]>
Commit | Line | Data |
---|---|---|
b2441318 | 1 | /* SPDX-License-Identifier: GPL-2.0 */ |
1da177e4 LT |
2 | #ifndef _LINUX_SCHED_H |
3 | #define _LINUX_SCHED_H | |
4 | ||
5eca1c10 IM |
5 | /* |
6 | * Define 'struct task_struct' and provide the main scheduler | |
7 | * APIs (schedule(), wakeup variants, etc.) | |
8 | */ | |
b7b3c76a | 9 | |
5eca1c10 | 10 | #include <uapi/linux/sched.h> |
5c228079 | 11 | |
5eca1c10 | 12 | #include <asm/current.h> |
1da177e4 | 13 | |
5eca1c10 | 14 | #include <linux/pid.h> |
1da177e4 | 15 | #include <linux/sem.h> |
ab602f79 | 16 | #include <linux/shm.h> |
5eca1c10 IM |
17 | #include <linux/kcov.h> |
18 | #include <linux/mutex.h> | |
19 | #include <linux/plist.h> | |
20 | #include <linux/hrtimer.h> | |
1da177e4 | 21 | #include <linux/seccomp.h> |
5eca1c10 | 22 | #include <linux/nodemask.h> |
b68070e1 | 23 | #include <linux/rcupdate.h> |
a3b6714e | 24 | #include <linux/resource.h> |
9745512c | 25 | #include <linux/latencytop.h> |
5eca1c10 IM |
26 | #include <linux/sched/prio.h> |
27 | #include <linux/signal_types.h> | |
28 | #include <linux/mm_types_task.h> | |
29 | #include <linux/task_io_accounting.h> | |
a3b6714e | 30 | |
5eca1c10 | 31 | /* task_struct member predeclarations (sorted alphabetically): */ |
c7af7877 | 32 | struct audit_context; |
c7af7877 | 33 | struct backing_dev_info; |
bddd87c7 | 34 | struct bio_list; |
73c10101 | 35 | struct blk_plug; |
c7af7877 | 36 | struct cfs_rq; |
c7af7877 IM |
37 | struct fs_struct; |
38 | struct futex_pi_state; | |
39 | struct io_context; | |
40 | struct mempolicy; | |
89076bc3 | 41 | struct nameidata; |
c7af7877 IM |
42 | struct nsproxy; |
43 | struct perf_event_context; | |
44 | struct pid_namespace; | |
45 | struct pipe_inode_info; | |
46 | struct rcu_node; | |
47 | struct reclaim_state; | |
48 | struct robust_list_head; | |
49 | struct sched_attr; | |
50 | struct sched_param; | |
43ae34cb | 51 | struct seq_file; |
c7af7877 IM |
52 | struct sighand_struct; |
53 | struct signal_struct; | |
54 | struct task_delay_info; | |
4cf86d77 | 55 | struct task_group; |
1da177e4 | 56 | |
4a8342d2 LT |
57 | /* |
58 | * Task state bitmask. NOTE! These bits are also | |
59 | * encoded in fs/proc/array.c: get_task_state(). | |
60 | * | |
61 | * We have two separate sets of flags: task->state | |
62 | * is about runnability, while task->exit_state are | |
63 | * about the task exiting. Confusing, but this way | |
64 | * modifying one set can't modify the other one by | |
65 | * mistake. | |
66 | */ | |
5eca1c10 IM |
67 | |
68 | /* Used in tsk->state: */ | |
92c4bc9f PZ |
69 | #define TASK_RUNNING 0x0000 |
70 | #define TASK_INTERRUPTIBLE 0x0001 | |
71 | #define TASK_UNINTERRUPTIBLE 0x0002 | |
72 | #define __TASK_STOPPED 0x0004 | |
73 | #define __TASK_TRACED 0x0008 | |
5eca1c10 | 74 | /* Used in tsk->exit_state: */ |
92c4bc9f PZ |
75 | #define EXIT_DEAD 0x0010 |
76 | #define EXIT_ZOMBIE 0x0020 | |
5eca1c10 IM |
77 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) |
78 | /* Used in tsk->state again: */ | |
8ef9925b PZ |
79 | #define TASK_PARKED 0x0040 |
80 | #define TASK_DEAD 0x0080 | |
81 | #define TASK_WAKEKILL 0x0100 | |
82 | #define TASK_WAKING 0x0200 | |
92c4bc9f PZ |
83 | #define TASK_NOLOAD 0x0400 |
84 | #define TASK_NEW 0x0800 | |
85 | #define TASK_STATE_MAX 0x1000 | |
5eca1c10 | 86 | |
5eca1c10 IM |
87 | /* Convenience macros for the sake of set_current_state: */ |
88 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) | |
89 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
90 | #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) | |
91 | ||
92 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) | |
93 | ||
94 | /* Convenience macros for the sake of wake_up(): */ | |
95 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
96 | #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) | |
97 | ||
98 | /* get_task_state(): */ | |
99 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
100 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ | |
8ef9925b PZ |
101 | __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \ |
102 | TASK_PARKED) | |
5eca1c10 IM |
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 | ||
ea19cfd1 PZ |
116 | /* |
117 | * Special states are those that do not use the normal wait-loop pattern. See | |
118 | * the comment with set_special_state(). | |
119 | */ | |
120 | #define is_special_task_state(state) \ | |
121 | ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_DEAD)) | |
122 | ||
8eb23b9f PZ |
123 | #define __set_current_state(state_value) \ |
124 | do { \ | |
ea19cfd1 | 125 | WARN_ON_ONCE(is_special_task_state(state_value));\ |
8eb23b9f PZ |
126 | current->task_state_change = _THIS_IP_; \ |
127 | current->state = (state_value); \ | |
128 | } while (0) | |
ea19cfd1 | 129 | |
8eb23b9f PZ |
130 | #define set_current_state(state_value) \ |
131 | do { \ | |
ea19cfd1 | 132 | WARN_ON_ONCE(is_special_task_state(state_value));\ |
8eb23b9f | 133 | current->task_state_change = _THIS_IP_; \ |
a2250238 | 134 | smp_store_mb(current->state, (state_value)); \ |
8eb23b9f PZ |
135 | } while (0) |
136 | ||
ea19cfd1 PZ |
137 | #define set_special_state(state_value) \ |
138 | do { \ | |
139 | unsigned long flags; /* may shadow */ \ | |
140 | WARN_ON_ONCE(!is_special_task_state(state_value)); \ | |
141 | raw_spin_lock_irqsave(¤t->pi_lock, flags); \ | |
142 | current->task_state_change = _THIS_IP_; \ | |
143 | current->state = (state_value); \ | |
144 | raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ | |
145 | } while (0) | |
8eb23b9f | 146 | #else |
498d0c57 AM |
147 | /* |
148 | * set_current_state() includes a barrier so that the write of current->state | |
149 | * is correctly serialised wrt the caller's subsequent test of whether to | |
150 | * actually sleep: | |
151 | * | |
a2250238 | 152 | * for (;;) { |
498d0c57 | 153 | * set_current_state(TASK_UNINTERRUPTIBLE); |
a2250238 PZ |
154 | * if (!need_sleep) |
155 | * break; | |
156 | * | |
157 | * schedule(); | |
158 | * } | |
159 | * __set_current_state(TASK_RUNNING); | |
160 | * | |
161 | * If the caller does not need such serialisation (because, for instance, the | |
162 | * condition test and condition change and wakeup are under the same lock) then | |
163 | * use __set_current_state(). | |
164 | * | |
165 | * The above is typically ordered against the wakeup, which does: | |
166 | * | |
ea19cfd1 PZ |
167 | * need_sleep = false; |
168 | * wake_up_state(p, TASK_UNINTERRUPTIBLE); | |
a2250238 PZ |
169 | * |
170 | * Where wake_up_state() (and all other wakeup primitives) imply enough | |
171 | * barriers to order the store of the variable against wakeup. | |
172 | * | |
173 | * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, | |
174 | * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a | |
175 | * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). | |
498d0c57 | 176 | * |
ea19cfd1 PZ |
177 | * However, with slightly different timing the wakeup TASK_RUNNING store can |
178 | * also collide with the TASK_UNINTERRUPTIBLE store. Loosing that store is not | |
179 | * a problem either because that will result in one extra go around the loop | |
180 | * and our @cond test will save the day. | |
498d0c57 | 181 | * |
a2250238 | 182 | * Also see the comments of try_to_wake_up(). |
498d0c57 | 183 | */ |
ea19cfd1 PZ |
184 | #define __set_current_state(state_value) \ |
185 | current->state = (state_value) | |
186 | ||
187 | #define set_current_state(state_value) \ | |
188 | smp_store_mb(current->state, (state_value)) | |
189 | ||
190 | /* | |
191 | * set_special_state() should be used for those states when the blocking task | |
192 | * can not use the regular condition based wait-loop. In that case we must | |
193 | * serialize against wakeups such that any possible in-flight TASK_RUNNING stores | |
194 | * will not collide with our state change. | |
195 | */ | |
196 | #define set_special_state(state_value) \ | |
197 | do { \ | |
198 | unsigned long flags; /* may shadow */ \ | |
199 | raw_spin_lock_irqsave(¤t->pi_lock, flags); \ | |
200 | current->state = (state_value); \ | |
201 | raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ | |
202 | } while (0) | |
203 | ||
8eb23b9f PZ |
204 | #endif |
205 | ||
5eca1c10 IM |
206 | /* Task command name length: */ |
207 | #define TASK_COMM_LEN 16 | |
1da177e4 | 208 | |
1da177e4 LT |
209 | extern void scheduler_tick(void); |
210 | ||
5eca1c10 IM |
211 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX |
212 | ||
213 | extern long schedule_timeout(long timeout); | |
214 | extern long schedule_timeout_interruptible(long timeout); | |
215 | extern long schedule_timeout_killable(long timeout); | |
216 | extern long schedule_timeout_uninterruptible(long timeout); | |
217 | extern long schedule_timeout_idle(long timeout); | |
1da177e4 | 218 | asmlinkage void schedule(void); |
c5491ea7 | 219 | extern void schedule_preempt_disabled(void); |
1da177e4 | 220 | |
10ab5643 TH |
221 | extern int __must_check io_schedule_prepare(void); |
222 | extern void io_schedule_finish(int token); | |
9cff8ade | 223 | extern long io_schedule_timeout(long timeout); |
10ab5643 | 224 | extern void io_schedule(void); |
9cff8ade | 225 | |
d37f761d | 226 | /** |
0ba42a59 | 227 | * struct prev_cputime - snapshot of system and user cputime |
d37f761d FW |
228 | * @utime: time spent in user mode |
229 | * @stime: time spent in system mode | |
9d7fb042 | 230 | * @lock: protects the above two fields |
d37f761d | 231 | * |
9d7fb042 PZ |
232 | * Stores previous user/system time values such that we can guarantee |
233 | * monotonicity. | |
d37f761d | 234 | */ |
9d7fb042 PZ |
235 | struct prev_cputime { |
236 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
5eca1c10 IM |
237 | u64 utime; |
238 | u64 stime; | |
239 | raw_spinlock_t lock; | |
9d7fb042 | 240 | #endif |
d37f761d FW |
241 | }; |
242 | ||
f06febc9 FM |
243 | /** |
244 | * struct task_cputime - collected CPU time counts | |
5613fda9 FW |
245 | * @utime: time spent in user mode, in nanoseconds |
246 | * @stime: time spent in kernel mode, in nanoseconds | |
f06febc9 | 247 | * @sum_exec_runtime: total time spent on the CPU, in nanoseconds |
5ce73a4a | 248 | * |
9d7fb042 PZ |
249 | * This structure groups together three kinds of CPU time that are tracked for |
250 | * threads and thread groups. Most things considering CPU time want to group | |
251 | * these counts together and treat all three of them in parallel. | |
f06febc9 FM |
252 | */ |
253 | struct task_cputime { | |
5eca1c10 IM |
254 | u64 utime; |
255 | u64 stime; | |
256 | unsigned long long sum_exec_runtime; | |
f06febc9 | 257 | }; |
9d7fb042 | 258 | |
5eca1c10 IM |
259 | /* Alternate field names when used on cache expirations: */ |
260 | #define virt_exp utime | |
261 | #define prof_exp stime | |
262 | #define sched_exp sum_exec_runtime | |
f06febc9 | 263 | |
bac5b6b6 FW |
264 | enum vtime_state { |
265 | /* Task is sleeping or running in a CPU with VTIME inactive: */ | |
266 | VTIME_INACTIVE = 0, | |
267 | /* Task runs in userspace in a CPU with VTIME active: */ | |
268 | VTIME_USER, | |
269 | /* Task runs in kernelspace in a CPU with VTIME active: */ | |
270 | VTIME_SYS, | |
271 | }; | |
272 | ||
273 | struct vtime { | |
274 | seqcount_t seqcount; | |
275 | unsigned long long starttime; | |
276 | enum vtime_state state; | |
2a42eb95 WL |
277 | u64 utime; |
278 | u64 stime; | |
279 | u64 gtime; | |
bac5b6b6 FW |
280 | }; |
281 | ||
1da177e4 | 282 | struct sched_info { |
7f5f8e8d | 283 | #ifdef CONFIG_SCHED_INFO |
5eca1c10 IM |
284 | /* Cumulative counters: */ |
285 | ||
286 | /* # of times we have run on this CPU: */ | |
287 | unsigned long pcount; | |
288 | ||
289 | /* Time spent waiting on a runqueue: */ | |
290 | unsigned long long run_delay; | |
291 | ||
292 | /* Timestamps: */ | |
293 | ||
294 | /* When did we last run on a CPU? */ | |
295 | unsigned long long last_arrival; | |
296 | ||
297 | /* When were we last queued to run? */ | |
298 | unsigned long long last_queued; | |
1da177e4 | 299 | |
f6db8347 | 300 | #endif /* CONFIG_SCHED_INFO */ |
7f5f8e8d | 301 | }; |
1da177e4 | 302 | |
6ecdd749 YD |
303 | /* |
304 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
305 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
306 | * | |
307 | * We define a basic fixed point arithmetic range, and then formalize | |
308 | * all these metrics based on that basic range. | |
309 | */ | |
5eca1c10 IM |
310 | # define SCHED_FIXEDPOINT_SHIFT 10 |
311 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
6ecdd749 | 312 | |
20b8a59f | 313 | struct load_weight { |
5eca1c10 IM |
314 | unsigned long weight; |
315 | u32 inv_weight; | |
20b8a59f IM |
316 | }; |
317 | ||
9d89c257 | 318 | /* |
7b595334 YD |
319 | * The load_avg/util_avg accumulates an infinite geometric series |
320 | * (see __update_load_avg() in kernel/sched/fair.c). | |
321 | * | |
322 | * [load_avg definition] | |
323 | * | |
324 | * load_avg = runnable% * scale_load_down(load) | |
325 | * | |
326 | * where runnable% is the time ratio that a sched_entity is runnable. | |
327 | * For cfs_rq, it is the aggregated load_avg of all runnable and | |
9d89c257 | 328 | * blocked sched_entities. |
7b595334 YD |
329 | * |
330 | * load_avg may also take frequency scaling into account: | |
331 | * | |
332 | * load_avg = runnable% * scale_load_down(load) * freq% | |
333 | * | |
334 | * where freq% is the CPU frequency normalized to the highest frequency. | |
335 | * | |
336 | * [util_avg definition] | |
337 | * | |
338 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
339 | * | |
340 | * where running% is the time ratio that a sched_entity is running on | |
341 | * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable | |
342 | * and blocked sched_entities. | |
343 | * | |
344 | * util_avg may also factor frequency scaling and CPU capacity scaling: | |
345 | * | |
346 | * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity% | |
347 | * | |
348 | * where freq% is the same as above, and capacity% is the CPU capacity | |
349 | * normalized to the greatest capacity (due to uarch differences, etc). | |
350 | * | |
351 | * N.B., the above ratios (runnable%, running%, freq%, and capacity%) | |
352 | * themselves are in the range of [0, 1]. To do fixed point arithmetics, | |
353 | * we therefore scale them to as large a range as necessary. This is for | |
354 | * example reflected by util_avg's SCHED_CAPACITY_SCALE. | |
355 | * | |
356 | * [Overflow issue] | |
357 | * | |
358 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
359 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
360 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
361 | * | |
362 | * For all other cases (including 32-bit kernels), struct load_weight's | |
363 | * weight will overflow first before we do, because: | |
364 | * | |
365 | * Max(load_avg) <= Max(load.weight) | |
366 | * | |
367 | * Then it is the load_weight's responsibility to consider overflow | |
368 | * issues. | |
9d89c257 | 369 | */ |
9d85f21c | 370 | struct sched_avg { |
5eca1c10 IM |
371 | u64 last_update_time; |
372 | u64 load_sum; | |
1ea6c46a | 373 | u64 runnable_load_sum; |
5eca1c10 IM |
374 | u32 util_sum; |
375 | u32 period_contrib; | |
376 | unsigned long load_avg; | |
1ea6c46a | 377 | unsigned long runnable_load_avg; |
5eca1c10 | 378 | unsigned long util_avg; |
9d85f21c PT |
379 | }; |
380 | ||
41acab88 | 381 | struct sched_statistics { |
7f5f8e8d | 382 | #ifdef CONFIG_SCHEDSTATS |
5eca1c10 IM |
383 | u64 wait_start; |
384 | u64 wait_max; | |
385 | u64 wait_count; | |
386 | u64 wait_sum; | |
387 | u64 iowait_count; | |
388 | u64 iowait_sum; | |
389 | ||
390 | u64 sleep_start; | |
391 | u64 sleep_max; | |
392 | s64 sum_sleep_runtime; | |
393 | ||
394 | u64 block_start; | |
395 | u64 block_max; | |
396 | u64 exec_max; | |
397 | u64 slice_max; | |
398 | ||
399 | u64 nr_migrations_cold; | |
400 | u64 nr_failed_migrations_affine; | |
401 | u64 nr_failed_migrations_running; | |
402 | u64 nr_failed_migrations_hot; | |
403 | u64 nr_forced_migrations; | |
404 | ||
405 | u64 nr_wakeups; | |
406 | u64 nr_wakeups_sync; | |
407 | u64 nr_wakeups_migrate; | |
408 | u64 nr_wakeups_local; | |
409 | u64 nr_wakeups_remote; | |
410 | u64 nr_wakeups_affine; | |
411 | u64 nr_wakeups_affine_attempts; | |
412 | u64 nr_wakeups_passive; | |
413 | u64 nr_wakeups_idle; | |
41acab88 | 414 | #endif |
7f5f8e8d | 415 | }; |
41acab88 LDM |
416 | |
417 | struct sched_entity { | |
5eca1c10 IM |
418 | /* For load-balancing: */ |
419 | struct load_weight load; | |
1ea6c46a | 420 | unsigned long runnable_weight; |
5eca1c10 IM |
421 | struct rb_node run_node; |
422 | struct list_head group_node; | |
423 | unsigned int on_rq; | |
41acab88 | 424 | |
5eca1c10 IM |
425 | u64 exec_start; |
426 | u64 sum_exec_runtime; | |
427 | u64 vruntime; | |
428 | u64 prev_sum_exec_runtime; | |
41acab88 | 429 | |
5eca1c10 | 430 | u64 nr_migrations; |
41acab88 | 431 | |
5eca1c10 | 432 | struct sched_statistics statistics; |
94c18227 | 433 | |
20b8a59f | 434 | #ifdef CONFIG_FAIR_GROUP_SCHED |
5eca1c10 IM |
435 | int depth; |
436 | struct sched_entity *parent; | |
20b8a59f | 437 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 438 | struct cfs_rq *cfs_rq; |
20b8a59f | 439 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 440 | struct cfs_rq *my_q; |
20b8a59f | 441 | #endif |
8bd75c77 | 442 | |
141965c7 | 443 | #ifdef CONFIG_SMP |
5a107804 JO |
444 | /* |
445 | * Per entity load average tracking. | |
446 | * | |
447 | * Put into separate cache line so it does not | |
448 | * collide with read-mostly values above. | |
449 | */ | |
5eca1c10 | 450 | struct sched_avg avg ____cacheline_aligned_in_smp; |
9d85f21c | 451 | #endif |
20b8a59f | 452 | }; |
70b97a7f | 453 | |
fa717060 | 454 | struct sched_rt_entity { |
5eca1c10 IM |
455 | struct list_head run_list; |
456 | unsigned long timeout; | |
457 | unsigned long watchdog_stamp; | |
458 | unsigned int time_slice; | |
459 | unsigned short on_rq; | |
460 | unsigned short on_list; | |
461 | ||
462 | struct sched_rt_entity *back; | |
052f1dc7 | 463 | #ifdef CONFIG_RT_GROUP_SCHED |
5eca1c10 | 464 | struct sched_rt_entity *parent; |
6f505b16 | 465 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 466 | struct rt_rq *rt_rq; |
6f505b16 | 467 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 468 | struct rt_rq *my_q; |
6f505b16 | 469 | #endif |
3859a271 | 470 | } __randomize_layout; |
fa717060 | 471 | |
aab03e05 | 472 | struct sched_dl_entity { |
5eca1c10 | 473 | struct rb_node rb_node; |
aab03e05 DF |
474 | |
475 | /* | |
476 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 477 | * during sched_setattr(), they will remain the same until |
478 | * the next sched_setattr(). | |
aab03e05 | 479 | */ |
5eca1c10 IM |
480 | u64 dl_runtime; /* Maximum runtime for each instance */ |
481 | u64 dl_deadline; /* Relative deadline of each instance */ | |
482 | u64 dl_period; /* Separation of two instances (period) */ | |
54d6d303 | 483 | u64 dl_bw; /* dl_runtime / dl_period */ |
3effcb42 | 484 | u64 dl_density; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
485 | |
486 | /* | |
487 | * Actual scheduling parameters. Initialized with the values above, | |
488 | * they are continously updated during task execution. Note that | |
489 | * the remaining runtime could be < 0 in case we are in overrun. | |
490 | */ | |
5eca1c10 IM |
491 | s64 runtime; /* Remaining runtime for this instance */ |
492 | u64 deadline; /* Absolute deadline for this instance */ | |
493 | unsigned int flags; /* Specifying the scheduler behaviour */ | |
aab03e05 DF |
494 | |
495 | /* | |
496 | * Some bool flags: | |
497 | * | |
498 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
499 | * task has to wait for a replenishment to be performed at the | |
500 | * next firing of dl_timer. | |
501 | * | |
2d3d891d DF |
502 | * @dl_boosted tells if we are boosted due to DI. If so we are |
503 | * outside bandwidth enforcement mechanism (but only until we | |
5bfd126e JL |
504 | * exit the critical section); |
505 | * | |
5eca1c10 | 506 | * @dl_yielded tells if task gave up the CPU before consuming |
5bfd126e | 507 | * all its available runtime during the last job. |
209a0cbd LA |
508 | * |
509 | * @dl_non_contending tells if the task is inactive while still | |
510 | * contributing to the active utilization. In other words, it | |
511 | * indicates if the inactive timer has been armed and its handler | |
512 | * has not been executed yet. This flag is useful to avoid race | |
513 | * conditions between the inactive timer handler and the wakeup | |
514 | * code. | |
aab03e05 | 515 | */ |
aa5222e9 DC |
516 | unsigned int dl_throttled : 1; |
517 | unsigned int dl_boosted : 1; | |
518 | unsigned int dl_yielded : 1; | |
519 | unsigned int dl_non_contending : 1; | |
aab03e05 DF |
520 | |
521 | /* | |
522 | * Bandwidth enforcement timer. Each -deadline task has its | |
523 | * own bandwidth to be enforced, thus we need one timer per task. | |
524 | */ | |
5eca1c10 | 525 | struct hrtimer dl_timer; |
209a0cbd LA |
526 | |
527 | /* | |
528 | * Inactive timer, responsible for decreasing the active utilization | |
529 | * at the "0-lag time". When a -deadline task blocks, it contributes | |
530 | * to GRUB's active utilization until the "0-lag time", hence a | |
531 | * timer is needed to decrease the active utilization at the correct | |
532 | * time. | |
533 | */ | |
534 | struct hrtimer inactive_timer; | |
aab03e05 | 535 | }; |
8bd75c77 | 536 | |
1d082fd0 PM |
537 | union rcu_special { |
538 | struct { | |
5eca1c10 IM |
539 | u8 blocked; |
540 | u8 need_qs; | |
541 | u8 exp_need_qs; | |
542 | ||
543 | /* Otherwise the compiler can store garbage here: */ | |
544 | u8 pad; | |
8203d6d0 PM |
545 | } b; /* Bits. */ |
546 | u32 s; /* Set of bits. */ | |
1d082fd0 | 547 | }; |
86848966 | 548 | |
8dc85d54 PZ |
549 | enum perf_event_task_context { |
550 | perf_invalid_context = -1, | |
551 | perf_hw_context = 0, | |
89a1e187 | 552 | perf_sw_context, |
8dc85d54 PZ |
553 | perf_nr_task_contexts, |
554 | }; | |
555 | ||
eb61baf6 IM |
556 | struct wake_q_node { |
557 | struct wake_q_node *next; | |
558 | }; | |
559 | ||
1da177e4 | 560 | struct task_struct { |
c65eacbe AL |
561 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
562 | /* | |
563 | * For reasons of header soup (see current_thread_info()), this | |
564 | * must be the first element of task_struct. | |
565 | */ | |
5eca1c10 | 566 | struct thread_info thread_info; |
c65eacbe | 567 | #endif |
5eca1c10 IM |
568 | /* -1 unrunnable, 0 runnable, >0 stopped: */ |
569 | volatile long state; | |
29e48ce8 KC |
570 | |
571 | /* | |
572 | * This begins the randomizable portion of task_struct. Only | |
573 | * scheduling-critical items should be added above here. | |
574 | */ | |
575 | randomized_struct_fields_start | |
576 | ||
5eca1c10 IM |
577 | void *stack; |
578 | atomic_t usage; | |
579 | /* Per task flags (PF_*), defined further below: */ | |
580 | unsigned int flags; | |
581 | unsigned int ptrace; | |
1da177e4 | 582 | |
2dd73a4f | 583 | #ifdef CONFIG_SMP |
5eca1c10 IM |
584 | struct llist_node wake_entry; |
585 | int on_cpu; | |
c65eacbe | 586 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 IM |
587 | /* Current CPU: */ |
588 | unsigned int cpu; | |
c65eacbe | 589 | #endif |
5eca1c10 IM |
590 | unsigned int wakee_flips; |
591 | unsigned long wakee_flip_decay_ts; | |
592 | struct task_struct *last_wakee; | |
ac66f547 | 593 | |
5eca1c10 | 594 | int wake_cpu; |
2dd73a4f | 595 | #endif |
5eca1c10 IM |
596 | int on_rq; |
597 | ||
598 | int prio; | |
599 | int static_prio; | |
600 | int normal_prio; | |
601 | unsigned int rt_priority; | |
50e645a8 | 602 | |
5eca1c10 IM |
603 | const struct sched_class *sched_class; |
604 | struct sched_entity se; | |
605 | struct sched_rt_entity rt; | |
8323f26c | 606 | #ifdef CONFIG_CGROUP_SCHED |
5eca1c10 | 607 | struct task_group *sched_task_group; |
8323f26c | 608 | #endif |
5eca1c10 | 609 | struct sched_dl_entity dl; |
1da177e4 | 610 | |
e107be36 | 611 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
5eca1c10 IM |
612 | /* List of struct preempt_notifier: */ |
613 | struct hlist_head preempt_notifiers; | |
e107be36 AK |
614 | #endif |
615 | ||
6c5c9341 | 616 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
5eca1c10 | 617 | unsigned int btrace_seq; |
6c5c9341 | 618 | #endif |
1da177e4 | 619 | |
5eca1c10 IM |
620 | unsigned int policy; |
621 | int nr_cpus_allowed; | |
622 | cpumask_t cpus_allowed; | |
1da177e4 | 623 | |
a57eb940 | 624 | #ifdef CONFIG_PREEMPT_RCU |
5eca1c10 IM |
625 | int rcu_read_lock_nesting; |
626 | union rcu_special rcu_read_unlock_special; | |
627 | struct list_head rcu_node_entry; | |
628 | struct rcu_node *rcu_blocked_node; | |
28f6569a | 629 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
5eca1c10 | 630 | |
8315f422 | 631 | #ifdef CONFIG_TASKS_RCU |
5eca1c10 | 632 | unsigned long rcu_tasks_nvcsw; |
ccdd29ff PM |
633 | u8 rcu_tasks_holdout; |
634 | u8 rcu_tasks_idx; | |
5eca1c10 | 635 | int rcu_tasks_idle_cpu; |
ccdd29ff | 636 | struct list_head rcu_tasks_holdout_list; |
8315f422 | 637 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 638 | |
5eca1c10 | 639 | struct sched_info sched_info; |
1da177e4 | 640 | |
5eca1c10 | 641 | struct list_head tasks; |
806c09a7 | 642 | #ifdef CONFIG_SMP |
5eca1c10 IM |
643 | struct plist_node pushable_tasks; |
644 | struct rb_node pushable_dl_tasks; | |
806c09a7 | 645 | #endif |
1da177e4 | 646 | |
5eca1c10 IM |
647 | struct mm_struct *mm; |
648 | struct mm_struct *active_mm; | |
314ff785 IM |
649 | |
650 | /* Per-thread vma caching: */ | |
5eca1c10 | 651 | struct vmacache vmacache; |
314ff785 | 652 | |
5eca1c10 IM |
653 | #ifdef SPLIT_RSS_COUNTING |
654 | struct task_rss_stat rss_stat; | |
34e55232 | 655 | #endif |
5eca1c10 IM |
656 | int exit_state; |
657 | int exit_code; | |
658 | int exit_signal; | |
659 | /* The signal sent when the parent dies: */ | |
660 | int pdeath_signal; | |
661 | /* JOBCTL_*, siglock protected: */ | |
662 | unsigned long jobctl; | |
663 | ||
664 | /* Used for emulating ABI behavior of previous Linux versions: */ | |
665 | unsigned int personality; | |
666 | ||
667 | /* Scheduler bits, serialized by scheduler locks: */ | |
668 | unsigned sched_reset_on_fork:1; | |
669 | unsigned sched_contributes_to_load:1; | |
670 | unsigned sched_migrated:1; | |
671 | unsigned sched_remote_wakeup:1; | |
672 | /* Force alignment to the next boundary: */ | |
673 | unsigned :0; | |
674 | ||
675 | /* Unserialized, strictly 'current' */ | |
676 | ||
677 | /* Bit to tell LSMs we're in execve(): */ | |
678 | unsigned in_execve:1; | |
679 | unsigned in_iowait:1; | |
680 | #ifndef TIF_RESTORE_SIGMASK | |
681 | unsigned restore_sigmask:1; | |
7e781418 | 682 | #endif |
626ebc41 | 683 | #ifdef CONFIG_MEMCG |
5eca1c10 | 684 | unsigned memcg_may_oom:1; |
127424c8 | 685 | #ifndef CONFIG_SLOB |
5eca1c10 | 686 | unsigned memcg_kmem_skip_account:1; |
6f185c29 | 687 | #endif |
127424c8 | 688 | #endif |
ff303e66 | 689 | #ifdef CONFIG_COMPAT_BRK |
5eca1c10 | 690 | unsigned brk_randomized:1; |
ff303e66 | 691 | #endif |
77f88796 TH |
692 | #ifdef CONFIG_CGROUPS |
693 | /* disallow userland-initiated cgroup migration */ | |
694 | unsigned no_cgroup_migration:1; | |
695 | #endif | |
6f185c29 | 696 | |
5eca1c10 | 697 | unsigned long atomic_flags; /* Flags requiring atomic access. */ |
1d4457f9 | 698 | |
5eca1c10 | 699 | struct restart_block restart_block; |
f56141e3 | 700 | |
5eca1c10 IM |
701 | pid_t pid; |
702 | pid_t tgid; | |
0a425405 | 703 | |
1314562a | 704 | #ifdef CONFIG_CC_STACKPROTECTOR |
5eca1c10 IM |
705 | /* Canary value for the -fstack-protector GCC feature: */ |
706 | unsigned long stack_canary; | |
1314562a | 707 | #endif |
4d1d61a6 | 708 | /* |
5eca1c10 | 709 | * Pointers to the (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 710 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 711 | * p->real_parent->pid) |
1da177e4 | 712 | */ |
5eca1c10 IM |
713 | |
714 | /* Real parent process: */ | |
715 | struct task_struct __rcu *real_parent; | |
716 | ||
717 | /* Recipient of SIGCHLD, wait4() reports: */ | |
718 | struct task_struct __rcu *parent; | |
719 | ||
1da177e4 | 720 | /* |
5eca1c10 | 721 | * Children/sibling form the list of natural children: |
1da177e4 | 722 | */ |
5eca1c10 IM |
723 | struct list_head children; |
724 | struct list_head sibling; | |
725 | struct task_struct *group_leader; | |
1da177e4 | 726 | |
f470021a | 727 | /* |
5eca1c10 IM |
728 | * 'ptraced' is the list of tasks this task is using ptrace() on. |
729 | * | |
f470021a | 730 | * This includes both natural children and PTRACE_ATTACH targets. |
5eca1c10 | 731 | * 'ptrace_entry' is this task's link on the p->parent->ptraced list. |
f470021a | 732 | */ |
5eca1c10 IM |
733 | struct list_head ptraced; |
734 | struct list_head ptrace_entry; | |
f470021a | 735 | |
1da177e4 | 736 | /* PID/PID hash table linkage. */ |
5eca1c10 IM |
737 | struct pid_link pids[PIDTYPE_MAX]; |
738 | struct list_head thread_group; | |
739 | struct list_head thread_node; | |
740 | ||
741 | struct completion *vfork_done; | |
1da177e4 | 742 | |
5eca1c10 IM |
743 | /* CLONE_CHILD_SETTID: */ |
744 | int __user *set_child_tid; | |
1da177e4 | 745 | |
5eca1c10 IM |
746 | /* CLONE_CHILD_CLEARTID: */ |
747 | int __user *clear_child_tid; | |
748 | ||
749 | u64 utime; | |
750 | u64 stime; | |
40565b5a | 751 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME |
5eca1c10 IM |
752 | u64 utimescaled; |
753 | u64 stimescaled; | |
40565b5a | 754 | #endif |
5eca1c10 IM |
755 | u64 gtime; |
756 | struct prev_cputime prev_cputime; | |
6a61671b | 757 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
bac5b6b6 | 758 | struct vtime vtime; |
d99ca3b9 | 759 | #endif |
d027d45d FW |
760 | |
761 | #ifdef CONFIG_NO_HZ_FULL | |
5eca1c10 | 762 | atomic_t tick_dep_mask; |
d027d45d | 763 | #endif |
5eca1c10 IM |
764 | /* Context switch counts: */ |
765 | unsigned long nvcsw; | |
766 | unsigned long nivcsw; | |
767 | ||
768 | /* Monotonic time in nsecs: */ | |
769 | u64 start_time; | |
770 | ||
771 | /* Boot based time in nsecs: */ | |
772 | u64 real_start_time; | |
773 | ||
774 | /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */ | |
775 | unsigned long min_flt; | |
776 | unsigned long maj_flt; | |
1da177e4 | 777 | |
b18b6a9c | 778 | #ifdef CONFIG_POSIX_TIMERS |
5eca1c10 IM |
779 | struct task_cputime cputime_expires; |
780 | struct list_head cpu_timers[3]; | |
b18b6a9c | 781 | #endif |
1da177e4 | 782 | |
5eca1c10 IM |
783 | /* Process credentials: */ |
784 | ||
785 | /* Tracer's credentials at attach: */ | |
786 | const struct cred __rcu *ptracer_cred; | |
787 | ||
788 | /* Objective and real subjective task credentials (COW): */ | |
789 | const struct cred __rcu *real_cred; | |
790 | ||
791 | /* Effective (overridable) subjective task credentials (COW): */ | |
792 | const struct cred __rcu *cred; | |
793 | ||
794 | /* | |
795 | * executable name, excluding path. | |
796 | * | |
797 | * - normally initialized setup_new_exec() | |
798 | * - access it with [gs]et_task_comm() | |
799 | * - lock it with task_lock() | |
800 | */ | |
801 | char comm[TASK_COMM_LEN]; | |
802 | ||
803 | struct nameidata *nameidata; | |
804 | ||
3d5b6fcc | 805 | #ifdef CONFIG_SYSVIPC |
5eca1c10 IM |
806 | struct sysv_sem sysvsem; |
807 | struct sysv_shm sysvshm; | |
3d5b6fcc | 808 | #endif |
e162b39a | 809 | #ifdef CONFIG_DETECT_HUNG_TASK |
5eca1c10 | 810 | unsigned long last_switch_count; |
82a1fcb9 | 811 | #endif |
5eca1c10 IM |
812 | /* Filesystem information: */ |
813 | struct fs_struct *fs; | |
814 | ||
815 | /* Open file information: */ | |
816 | struct files_struct *files; | |
817 | ||
818 | /* Namespaces: */ | |
819 | struct nsproxy *nsproxy; | |
820 | ||
821 | /* Signal handlers: */ | |
822 | struct signal_struct *signal; | |
823 | struct sighand_struct *sighand; | |
824 | sigset_t blocked; | |
825 | sigset_t real_blocked; | |
826 | /* Restored if set_restore_sigmask() was used: */ | |
827 | sigset_t saved_sigmask; | |
828 | struct sigpending pending; | |
829 | unsigned long sas_ss_sp; | |
830 | size_t sas_ss_size; | |
831 | unsigned int sas_ss_flags; | |
832 | ||
833 | struct callback_head *task_works; | |
834 | ||
835 | struct audit_context *audit_context; | |
bfef93a5 | 836 | #ifdef CONFIG_AUDITSYSCALL |
5eca1c10 IM |
837 | kuid_t loginuid; |
838 | unsigned int sessionid; | |
bfef93a5 | 839 | #endif |
5eca1c10 IM |
840 | struct seccomp seccomp; |
841 | ||
842 | /* Thread group tracking: */ | |
843 | u32 parent_exec_id; | |
844 | u32 self_exec_id; | |
1da177e4 | 845 | |
5eca1c10 IM |
846 | /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */ |
847 | spinlock_t alloc_lock; | |
1da177e4 | 848 | |
b29739f9 | 849 | /* Protection of the PI data structures: */ |
5eca1c10 | 850 | raw_spinlock_t pi_lock; |
b29739f9 | 851 | |
5eca1c10 | 852 | struct wake_q_node wake_q; |
76751049 | 853 | |
23f78d4a | 854 | #ifdef CONFIG_RT_MUTEXES |
5eca1c10 | 855 | /* PI waiters blocked on a rt_mutex held by this task: */ |
a23ba907 | 856 | struct rb_root_cached pi_waiters; |
e96a7705 XP |
857 | /* Updated under owner's pi_lock and rq lock */ |
858 | struct task_struct *pi_top_task; | |
5eca1c10 IM |
859 | /* Deadlock detection and priority inheritance handling: */ |
860 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
861 | #endif |
862 | ||
408894ee | 863 | #ifdef CONFIG_DEBUG_MUTEXES |
5eca1c10 IM |
864 | /* Mutex deadlock detection: */ |
865 | struct mutex_waiter *blocked_on; | |
408894ee | 866 | #endif |
5eca1c10 | 867 | |
de30a2b3 | 868 | #ifdef CONFIG_TRACE_IRQFLAGS |
5eca1c10 IM |
869 | unsigned int irq_events; |
870 | unsigned long hardirq_enable_ip; | |
871 | unsigned long hardirq_disable_ip; | |
872 | unsigned int hardirq_enable_event; | |
873 | unsigned int hardirq_disable_event; | |
874 | int hardirqs_enabled; | |
875 | int hardirq_context; | |
876 | unsigned long softirq_disable_ip; | |
877 | unsigned long softirq_enable_ip; | |
878 | unsigned int softirq_disable_event; | |
879 | unsigned int softirq_enable_event; | |
880 | int softirqs_enabled; | |
881 | int softirq_context; | |
de30a2b3 | 882 | #endif |
5eca1c10 | 883 | |
fbb9ce95 | 884 | #ifdef CONFIG_LOCKDEP |
5eca1c10 IM |
885 | # define MAX_LOCK_DEPTH 48UL |
886 | u64 curr_chain_key; | |
887 | int lockdep_depth; | |
888 | unsigned int lockdep_recursion; | |
889 | struct held_lock held_locks[MAX_LOCK_DEPTH]; | |
fbb9ce95 | 890 | #endif |
5eca1c10 | 891 | |
c6d30853 | 892 | #ifdef CONFIG_UBSAN |
5eca1c10 | 893 | unsigned int in_ubsan; |
c6d30853 | 894 | #endif |
408894ee | 895 | |
5eca1c10 IM |
896 | /* Journalling filesystem info: */ |
897 | void *journal_info; | |
1da177e4 | 898 | |
5eca1c10 IM |
899 | /* Stacked block device info: */ |
900 | struct bio_list *bio_list; | |
d89d8796 | 901 | |
73c10101 | 902 | #ifdef CONFIG_BLOCK |
5eca1c10 IM |
903 | /* Stack plugging: */ |
904 | struct blk_plug *plug; | |
73c10101 JA |
905 | #endif |
906 | ||
5eca1c10 IM |
907 | /* VM state: */ |
908 | struct reclaim_state *reclaim_state; | |
909 | ||
910 | struct backing_dev_info *backing_dev_info; | |
1da177e4 | 911 | |
5eca1c10 | 912 | struct io_context *io_context; |
1da177e4 | 913 | |
5eca1c10 IM |
914 | /* Ptrace state: */ |
915 | unsigned long ptrace_message; | |
916 | siginfo_t *last_siginfo; | |
1da177e4 | 917 | |
5eca1c10 IM |
918 | struct task_io_accounting ioac; |
919 | #ifdef CONFIG_TASK_XACCT | |
920 | /* Accumulated RSS usage: */ | |
921 | u64 acct_rss_mem1; | |
922 | /* Accumulated virtual memory usage: */ | |
923 | u64 acct_vm_mem1; | |
924 | /* stime + utime since last update: */ | |
925 | u64 acct_timexpd; | |
1da177e4 LT |
926 | #endif |
927 | #ifdef CONFIG_CPUSETS | |
5eca1c10 IM |
928 | /* Protected by ->alloc_lock: */ |
929 | nodemask_t mems_allowed; | |
930 | /* Seqence number to catch updates: */ | |
931 | seqcount_t mems_allowed_seq; | |
932 | int cpuset_mem_spread_rotor; | |
933 | int cpuset_slab_spread_rotor; | |
1da177e4 | 934 | #endif |
ddbcc7e8 | 935 | #ifdef CONFIG_CGROUPS |
5eca1c10 IM |
936 | /* Control Group info protected by css_set_lock: */ |
937 | struct css_set __rcu *cgroups; | |
938 | /* cg_list protected by css_set_lock and tsk->alloc_lock: */ | |
939 | struct list_head cg_list; | |
ddbcc7e8 | 940 | #endif |
f01d7d51 | 941 | #ifdef CONFIG_INTEL_RDT |
0734ded1 | 942 | u32 closid; |
d6aaba61 | 943 | u32 rmid; |
e02737d5 | 944 | #endif |
42b2dd0a | 945 | #ifdef CONFIG_FUTEX |
5eca1c10 | 946 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
947 | #ifdef CONFIG_COMPAT |
948 | struct compat_robust_list_head __user *compat_robust_list; | |
949 | #endif | |
5eca1c10 IM |
950 | struct list_head pi_state_list; |
951 | struct futex_pi_state *pi_state_cache; | |
c7aceaba | 952 | #endif |
cdd6c482 | 953 | #ifdef CONFIG_PERF_EVENTS |
5eca1c10 IM |
954 | struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; |
955 | struct mutex perf_event_mutex; | |
956 | struct list_head perf_event_list; | |
a63eaf34 | 957 | #endif |
8f47b187 | 958 | #ifdef CONFIG_DEBUG_PREEMPT |
5eca1c10 | 959 | unsigned long preempt_disable_ip; |
8f47b187 | 960 | #endif |
c7aceaba | 961 | #ifdef CONFIG_NUMA |
5eca1c10 IM |
962 | /* Protected by alloc_lock: */ |
963 | struct mempolicy *mempolicy; | |
45816682 | 964 | short il_prev; |
5eca1c10 | 965 | short pref_node_fork; |
42b2dd0a | 966 | #endif |
cbee9f88 | 967 | #ifdef CONFIG_NUMA_BALANCING |
5eca1c10 IM |
968 | int numa_scan_seq; |
969 | unsigned int numa_scan_period; | |
970 | unsigned int numa_scan_period_max; | |
971 | int numa_preferred_nid; | |
972 | unsigned long numa_migrate_retry; | |
973 | /* Migration stamp: */ | |
974 | u64 node_stamp; | |
975 | u64 last_task_numa_placement; | |
976 | u64 last_sum_exec_runtime; | |
977 | struct callback_head numa_work; | |
978 | ||
979 | struct list_head numa_entry; | |
980 | struct numa_group *numa_group; | |
8c8a743c | 981 | |
745d6147 | 982 | /* |
44dba3d5 IM |
983 | * numa_faults is an array split into four regions: |
984 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
985 | * in this precise order. | |
986 | * | |
987 | * faults_memory: Exponential decaying average of faults on a per-node | |
988 | * basis. Scheduling placement decisions are made based on these | |
989 | * counts. The values remain static for the duration of a PTE scan. | |
990 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
991 | * hinting fault was incurred. | |
992 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
993 | * during the current scan window. When the scan completes, the counts | |
994 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 995 | */ |
5eca1c10 IM |
996 | unsigned long *numa_faults; |
997 | unsigned long total_numa_faults; | |
745d6147 | 998 | |
04bb2f94 RR |
999 | /* |
1000 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
1001 | * scan window were remote/local or failed to migrate. The task scan |
1002 | * period is adapted based on the locality of the faults with different | |
1003 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 1004 | */ |
5eca1c10 | 1005 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 1006 | |
5eca1c10 | 1007 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
1008 | #endif /* CONFIG_NUMA_BALANCING */ |
1009 | ||
5eca1c10 | 1010 | struct tlbflush_unmap_batch tlb_ubc; |
72b252ae | 1011 | |
5eca1c10 | 1012 | struct rcu_head rcu; |
b92ce558 | 1013 | |
5eca1c10 IM |
1014 | /* Cache last used pipe for splice(): */ |
1015 | struct pipe_inode_info *splice_pipe; | |
5640f768 | 1016 | |
5eca1c10 | 1017 | struct page_frag task_frag; |
5640f768 | 1018 | |
47913d4e IM |
1019 | #ifdef CONFIG_TASK_DELAY_ACCT |
1020 | struct task_delay_info *delays; | |
f4f154fd | 1021 | #endif |
47913d4e | 1022 | |
f4f154fd | 1023 | #ifdef CONFIG_FAULT_INJECTION |
5eca1c10 | 1024 | int make_it_fail; |
9049f2f6 | 1025 | unsigned int fail_nth; |
ca74e92b | 1026 | #endif |
9d823e8f | 1027 | /* |
5eca1c10 IM |
1028 | * When (nr_dirtied >= nr_dirtied_pause), it's time to call |
1029 | * balance_dirty_pages() for a dirty throttling pause: | |
9d823e8f | 1030 | */ |
5eca1c10 IM |
1031 | int nr_dirtied; |
1032 | int nr_dirtied_pause; | |
1033 | /* Start of a write-and-pause period: */ | |
1034 | unsigned long dirty_paused_when; | |
9d823e8f | 1035 | |
9745512c | 1036 | #ifdef CONFIG_LATENCYTOP |
5eca1c10 IM |
1037 | int latency_record_count; |
1038 | struct latency_record latency_record[LT_SAVECOUNT]; | |
9745512c | 1039 | #endif |
6976675d | 1040 | /* |
5eca1c10 | 1041 | * Time slack values; these are used to round up poll() and |
6976675d AV |
1042 | * select() etc timeout values. These are in nanoseconds. |
1043 | */ | |
5eca1c10 IM |
1044 | u64 timer_slack_ns; |
1045 | u64 default_timer_slack_ns; | |
f8d570a4 | 1046 | |
0b24becc | 1047 | #ifdef CONFIG_KASAN |
5eca1c10 | 1048 | unsigned int kasan_depth; |
0b24becc | 1049 | #endif |
5eca1c10 | 1050 | |
fb52607a | 1051 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
5eca1c10 IM |
1052 | /* Index of current stored address in ret_stack: */ |
1053 | int curr_ret_stack; | |
1054 | ||
1055 | /* Stack of return addresses for return function tracing: */ | |
1056 | struct ftrace_ret_stack *ret_stack; | |
1057 | ||
1058 | /* Timestamp for last schedule: */ | |
1059 | unsigned long long ftrace_timestamp; | |
1060 | ||
f201ae23 FW |
1061 | /* |
1062 | * Number of functions that haven't been traced | |
5eca1c10 | 1063 | * because of depth overrun: |
f201ae23 | 1064 | */ |
5eca1c10 IM |
1065 | atomic_t trace_overrun; |
1066 | ||
1067 | /* Pause tracing: */ | |
1068 | atomic_t tracing_graph_pause; | |
f201ae23 | 1069 | #endif |
5eca1c10 | 1070 | |
ea4e2bc4 | 1071 | #ifdef CONFIG_TRACING |
5eca1c10 IM |
1072 | /* State flags for use by tracers: */ |
1073 | unsigned long trace; | |
1074 | ||
1075 | /* Bitmask and counter of trace recursion: */ | |
1076 | unsigned long trace_recursion; | |
261842b7 | 1077 | #endif /* CONFIG_TRACING */ |
5eca1c10 | 1078 | |
5c9a8750 | 1079 | #ifdef CONFIG_KCOV |
5eca1c10 IM |
1080 | /* Coverage collection mode enabled for this task (0 if disabled): */ |
1081 | enum kcov_mode kcov_mode; | |
1082 | ||
1083 | /* Size of the kcov_area: */ | |
1084 | unsigned int kcov_size; | |
1085 | ||
1086 | /* Buffer for coverage collection: */ | |
1087 | void *kcov_area; | |
1088 | ||
1089 | /* KCOV descriptor wired with this task or NULL: */ | |
1090 | struct kcov *kcov; | |
5c9a8750 | 1091 | #endif |
5eca1c10 | 1092 | |
6f185c29 | 1093 | #ifdef CONFIG_MEMCG |
5eca1c10 IM |
1094 | struct mem_cgroup *memcg_in_oom; |
1095 | gfp_t memcg_oom_gfp_mask; | |
1096 | int memcg_oom_order; | |
b23afb93 | 1097 | |
5eca1c10 IM |
1098 | /* Number of pages to reclaim on returning to userland: */ |
1099 | unsigned int memcg_nr_pages_over_high; | |
569b846d | 1100 | #endif |
5eca1c10 | 1101 | |
0326f5a9 | 1102 | #ifdef CONFIG_UPROBES |
5eca1c10 | 1103 | struct uprobe_task *utask; |
0326f5a9 | 1104 | #endif |
cafe5635 | 1105 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
5eca1c10 IM |
1106 | unsigned int sequential_io; |
1107 | unsigned int sequential_io_avg; | |
cafe5635 | 1108 | #endif |
8eb23b9f | 1109 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
5eca1c10 | 1110 | unsigned long task_state_change; |
8eb23b9f | 1111 | #endif |
5eca1c10 | 1112 | int pagefault_disabled; |
03049269 | 1113 | #ifdef CONFIG_MMU |
5eca1c10 | 1114 | struct task_struct *oom_reaper_list; |
03049269 | 1115 | #endif |
ba14a194 | 1116 | #ifdef CONFIG_VMAP_STACK |
5eca1c10 | 1117 | struct vm_struct *stack_vm_area; |
ba14a194 | 1118 | #endif |
68f24b08 | 1119 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 IM |
1120 | /* A live task holds one reference: */ |
1121 | atomic_t stack_refcount; | |
d83a7cb3 JP |
1122 | #endif |
1123 | #ifdef CONFIG_LIVEPATCH | |
1124 | int patch_state; | |
0302e28d | 1125 | #endif |
e4e55b47 TH |
1126 | #ifdef CONFIG_SECURITY |
1127 | /* Used by LSM modules for access restriction: */ | |
1128 | void *security; | |
68f24b08 | 1129 | #endif |
29e48ce8 KC |
1130 | |
1131 | /* | |
1132 | * New fields for task_struct should be added above here, so that | |
1133 | * they are included in the randomized portion of task_struct. | |
1134 | */ | |
1135 | randomized_struct_fields_end | |
1136 | ||
5eca1c10 IM |
1137 | /* CPU-specific state of this task: */ |
1138 | struct thread_struct thread; | |
1139 | ||
1140 | /* | |
1141 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1142 | * structure. It *MUST* be at the end of 'task_struct'. | |
1143 | * | |
1144 | * Do not put anything below here! | |
1145 | */ | |
1da177e4 LT |
1146 | }; |
1147 | ||
e868171a | 1148 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 EB |
1149 | { |
1150 | return task->pids[PIDTYPE_PID].pid; | |
1151 | } | |
1152 | ||
e868171a | 1153 | static inline struct pid *task_tgid(struct task_struct *task) |
22c935f4 EB |
1154 | { |
1155 | return task->group_leader->pids[PIDTYPE_PID].pid; | |
1156 | } | |
1157 | ||
6dda81f4 | 1158 | /* |
5eca1c10 | 1159 | * Without tasklist or RCU lock it is not safe to dereference |
6dda81f4 ON |
1160 | * the result of task_pgrp/task_session even if task == current, |
1161 | * we can race with another thread doing sys_setsid/sys_setpgid. | |
1162 | */ | |
e868171a | 1163 | static inline struct pid *task_pgrp(struct task_struct *task) |
22c935f4 EB |
1164 | { |
1165 | return task->group_leader->pids[PIDTYPE_PGID].pid; | |
1166 | } | |
1167 | ||
e868171a | 1168 | static inline struct pid *task_session(struct task_struct *task) |
22c935f4 EB |
1169 | { |
1170 | return task->group_leader->pids[PIDTYPE_SID].pid; | |
1171 | } | |
1172 | ||
7af57294 PE |
1173 | /* |
1174 | * the helpers to get the task's different pids as they are seen | |
1175 | * from various namespaces | |
1176 | * | |
1177 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
1178 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
1179 | * current. | |
7af57294 PE |
1180 | * task_xid_nr_ns() : id seen from the ns specified; |
1181 | * | |
7af57294 PE |
1182 | * see also pid_nr() etc in include/linux/pid.h |
1183 | */ | |
5eca1c10 | 1184 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns); |
7af57294 | 1185 | |
e868171a | 1186 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
1187 | { |
1188 | return tsk->pid; | |
1189 | } | |
1190 | ||
5eca1c10 | 1191 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
52ee2dfd ON |
1192 | { |
1193 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
1194 | } | |
7af57294 PE |
1195 | |
1196 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
1197 | { | |
52ee2dfd | 1198 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
1199 | } |
1200 | ||
1201 | ||
e868171a | 1202 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
1203 | { |
1204 | return tsk->tgid; | |
1205 | } | |
1206 | ||
5eca1c10 IM |
1207 | /** |
1208 | * pid_alive - check that a task structure is not stale | |
1209 | * @p: Task structure to be checked. | |
1210 | * | |
1211 | * Test if a process is not yet dead (at most zombie state) | |
1212 | * If pid_alive fails, then pointers within the task structure | |
1213 | * can be stale and must not be dereferenced. | |
1214 | * | |
1215 | * Return: 1 if the process is alive. 0 otherwise. | |
1216 | */ | |
1217 | static inline int pid_alive(const struct task_struct *p) | |
1218 | { | |
1219 | return p->pids[PIDTYPE_PID].pid != NULL; | |
1220 | } | |
7af57294 | 1221 | |
5eca1c10 | 1222 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1223 | { |
52ee2dfd | 1224 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
1225 | } |
1226 | ||
7af57294 PE |
1227 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
1228 | { | |
52ee2dfd | 1229 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
1230 | } |
1231 | ||
1232 | ||
5eca1c10 | 1233 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1234 | { |
52ee2dfd | 1235 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
1236 | } |
1237 | ||
7af57294 PE |
1238 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
1239 | { | |
52ee2dfd | 1240 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
1241 | } |
1242 | ||
dd1c1f2f ON |
1243 | static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
1244 | { | |
1245 | return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, ns); | |
1246 | } | |
1247 | ||
1248 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
1249 | { | |
1250 | return __task_pid_nr_ns(tsk, __PIDTYPE_TGID, NULL); | |
1251 | } | |
1252 | ||
1253 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) | |
1254 | { | |
1255 | pid_t pid = 0; | |
1256 | ||
1257 | rcu_read_lock(); | |
1258 | if (pid_alive(tsk)) | |
1259 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
1260 | rcu_read_unlock(); | |
1261 | ||
1262 | return pid; | |
1263 | } | |
1264 | ||
1265 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
1266 | { | |
1267 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
1268 | } | |
1269 | ||
5eca1c10 | 1270 | /* Obsolete, do not use: */ |
1b0f7ffd ON |
1271 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) |
1272 | { | |
1273 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
1274 | } | |
7af57294 | 1275 | |
06eb6184 PZ |
1276 | #define TASK_REPORT_IDLE (TASK_REPORT + 1) |
1277 | #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1) | |
1278 | ||
1d48b080 | 1279 | static inline unsigned int task_state_index(struct task_struct *tsk) |
20435d84 | 1280 | { |
1593baab PZ |
1281 | unsigned int tsk_state = READ_ONCE(tsk->state); |
1282 | unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT; | |
20435d84 | 1283 | |
06eb6184 PZ |
1284 | BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX); |
1285 | ||
06eb6184 PZ |
1286 | if (tsk_state == TASK_IDLE) |
1287 | state = TASK_REPORT_IDLE; | |
1288 | ||
1593baab PZ |
1289 | return fls(state); |
1290 | } | |
1291 | ||
1d48b080 | 1292 | static inline char task_index_to_char(unsigned int state) |
1593baab | 1293 | { |
8ef9925b | 1294 | static const char state_char[] = "RSDTtXZPI"; |
1593baab | 1295 | |
06eb6184 | 1296 | BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1); |
20435d84 | 1297 | |
1593baab PZ |
1298 | return state_char[state]; |
1299 | } | |
1300 | ||
1301 | static inline char task_state_to_char(struct task_struct *tsk) | |
1302 | { | |
1d48b080 | 1303 | return task_index_to_char(task_state_index(tsk)); |
20435d84 XX |
1304 | } |
1305 | ||
f400e198 | 1306 | /** |
570f5241 SS |
1307 | * is_global_init - check if a task structure is init. Since init |
1308 | * is free to have sub-threads we need to check tgid. | |
3260259f HK |
1309 | * @tsk: Task structure to be checked. |
1310 | * | |
1311 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
1312 | * |
1313 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 1314 | */ |
e868171a | 1315 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 1316 | { |
570f5241 | 1317 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 1318 | } |
b460cbc5 | 1319 | |
9ec52099 CLG |
1320 | extern struct pid *cad_pid; |
1321 | ||
1da177e4 LT |
1322 | /* |
1323 | * Per process flags | |
1324 | */ | |
5eca1c10 IM |
1325 | #define PF_IDLE 0x00000002 /* I am an IDLE thread */ |
1326 | #define PF_EXITING 0x00000004 /* Getting shut down */ | |
1327 | #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */ | |
1328 | #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ | |
1329 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ | |
1330 | #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */ | |
1331 | #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */ | |
1332 | #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */ | |
1333 | #define PF_DUMPCORE 0x00000200 /* Dumped core */ | |
1334 | #define PF_SIGNALED 0x00000400 /* Killed by a signal */ | |
1335 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
1336 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */ | |
1337 | #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */ | |
1338 | #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */ | |
1339 | #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */ | |
1340 | #define PF_FROZEN 0x00010000 /* Frozen for system suspend */ | |
7dea19f9 MH |
1341 | #define PF_KSWAPD 0x00020000 /* I am kswapd */ |
1342 | #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */ | |
1343 | #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */ | |
5eca1c10 IM |
1344 | #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ |
1345 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ | |
1346 | #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */ | |
1347 | #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ | |
1348 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */ | |
1349 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ | |
1350 | #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ | |
1351 | #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ | |
1352 | #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */ | |
1da177e4 LT |
1353 | |
1354 | /* | |
1355 | * Only the _current_ task can read/write to tsk->flags, but other | |
1356 | * tasks can access tsk->flags in readonly mode for example | |
1357 | * with tsk_used_math (like during threaded core dumping). | |
1358 | * There is however an exception to this rule during ptrace | |
1359 | * or during fork: the ptracer task is allowed to write to the | |
1360 | * child->flags of its traced child (same goes for fork, the parent | |
1361 | * can write to the child->flags), because we're guaranteed the | |
1362 | * child is not running and in turn not changing child->flags | |
1363 | * at the same time the parent does it. | |
1364 | */ | |
5eca1c10 IM |
1365 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) |
1366 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
1367 | #define clear_used_math() clear_stopped_child_used_math(current) | |
1368 | #define set_used_math() set_stopped_child_used_math(current) | |
1369 | ||
1da177e4 LT |
1370 | #define conditional_stopped_child_used_math(condition, child) \ |
1371 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
5eca1c10 IM |
1372 | |
1373 | #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current) | |
1374 | ||
1da177e4 LT |
1375 | #define copy_to_stopped_child_used_math(child) \ |
1376 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
5eca1c10 | 1377 | |
1da177e4 | 1378 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ |
5eca1c10 IM |
1379 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) |
1380 | #define used_math() tsk_used_math(current) | |
1da177e4 | 1381 | |
62ec05dd TG |
1382 | static inline bool is_percpu_thread(void) |
1383 | { | |
1384 | #ifdef CONFIG_SMP | |
1385 | return (current->flags & PF_NO_SETAFFINITY) && | |
1386 | (current->nr_cpus_allowed == 1); | |
1387 | #else | |
1388 | return true; | |
1389 | #endif | |
1390 | } | |
1391 | ||
1d4457f9 | 1392 | /* Per-process atomic flags. */ |
5eca1c10 IM |
1393 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
1394 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ | |
1395 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
733f4234 TG |
1396 | #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */ |
1397 | #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/ | |
1d4457f9 | 1398 | |
e0e5070b ZL |
1399 | #define TASK_PFA_TEST(name, func) \ |
1400 | static inline bool task_##func(struct task_struct *p) \ | |
1401 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1402 | |
e0e5070b ZL |
1403 | #define TASK_PFA_SET(name, func) \ |
1404 | static inline void task_set_##func(struct task_struct *p) \ | |
1405 | { set_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1406 | |
e0e5070b ZL |
1407 | #define TASK_PFA_CLEAR(name, func) \ |
1408 | static inline void task_clear_##func(struct task_struct *p) \ | |
1409 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
1410 | ||
1411 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
1412 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 1413 | |
2ad654bc ZL |
1414 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
1415 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
1416 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
1417 | ||
1418 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
1419 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
1420 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 1421 | |
733f4234 TG |
1422 | TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable) |
1423 | TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1424 | TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1425 | ||
1426 | TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) | |
1427 | TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) | |
1428 | ||
5eca1c10 | 1429 | static inline void |
717a94b5 | 1430 | current_restore_flags(unsigned long orig_flags, unsigned long flags) |
907aed48 | 1431 | { |
717a94b5 N |
1432 | current->flags &= ~flags; |
1433 | current->flags |= orig_flags & flags; | |
907aed48 MG |
1434 | } |
1435 | ||
5eca1c10 IM |
1436 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial); |
1437 | extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed); | |
1da177e4 | 1438 | #ifdef CONFIG_SMP |
5eca1c10 IM |
1439 | extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); |
1440 | extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask); | |
1da177e4 | 1441 | #else |
5eca1c10 | 1442 | static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) |
1e1b6c51 KM |
1443 | { |
1444 | } | |
5eca1c10 | 1445 | static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) |
1da177e4 | 1446 | { |
96f874e2 | 1447 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
1448 | return -EINVAL; |
1449 | return 0; | |
1450 | } | |
1451 | #endif | |
e0ad9556 | 1452 | |
6d0d2878 CB |
1453 | #ifndef cpu_relax_yield |
1454 | #define cpu_relax_yield() cpu_relax() | |
1455 | #endif | |
1456 | ||
fa93384f | 1457 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
1458 | extern void set_user_nice(struct task_struct *p, long nice); |
1459 | extern int task_prio(const struct task_struct *p); | |
5eca1c10 | 1460 | |
d0ea0268 DY |
1461 | /** |
1462 | * task_nice - return the nice value of a given task. | |
1463 | * @p: the task in question. | |
1464 | * | |
1465 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
1466 | */ | |
1467 | static inline int task_nice(const struct task_struct *p) | |
1468 | { | |
1469 | return PRIO_TO_NICE((p)->static_prio); | |
1470 | } | |
5eca1c10 | 1471 | |
36c8b586 IM |
1472 | extern int can_nice(const struct task_struct *p, const int nice); |
1473 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 1474 | extern int idle_cpu(int cpu); |
5eca1c10 IM |
1475 | extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *); |
1476 | extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *); | |
1477 | extern int sched_setattr(struct task_struct *, const struct sched_attr *); | |
36c8b586 | 1478 | extern struct task_struct *idle_task(int cpu); |
5eca1c10 | 1479 | |
c4f30608 PM |
1480 | /** |
1481 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 1482 | * @p: the task in question. |
e69f6186 YB |
1483 | * |
1484 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 1485 | */ |
7061ca3b | 1486 | static inline bool is_idle_task(const struct task_struct *p) |
c4f30608 | 1487 | { |
c1de45ca | 1488 | return !!(p->flags & PF_IDLE); |
c4f30608 | 1489 | } |
5eca1c10 | 1490 | |
36c8b586 | 1491 | extern struct task_struct *curr_task(int cpu); |
a458ae2e | 1492 | extern void ia64_set_curr_task(int cpu, struct task_struct *p); |
1da177e4 LT |
1493 | |
1494 | void yield(void); | |
1495 | ||
1da177e4 | 1496 | union thread_union { |
c65eacbe | 1497 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1da177e4 | 1498 | struct thread_info thread_info; |
c65eacbe | 1499 | #endif |
1da177e4 LT |
1500 | unsigned long stack[THREAD_SIZE/sizeof(long)]; |
1501 | }; | |
1502 | ||
f3ac6067 IM |
1503 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1504 | static inline struct thread_info *task_thread_info(struct task_struct *task) | |
1505 | { | |
1506 | return &task->thread_info; | |
1507 | } | |
1508 | #elif !defined(__HAVE_THREAD_FUNCTIONS) | |
1509 | # define task_thread_info(task) ((struct thread_info *)(task)->stack) | |
1510 | #endif | |
1511 | ||
198fe21b PE |
1512 | /* |
1513 | * find a task by one of its numerical ids | |
1514 | * | |
198fe21b PE |
1515 | * find_task_by_pid_ns(): |
1516 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
1517 | * find_task_by_vpid(): |
1518 | * finds a task by its virtual pid | |
198fe21b | 1519 | * |
e49859e7 | 1520 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
1521 | */ |
1522 | ||
228ebcbe | 1523 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
5eca1c10 | 1524 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns); |
198fe21b | 1525 | |
b3c97528 HH |
1526 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
1527 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 1528 | extern void wake_up_new_task(struct task_struct *tsk); |
5eca1c10 | 1529 | |
1da177e4 | 1530 | #ifdef CONFIG_SMP |
5eca1c10 | 1531 | extern void kick_process(struct task_struct *tsk); |
1da177e4 | 1532 | #else |
5eca1c10 | 1533 | static inline void kick_process(struct task_struct *tsk) { } |
1da177e4 | 1534 | #endif |
1da177e4 | 1535 | |
82b89778 | 1536 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
5eca1c10 | 1537 | |
82b89778 AH |
1538 | static inline void set_task_comm(struct task_struct *tsk, const char *from) |
1539 | { | |
1540 | __set_task_comm(tsk, from, false); | |
1541 | } | |
5eca1c10 | 1542 | |
3756f640 AB |
1543 | extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk); |
1544 | #define get_task_comm(buf, tsk) ({ \ | |
1545 | BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \ | |
1546 | __get_task_comm(buf, sizeof(buf), tsk); \ | |
1547 | }) | |
1da177e4 LT |
1548 | |
1549 | #ifdef CONFIG_SMP | |
317f3941 | 1550 | void scheduler_ipi(void); |
85ba2d86 | 1551 | extern unsigned long wait_task_inactive(struct task_struct *, long match_state); |
1da177e4 | 1552 | #else |
184748cc | 1553 | static inline void scheduler_ipi(void) { } |
5eca1c10 | 1554 | static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state) |
85ba2d86 RM |
1555 | { |
1556 | return 1; | |
1557 | } | |
1da177e4 LT |
1558 | #endif |
1559 | ||
5eca1c10 IM |
1560 | /* |
1561 | * Set thread flags in other task's structures. | |
1562 | * See asm/thread_info.h for TIF_xxxx flags available: | |
1da177e4 LT |
1563 | */ |
1564 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1565 | { | |
a1261f54 | 1566 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1567 | } |
1568 | ||
1569 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1570 | { | |
a1261f54 | 1571 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1572 | } |
1573 | ||
1574 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1575 | { | |
a1261f54 | 1576 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1577 | } |
1578 | ||
1579 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1580 | { | |
a1261f54 | 1581 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1582 | } |
1583 | ||
1584 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1585 | { | |
a1261f54 | 1586 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1587 | } |
1588 | ||
1589 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
1590 | { | |
1591 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1592 | } | |
1593 | ||
1594 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
1595 | { | |
1596 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1597 | } | |
1598 | ||
8ae121ac GH |
1599 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
1600 | { | |
1601 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
1602 | } | |
1603 | ||
1da177e4 LT |
1604 | /* |
1605 | * cond_resched() and cond_resched_lock(): latency reduction via | |
1606 | * explicit rescheduling in places that are safe. The return | |
1607 | * value indicates whether a reschedule was done in fact. | |
1608 | * cond_resched_lock() will drop the spinlock before scheduling, | |
1609 | * cond_resched_softirq() will enable bhs before scheduling. | |
1610 | */ | |
35a773a0 | 1611 | #ifndef CONFIG_PREEMPT |
c3921ab7 | 1612 | extern int _cond_resched(void); |
35a773a0 PZ |
1613 | #else |
1614 | static inline int _cond_resched(void) { return 0; } | |
1615 | #endif | |
6f80bd98 | 1616 | |
613afbf8 | 1617 | #define cond_resched() ({ \ |
3427445a | 1618 | ___might_sleep(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
1619 | _cond_resched(); \ |
1620 | }) | |
6f80bd98 | 1621 | |
613afbf8 FW |
1622 | extern int __cond_resched_lock(spinlock_t *lock); |
1623 | ||
1624 | #define cond_resched_lock(lock) ({ \ | |
3427445a | 1625 | ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ |
613afbf8 FW |
1626 | __cond_resched_lock(lock); \ |
1627 | }) | |
1628 | ||
1629 | extern int __cond_resched_softirq(void); | |
1630 | ||
75e1056f | 1631 | #define cond_resched_softirq() ({ \ |
3427445a | 1632 | ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \ |
75e1056f | 1633 | __cond_resched_softirq(); \ |
613afbf8 | 1634 | }) |
1da177e4 | 1635 | |
f6f3c437 SH |
1636 | static inline void cond_resched_rcu(void) |
1637 | { | |
1638 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
1639 | rcu_read_unlock(); | |
1640 | cond_resched(); | |
1641 | rcu_read_lock(); | |
1642 | #endif | |
1643 | } | |
1644 | ||
1da177e4 LT |
1645 | /* |
1646 | * Does a critical section need to be broken due to another | |
95c354fe NP |
1647 | * task waiting?: (technically does not depend on CONFIG_PREEMPT, |
1648 | * but a general need for low latency) | |
1da177e4 | 1649 | */ |
95c354fe | 1650 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 1651 | { |
95c354fe NP |
1652 | #ifdef CONFIG_PREEMPT |
1653 | return spin_is_contended(lock); | |
1654 | #else | |
1da177e4 | 1655 | return 0; |
95c354fe | 1656 | #endif |
1da177e4 LT |
1657 | } |
1658 | ||
75f93fed PZ |
1659 | static __always_inline bool need_resched(void) |
1660 | { | |
1661 | return unlikely(tif_need_resched()); | |
1662 | } | |
1663 | ||
1da177e4 LT |
1664 | /* |
1665 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
1666 | */ | |
1667 | #ifdef CONFIG_SMP | |
1668 | ||
1669 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1670 | { | |
c65eacbe AL |
1671 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1672 | return p->cpu; | |
1673 | #else | |
a1261f54 | 1674 | return task_thread_info(p)->cpu; |
c65eacbe | 1675 | #endif |
1da177e4 LT |
1676 | } |
1677 | ||
c65cc870 | 1678 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
1679 | |
1680 | #else | |
1681 | ||
1682 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1683 | { | |
1684 | return 0; | |
1685 | } | |
1686 | ||
1687 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
1688 | { | |
1689 | } | |
1690 | ||
1691 | #endif /* CONFIG_SMP */ | |
1692 | ||
d9345c65 PX |
1693 | /* |
1694 | * In order to reduce various lock holder preemption latencies provide an | |
1695 | * interface to see if a vCPU is currently running or not. | |
1696 | * | |
1697 | * This allows us to terminate optimistic spin loops and block, analogous to | |
1698 | * the native optimistic spin heuristic of testing if the lock owner task is | |
1699 | * running or not. | |
1700 | */ | |
1701 | #ifndef vcpu_is_preempted | |
1702 | # define vcpu_is_preempted(cpu) false | |
1703 | #endif | |
1704 | ||
96f874e2 RR |
1705 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
1706 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 1707 | |
82455257 DH |
1708 | #ifndef TASK_SIZE_OF |
1709 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
1710 | #endif | |
1711 | ||
1da177e4 | 1712 | #endif |