
867 | /* Ptrace state: */ |
868 | unsigned long ptrace_message; | |
869 | siginfo_t *last_siginfo; | |
1da177e4 | 870 | |
5eca1c10 IM |
871 | struct task_io_accounting ioac; |
872 | #ifdef CONFIG_TASK_XACCT | |
873 | /* Accumulated RSS usage: */ | |
874 | u64 acct_rss_mem1; | |
875 | /* Accumulated virtual memory usage: */ | |
876 | u64 acct_vm_mem1; | |
877 | /* stime + utime since last update: */ | |
878 | u64 acct_timexpd; | |
1da177e4 LT |
879 | #endif |
880 | #ifdef CONFIG_CPUSETS | |
5eca1c10 IM |
881 | /* Protected by ->alloc_lock: */ |
882 | nodemask_t mems_allowed; | |
883 | /* Seqence number to catch updates: */ | |
884 | seqcount_t mems_allowed_seq; | |
885 | int cpuset_mem_spread_rotor; | |
886 | int cpuset_slab_spread_rotor; | |
1da177e4 | 887 | #endif |
ddbcc7e8 | 888 | #ifdef CONFIG_CGROUPS |
5eca1c10 IM |
889 | /* Control Group info protected by css_set_lock: */ |
890 | struct css_set __rcu *cgroups; | |
891 | /* cg_list protected by css_set_lock and tsk->alloc_lock: */ | |
892 | struct list_head cg_list; | |
ddbcc7e8 | 893 | #endif |
e02737d5 | 894 | #ifdef CONFIG_INTEL_RDT_A |
5eca1c10 | 895 | int closid; |
e02737d5 | 896 | #endif |
42b2dd0a | 897 | #ifdef CONFIG_FUTEX |
5eca1c10 | 898 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
899 | #ifdef CONFIG_COMPAT |
900 | struct compat_robust_list_head __user *compat_robust_list; | |
901 | #endif | |
5eca1c10 IM |
902 | struct list_head pi_state_list; |
903 | struct futex_pi_state *pi_state_cache; | |
c7aceaba | 904 | #endif |
cdd6c482 | 905 | #ifdef CONFIG_PERF_EVENTS |
5eca1c10 IM |
906 | struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; |
907 | struct mutex perf_event_mutex; | |
908 | struct list_head perf_event_list; | |
a63eaf34 | 909 | #endif |
8f47b187 | 910 | #ifdef CONFIG_DEBUG_PREEMPT |
5eca1c10 | 911 | unsigned long preempt_disable_ip; |
8f47b187 | 912 | #endif |
c7aceaba | 913 | #ifdef CONFIG_NUMA |
5eca1c10 IM |
914 | /* Protected by alloc_lock: */ |
915 | struct mempolicy *mempolicy; | |
45816682 | 916 | short il_prev; |
5eca1c10 | 917 | short pref_node_fork; |
42b2dd0a | 918 | #endif |
cbee9f88 | 919 | #ifdef CONFIG_NUMA_BALANCING |
5eca1c10 IM |
920 | int numa_scan_seq; |
921 | unsigned int numa_scan_period; | |
922 | unsigned int numa_scan_period_max; | |
923 | int numa_preferred_nid; | |
924 | unsigned long numa_migrate_retry; | |
925 | /* Migration stamp: */ | |
926 | u64 node_stamp; | |
927 | u64 last_task_numa_placement; | |
928 | u64 last_sum_exec_runtime; | |
929 | struct callback_head numa_work; | |
930 | ||
931 | struct list_head numa_entry; | |
932 | struct numa_group *numa_group; | |
8c8a743c | 933 | |
745d6147 | 934 | /* |
44dba3d5 IM |
935 | * numa_faults is an array split into four regions: |
936 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
937 | * in this precise order. | |
938 | * | |
939 | * faults_memory: Exponential decaying average of faults on a per-node | |
940 | * basis. Scheduling placement decisions are made based on these | |
941 | * counts. The values remain static for the duration of a PTE scan. | |
942 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
943 | * hinting fault was incurred. | |
944 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
945 | * during the current scan window. When the scan completes, the counts | |
946 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 947 | */ |
5eca1c10 IM |
948 | unsigned long *numa_faults; |
949 | unsigned long total_numa_faults; | |
745d6147 | 950 | |
04bb2f94 RR |
951 | /* |
952 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
953 | * scan window were remote/local or failed to migrate. The task scan |
954 | * period is adapted based on the locality of the faults with different | |
955 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 956 | */ |
5eca1c10 | 957 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 958 | |
5eca1c10 | 959 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
960 | #endif /* CONFIG_NUMA_BALANCING */ |
961 | ||
5eca1c10 | 962 | struct tlbflush_unmap_batch tlb_ubc; |
72b252ae | 963 | |
5eca1c10 | 964 | struct rcu_head rcu; |
b92ce558 | 965 | |
5eca1c10 IM |
966 | /* Cache last used pipe for splice(): */ |
967 | struct pipe_inode_info *splice_pipe; | |
5640f768 | 968 | |
5eca1c10 | 969 | struct page_frag task_frag; |
5640f768 | 970 | |
47913d4e IM |
971 | #ifdef CONFIG_TASK_DELAY_ACCT |
972 | struct task_delay_info *delays; | |
f4f154fd | 973 | #endif |
47913d4e | 974 | |
f4f154fd | 975 | #ifdef CONFIG_FAULT_INJECTION |
5eca1c10 | 976 | int make_it_fail; |
ca74e92b | 977 | #endif |
9d823e8f | 978 | /* |
5eca1c10 IM |
979 | * When (nr_dirtied >= nr_dirtied_pause), it's time to call |
980 | * balance_dirty_pages() for a dirty throttling pause: | |
9d823e8f | 981 | */ |
5eca1c10 IM |
982 | int nr_dirtied; |
983 | int nr_dirtied_pause; | |
984 | /* Start of a write-and-pause period: */ | |
985 | unsigned long dirty_paused_when; | |
9d823e8f | 986 | |
9745512c | 987 | #ifdef CONFIG_LATENCYTOP |
5eca1c10 IM |
988 | int latency_record_count; |
989 | struct latency_record latency_record[LT_SAVECOUNT]; | |
9745512c | 990 | #endif |
6976675d | 991 | /* |
5eca1c10 | 992 | * Time slack values; these are used to round up poll() and |
6976675d AV |
993 | * select() etc timeout values. These are in nanoseconds. |
994 | */ | |
5eca1c10 IM |
995 | u64 timer_slack_ns; |
996 | u64 default_timer_slack_ns; | |
f8d570a4 | 997 | |
0b24becc | 998 | #ifdef CONFIG_KASAN |
5eca1c10 | 999 | unsigned int kasan_depth; |
0b24becc | 1000 | #endif |
5eca1c10 | 1001 | |
fb52607a | 1002 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
5eca1c10 IM |
1003 | /* Index of current stored address in ret_stack: */ |
1004 | int curr_ret_stack; | |
1005 | ||
1006 | /* Stack of return addresses for return function tracing: */ | |
1007 | struct ftrace_ret_stack *ret_stack; | |
1008 | ||
1009 | /* Timestamp for last schedule: */ | |
1010 | unsigned long long ftrace_timestamp; | |
1011 | ||
f201ae23 FW |
1012 | /* |
1013 | * Number of functions that haven't been traced | |
5eca1c10 | 1014 | * because of depth overrun: |
f201ae23 | 1015 | */ |
5eca1c10 IM |
1016 | atomic_t trace_overrun; |
1017 | ||
1018 | /* Pause tracing: */ | |
1019 | atomic_t tracing_graph_pause; | |
f201ae23 | 1020 | #endif |
5eca1c10 | 1021 | |
ea4e2bc4 | 1022 | #ifdef CONFIG_TRACING |
5eca1c10 IM |
1023 | /* State flags for use by tracers: */ |
1024 | unsigned long trace; | |
1025 | ||
1026 | /* Bitmask and counter of trace recursion: */ | |
1027 | unsigned long trace_recursion; | |
261842b7 | 1028 | #endif /* CONFIG_TRACING */ |
5eca1c10 | 1029 | |
5c9a8750 | 1030 | #ifdef CONFIG_KCOV |
5eca1c10 IM |
1031 | /* Coverage collection mode enabled for this task (0 if disabled): */ |
1032 | enum kcov_mode kcov_mode; | |
1033 | ||
1034 | /* Size of the kcov_area: */ | |
1035 | unsigned int kcov_size; | |
1036 | ||
1037 | /* Buffer for coverage collection: */ | |
1038 | void *kcov_area; | |
1039 | ||
1040 | /* KCOV descriptor wired with this task or NULL: */ | |
1041 | struct kcov *kcov; | |
5c9a8750 | 1042 | #endif |
5eca1c10 | 1043 | |
6f185c29 | 1044 | #ifdef CONFIG_MEMCG |
5eca1c10 IM |
1045 | struct mem_cgroup *memcg_in_oom; |
1046 | gfp_t memcg_oom_gfp_mask; | |
1047 | int memcg_oom_order; | |
b23afb93 | 1048 | |
5eca1c10 IM |
1049 | /* Number of pages to reclaim on returning to userland: */ |
1050 | unsigned int memcg_nr_pages_over_high; | |
569b846d | 1051 | #endif |
5eca1c10 | 1052 | |
0326f5a9 | 1053 | #ifdef CONFIG_UPROBES |
5eca1c10 | 1054 | struct uprobe_task *utask; |
0326f5a9 | 1055 | #endif |
cafe5635 | 1056 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
5eca1c10 IM |
1057 | unsigned int sequential_io; |
1058 | unsigned int sequential_io_avg; | |
cafe5635 | 1059 | #endif |
8eb23b9f | 1060 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
5eca1c10 | 1061 | unsigned long task_state_change; |
8eb23b9f | 1062 | #endif |
5eca1c10 | 1063 | int pagefault_disabled; |
03049269 | 1064 | #ifdef CONFIG_MMU |
5eca1c10 | 1065 | struct task_struct *oom_reaper_list; |
03049269 | 1066 | #endif |
ba14a194 | 1067 | #ifdef CONFIG_VMAP_STACK |
5eca1c10 | 1068 | struct vm_struct *stack_vm_area; |
ba14a194 | 1069 | #endif |
68f24b08 | 1070 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 IM |
1071 | /* A live task holds one reference: */ |
1072 | atomic_t stack_refcount; | |
d83a7cb3 JP |
1073 | #endif |
1074 | #ifdef CONFIG_LIVEPATCH | |
1075 | int patch_state; | |
0302e28d | 1076 | #endif |
e4e55b47 TH |
1077 | #ifdef CONFIG_SECURITY |
1078 | /* Used by LSM modules for access restriction: */ | |
1079 | void *security; | |
68f24b08 | 1080 | #endif |
5eca1c10 IM |
1081 | /* CPU-specific state of this task: */ |
1082 | struct thread_struct thread; | |
1083 | ||
1084 | /* | |
1085 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1086 | * structure. It *MUST* be at the end of 'task_struct'. | |
1087 | * | |
1088 | * Do not put anything below here! | |
1089 | */ | |
1da177e4 LT |
1090 | }; |
1091 | ||
e868171a | 1092 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 EB |
1093 | { |
1094 | return task->pids[PIDTYPE_PID].pid; | |
1095 | } | |
1096 | ||
e868171a | 1097 | static inline struct pid *task_tgid(struct task_struct *task) |
22c935f4 EB |
1098 | { |
1099 | return task->group_leader->pids[PIDTYPE_PID].pid; | |
1100 | } | |
1101 | ||
6dda81f4 | 1102 | /* |
5eca1c10 | 1103 | * Without tasklist or RCU lock it is not safe to dereference |
6dda81f4 ON |
1104 | * the result of task_pgrp/task_session even if task == current, |
1105 | * we can race with another thread doing sys_setsid/sys_setpgid. | |
1106 | */ | |
e868171a | 1107 | static inline struct pid *task_pgrp(struct task_struct *task) |
22c935f4 EB |
1108 | { |
1109 | return task->group_leader->pids[PIDTYPE_PGID].pid; | |
1110 | } | |
1111 | ||
e868171a | 1112 | static inline struct pid *task_session(struct task_struct *task) |
22c935f4 EB |
1113 | { |
1114 | return task->group_leader->pids[PIDTYPE_SID].pid; | |
1115 | } | |
1116 | ||
7af57294 PE |
1117 | /* |
1118 | * the helpers to get the task's different pids as they are seen | |
1119 | * from various namespaces | |
1120 | * | |
1121 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
1122 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
1123 | * current. | |
7af57294 PE |
1124 | * task_xid_nr_ns() : id seen from the ns specified; |
1125 | * | |
7af57294 PE |
1126 | * see also pid_nr() etc in include/linux/pid.h |
1127 | */ | |
5eca1c10 | 1128 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns); |
7af57294 | 1129 | |
e868171a | 1130 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
1131 | { |
1132 | return tsk->pid; | |
1133 | } | |
1134 | ||
5eca1c10 | 1135 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
52ee2dfd ON |
1136 | { |
1137 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
1138 | } | |
7af57294 PE |
1139 | |
1140 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
1141 | { | |
52ee2dfd | 1142 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
1143 | } |
1144 | ||
1145 | ||
e868171a | 1146 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
1147 | { |
1148 | return tsk->tgid; | |
1149 | } | |
1150 | ||
5eca1c10 | 1151 | extern pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); |
7af57294 PE |
1152 | |
1153 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
1154 | { | |
1155 | return pid_vnr(task_tgid(tsk)); | |
1156 | } | |
1157 | ||
5eca1c10 IM |
1158 | /** |
1159 | * pid_alive - check that a task structure is not stale | |
1160 | * @p: Task structure to be checked. | |
1161 | * | |
1162 | * Test if a process is not yet dead (at most zombie state) | |
1163 | * If pid_alive fails, then pointers within the task structure | |
1164 | * can be stale and must not be dereferenced. | |
1165 | * | |
1166 | * Return: 1 if the process is alive. 0 otherwise. | |
1167 | */ | |
1168 | static inline int pid_alive(const struct task_struct *p) | |
1169 | { | |
1170 | return p->pids[PIDTYPE_PID].pid != NULL; | |
1171 | } | |
7af57294 | 1172 | |
ad36d282 RGB |
1173 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) |
1174 | { | |
1175 | pid_t pid = 0; | |
1176 | ||
1177 | rcu_read_lock(); | |
1178 | if (pid_alive(tsk)) | |
1179 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
1180 | rcu_read_unlock(); | |
1181 | ||
1182 | return pid; | |
1183 | } | |
1184 | ||
1185 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
1186 | { | |
1187 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
1188 | } | |
1189 | ||
5eca1c10 | 1190 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1191 | { |
52ee2dfd | 1192 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
1193 | } |
1194 | ||
7af57294 PE |
1195 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
1196 | { | |
52ee2dfd | 1197 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
1198 | } |
1199 | ||
1200 | ||
5eca1c10 | 1201 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1202 | { |
52ee2dfd | 1203 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
1204 | } |
1205 | ||
7af57294 PE |
1206 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
1207 | { | |
52ee2dfd | 1208 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
1209 | } |
1210 | ||
5eca1c10 | 1211 | /* Obsolete, do not use: */ |
1b0f7ffd ON |
1212 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) |
1213 | { | |
1214 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
1215 | } | |
7af57294 | 1216 | |
f400e198 | 1217 | /** |
570f5241 SS |
1218 | * is_global_init - check if a task structure is init. Since init |
1219 | * is free to have sub-threads we need to check tgid. | |
3260259f HK |
1220 | * @tsk: Task structure to be checked. |
1221 | * | |
1222 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
1223 | * |
1224 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 1225 | */ |
e868171a | 1226 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 1227 | { |
570f5241 | 1228 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 1229 | } |
b460cbc5 | 1230 | |
9ec52099 CLG |
1231 | extern struct pid *cad_pid; |
1232 | ||
1da177e4 LT |
1233 | /* |
1234 | * Per process flags | |
1235 | */ | |
5eca1c10 IM |
1236 | #define PF_IDLE 0x00000002 /* I am an IDLE thread */ |
1237 | #define PF_EXITING 0x00000004 /* Getting shut down */ | |
1238 | #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */ | |
1239 | #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ | |
1240 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ | |
1241 | #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */ | |
1242 | #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */ | |
1243 | #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */ | |
1244 | #define PF_DUMPCORE 0x00000200 /* Dumped core */ | |
1245 | #define PF_SIGNALED 0x00000400 /* Killed by a signal */ | |
1246 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
1247 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */ | |
1248 | #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */ | |
1249 | #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */ | |
1250 | #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */ | |
1251 | #define PF_FROZEN 0x00010000 /* Frozen for system suspend */ | |
7dea19f9 MH |
1252 | #define PF_KSWAPD 0x00020000 /* I am kswapd */ |
1253 | #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */ | |
1254 | #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */ | |
5eca1c10 IM |
1255 | #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ |
1256 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ | |
1257 | #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */ | |
1258 | #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ | |
1259 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */ | |
1260 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ | |
1261 | #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ | |
1262 | #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ | |
1263 | #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */ | |
1da177e4 LT |
1264 | |
1265 | /* | |
1266 | * Only the _current_ task can read/write to tsk->flags, but other | |
1267 | * tasks can access tsk->flags in readonly mode for example | |
1268 | * with tsk_used_math (like during threaded core dumping). | |
1269 | * There is however an exception to this rule during ptrace | |
1270 | * or during fork: the ptracer task is allowed to write to the | |
1271 | * child->flags of its traced child (same goes for fork, the parent | |
1272 | * can write to the child->flags), because we're guaranteed the | |
1273 | * child is not running and in turn not changing child->flags | |
1274 | * at the same time the parent does it. | |
1275 | */ | |
5eca1c10 IM |
1276 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) |
1277 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
1278 | #define clear_used_math() clear_stopped_child_used_math(current) | |
1279 | #define set_used_math() set_stopped_child_used_math(current) | |
1280 | ||
1da177e4 LT |
1281 | #define conditional_stopped_child_used_math(condition, child) \ |
1282 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
5eca1c10 IM |
1283 | |
1284 | #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current) | |
1285 | ||
1da177e4 LT |
1286 | #define copy_to_stopped_child_used_math(child) \ |
1287 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
5eca1c10 | 1288 | |
1da177e4 | 1289 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ |
5eca1c10 IM |
1290 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) |
1291 | #define used_math() tsk_used_math(current) | |
1da177e4 | 1292 | |
62ec05dd TG |
1293 | static inline bool is_percpu_thread(void) |
1294 | { | |
1295 | #ifdef CONFIG_SMP | |
1296 | return (current->flags & PF_NO_SETAFFINITY) && | |
1297 | (current->nr_cpus_allowed == 1); | |
1298 | #else | |
1299 | return true; | |
1300 | #endif | |
1301 | } | |
1302 | ||
1d4457f9 | 1303 | /* Per-process atomic flags. */ |
5eca1c10 IM |
1304 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
1305 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ | |
1306 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
2ad654bc | 1307 | |
1d4457f9 | 1308 | |
e0e5070b ZL |
1309 | #define TASK_PFA_TEST(name, func) \ |
1310 | static inline bool task_##func(struct task_struct *p) \ | |
1311 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1312 | |
e0e5070b ZL |
1313 | #define TASK_PFA_SET(name, func) \ |
1314 | static inline void task_set_##func(struct task_struct *p) \ | |
1315 | { set_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1316 | |
e0e5070b ZL |
1317 | #define TASK_PFA_CLEAR(name, func) \ |
1318 | static inline void task_clear_##func(struct task_struct *p) \ | |
1319 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
1320 | ||
1321 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
1322 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 1323 | |
2ad654bc ZL |
1324 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
1325 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
1326 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
1327 | ||
1328 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
1329 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
1330 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 1331 | |
5eca1c10 | 1332 | static inline void |
717a94b5 | 1333 | current_restore_flags(unsigned long orig_flags, unsigned long flags) |
907aed48 | 1334 | { |
717a94b5 N |
1335 | current->flags &= ~flags; |
1336 | current->flags |= orig_flags & flags; | |
907aed48 MG |
1337 | } |
1338 | ||
5eca1c10 IM |
1339 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial); |
1340 | extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed); | |
1da177e4 | 1341 | #ifdef CONFIG_SMP |
5eca1c10 IM |
1342 | extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); |
1343 | extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask); | |
1da177e4 | 1344 | #else |
5eca1c10 | 1345 | static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) |
1e1b6c51 KM |
1346 | { |
1347 | } | |
5eca1c10 | 1348 | static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) |
1da177e4 | 1349 | { |
96f874e2 | 1350 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
1351 | return -EINVAL; |
1352 | return 0; | |
1353 | } | |
1354 | #endif | |
e0ad9556 | 1355 | |
6d0d2878 CB |
1356 | #ifndef cpu_relax_yield |
1357 | #define cpu_relax_yield() cpu_relax() | |
1358 | #endif | |
1359 | ||
fa93384f | 1360 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
1361 | extern void set_user_nice(struct task_struct *p, long nice); |
1362 | extern int task_prio(const struct task_struct *p); | |
5eca1c10 | 1363 | |
d0ea0268 DY |
1364 | /** |
1365 | * task_nice - return the nice value of a given task. | |
1366 | * @p: the task in question. | |
1367 | * | |
1368 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
1369 | */ | |
1370 | static inline int task_nice(const struct task_struct *p) | |
1371 | { | |
1372 | return PRIO_TO_NICE((p)->static_prio); | |
1373 | } | |
5eca1c10 | 1374 | |
36c8b586 IM |
1375 | extern int can_nice(const struct task_struct *p, const int nice); |
1376 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 1377 | extern int idle_cpu(int cpu); |
5eca1c10 IM |
1378 | extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *); |
1379 | extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *); | |
1380 | extern int sched_setattr(struct task_struct *, const struct sched_attr *); | |
36c8b586 | 1381 | extern struct task_struct *idle_task(int cpu); |
5eca1c10 | 1382 | |
c4f30608 PM |
1383 | /** |
1384 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 1385 | * @p: the task in question. |
e69f6186 YB |
1386 | * |
1387 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 1388 | */ |
7061ca3b | 1389 | static inline bool is_idle_task(const struct task_struct *p) |
c4f30608 | 1390 | { |
c1de45ca | 1391 | return !!(p->flags & PF_IDLE); |
c4f30608 | 1392 | } |
5eca1c10 | 1393 | |
36c8b586 | 1394 | extern struct task_struct *curr_task(int cpu); |
a458ae2e | 1395 | extern void ia64_set_curr_task(int cpu, struct task_struct *p); |
1da177e4 LT |
1396 | |
1397 | void yield(void); | |
1398 | ||
1da177e4 | 1399 | union thread_union { |
c65eacbe | 1400 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1da177e4 | 1401 | struct thread_info thread_info; |
c65eacbe | 1402 | #endif |
1da177e4 LT |
1403 | unsigned long stack[THREAD_SIZE/sizeof(long)]; |
1404 | }; | |
1405 | ||
f3ac6067 IM |
1406 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1407 | static inline struct thread_info *task_thread_info(struct task_struct *task) | |
1408 | { | |
1409 | return &task->thread_info; | |
1410 | } | |
1411 | #elif !defined(__HAVE_THREAD_FUNCTIONS) | |
1412 | # define task_thread_info(task) ((struct thread_info *)(task)->stack) | |
1413 | #endif | |
1414 | ||
198fe21b PE |
1415 | /* |
1416 | * find a task by one of its numerical ids | |
1417 | * | |
198fe21b PE |
1418 | * find_task_by_pid_ns(): |
1419 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
1420 | * find_task_by_vpid(): |
1421 | * finds a task by its virtual pid | |
198fe21b | 1422 | * |
e49859e7 | 1423 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
1424 | */ |
1425 | ||
228ebcbe | 1426 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
5eca1c10 | 1427 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns); |
198fe21b | 1428 | |
b3c97528 HH |
1429 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
1430 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 1431 | extern void wake_up_new_task(struct task_struct *tsk); |
5eca1c10 | 1432 | |
1da177e4 | 1433 | #ifdef CONFIG_SMP |
5eca1c10 | 1434 | extern void kick_process(struct task_struct *tsk); |
1da177e4 | 1435 | #else |
5eca1c10 | 1436 | static inline void kick_process(struct task_struct *tsk) { } |
1da177e4 | 1437 | #endif |
1da177e4 | 1438 | |
82b89778 | 1439 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
5eca1c10 | 1440 | |
82b89778 AH |
1441 | static inline void set_task_comm(struct task_struct *tsk, const char *from) |
1442 | { | |
1443 | __set_task_comm(tsk, from, false); | |
1444 | } | |
5eca1c10 | 1445 | |
59714d65 | 1446 | extern char *get_task_comm(char *to, struct task_struct *tsk); |
1da177e4 LT |
1447 | |
1448 | #ifdef CONFIG_SMP | |
317f3941 | 1449 | void scheduler_ipi(void); |
85ba2d86 | 1450 | extern unsigned long wait_task_inactive(struct task_struct *, long match_state); |
1da177e4 | 1451 | #else |
184748cc | 1452 | static inline void scheduler_ipi(void) { } |
5eca1c10 | 1453 | static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state) |
85ba2d86 RM |
1454 | { |
1455 | return 1; | |
1456 | } | |
1da177e4 LT |
1457 | #endif |
1458 | ||
5eca1c10 IM |
1459 | /* |
1460 | * Set thread flags in other task's structures. | |
1461 | * See asm/thread_info.h for TIF_xxxx flags available: | |
1da177e4 LT |
1462 | */ |
1463 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1464 | { | |
a1261f54 | 1465 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1466 | } |
1467 | ||
1468 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1469 | { | |
a1261f54 | 1470 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1471 | } |
1472 | ||
1473 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1474 | { | |
a1261f54 | 1475 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1476 | } |
1477 | ||
1478 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1479 | { | |
a1261f54 | 1480 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1481 | } |
1482 | ||
1483 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1484 | { | |
a1261f54 | 1485 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1486 | } |
1487 | ||
1488 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
1489 | { | |
1490 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1491 | } | |
1492 | ||
1493 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
1494 | { | |
1495 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1496 | } | |
1497 | ||
8ae121ac GH |
1498 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
1499 | { | |
1500 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
1501 | } | |
1502 | ||
1da177e4 LT |
1503 | /* |
1504 | * cond_resched() and cond_resched_lock(): latency reduction via | |
1505 | * explicit rescheduling in places that are safe. The return | |
1506 | * value indicates whether a reschedule was done in fact. | |
1507 | * cond_resched_lock() will drop the spinlock before scheduling, | |
1508 | * cond_resched_softirq() will enable bhs before scheduling. | |
1509 | */ | |
35a773a0 | 1510 | #ifndef CONFIG_PREEMPT |
c3921ab7 | 1511 | extern int _cond_resched(void); |
35a773a0 PZ |
1512 | #else |
1513 | static inline int _cond_resched(void) { return 0; } | |
1514 | #endif | |
6f80bd98 | 1515 | |
613afbf8 | 1516 | #define cond_resched() ({ \ |
3427445a | 1517 | ___might_sleep(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
1518 | _cond_resched(); \ |
1519 | }) | |
6f80bd98 | 1520 | |
613afbf8 FW |
1521 | extern int __cond_resched_lock(spinlock_t *lock); |
1522 | ||
1523 | #define cond_resched_lock(lock) ({ \ | |
3427445a | 1524 | ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ |
613afbf8 FW |
1525 | __cond_resched_lock(lock); \ |
1526 | }) | |
1527 | ||
1528 | extern int __cond_resched_softirq(void); | |
1529 | ||
75e1056f | 1530 | #define cond_resched_softirq() ({ \ |
3427445a | 1531 | ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \ |
75e1056f | 1532 | __cond_resched_softirq(); \ |
613afbf8 | 1533 | }) |
1da177e4 | 1534 | |
f6f3c437 SH |
1535 | static inline void cond_resched_rcu(void) |
1536 | { | |
1537 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
1538 | rcu_read_unlock(); | |
1539 | cond_resched(); | |
1540 | rcu_read_lock(); | |
1541 | #endif | |
1542 | } | |
1543 | ||
1da177e4 LT |
1544 | /* |
1545 | * Does a critical section need to be broken due to another | |
95c354fe NP |
1546 | * task waiting?: (technically does not depend on CONFIG_PREEMPT, |
1547 | * but a general need for low latency) | |
1da177e4 | 1548 | */ |
95c354fe | 1549 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 1550 | { |
95c354fe NP |
1551 | #ifdef CONFIG_PREEMPT |
1552 | return spin_is_contended(lock); | |
1553 | #else | |
1da177e4 | 1554 | return 0; |
95c354fe | 1555 | #endif |
1da177e4 LT |
1556 | } |
1557 | ||
75f93fed PZ |
1558 | static __always_inline bool need_resched(void) |
1559 | { | |
1560 | return unlikely(tif_need_resched()); | |
1561 | } | |
1562 | ||
1da177e4 LT |
1563 | /* |
1564 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
1565 | */ | |
1566 | #ifdef CONFIG_SMP | |
1567 | ||
1568 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1569 | { | |
c65eacbe AL |
1570 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1571 | return p->cpu; | |
1572 | #else | |
a1261f54 | 1573 | return task_thread_info(p)->cpu; |
c65eacbe | 1574 | #endif |
1da177e4 LT |
1575 | } |
1576 | ||
c65cc870 | 1577 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
1578 | |
1579 | #else | |
1580 | ||
1581 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1582 | { | |
1583 | return 0; | |
1584 | } | |
1585 | ||
1586 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
1587 | { | |
1588 | } | |
1589 | ||
1590 | #endif /* CONFIG_SMP */ | |
1591 | ||
d9345c65 PX |
1592 | /* |
1593 | * In order to reduce various lock holder preemption latencies provide an | |
1594 | * interface to see if a vCPU is currently running or not. | |
1595 | * | |
1596 | * This allows us to terminate optimistic spin loops and block, analogous to | |
1597 | * the native optimistic spin heuristic of testing if the lock owner task is | |
1598 | * running or not. | |
1599 | */ | |
1600 | #ifndef vcpu_is_preempted | |
1601 | # define vcpu_is_preempted(cpu) false | |
1602 | #endif | |
1603 | ||
96f874e2 RR |
1604 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
1605 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 1606 | |
82455257 DH |
1607 | #ifndef TASK_SIZE_OF |
1608 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
1609 | #endif | |
1610 | ||
1da177e4 | 1611 | #endif |