]>
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> |
ec1d2819 | 24 | #include <linux/refcount.h> |
a3b6714e | 25 | #include <linux/resource.h> |
9745512c | 26 | #include <linux/latencytop.h> |
5eca1c10 IM |
27 | #include <linux/sched/prio.h> |
28 | #include <linux/signal_types.h> | |
29 | #include <linux/mm_types_task.h> | |
30 | #include <linux/task_io_accounting.h> | |
d7822b1e | 31 | #include <linux/rseq.h> |
a3b6714e | 32 | |
5eca1c10 | 33 | /* task_struct member predeclarations (sorted alphabetically): */ |
c7af7877 | 34 | struct audit_context; |
c7af7877 | 35 | struct backing_dev_info; |
bddd87c7 | 36 | struct bio_list; |
73c10101 | 37 | struct blk_plug; |
3c93a0c0 | 38 | struct capture_control; |
c7af7877 | 39 | struct cfs_rq; |
c7af7877 IM |
40 | struct fs_struct; |
41 | struct futex_pi_state; | |
42 | struct io_context; | |
43 | struct mempolicy; | |
89076bc3 | 44 | struct nameidata; |
c7af7877 IM |
45 | struct nsproxy; |
46 | struct perf_event_context; | |
47 | struct pid_namespace; | |
48 | struct pipe_inode_info; | |
49 | struct rcu_node; | |
50 | struct reclaim_state; | |
51 | struct robust_list_head; | |
3c93a0c0 QY |
52 | struct root_domain; |
53 | struct rq; | |
c7af7877 IM |
54 | struct sched_attr; |
55 | struct sched_param; | |
43ae34cb | 56 | struct seq_file; |
c7af7877 IM |
57 | struct sighand_struct; |
58 | struct signal_struct; | |
59 | struct task_delay_info; | |
4cf86d77 | 60 | struct task_group; |
1da177e4 | 61 | |
4a8342d2 LT |
62 | /* |
63 | * Task state bitmask. NOTE! These bits are also | |
64 | * encoded in fs/proc/array.c: get_task_state(). | |
65 | * | |
66 | * We have two separate sets of flags: task->state | |
67 | * is about runnability, while task->exit_state are | |
68 | * about the task exiting. Confusing, but this way | |
69 | * modifying one set can't modify the other one by | |
70 | * mistake. | |
71 | */ | |
5eca1c10 IM |
72 | |
73 | /* Used in tsk->state: */ | |
92c4bc9f PZ |
74 | #define TASK_RUNNING 0x0000 |
75 | #define TASK_INTERRUPTIBLE 0x0001 | |
76 | #define TASK_UNINTERRUPTIBLE 0x0002 | |
77 | #define __TASK_STOPPED 0x0004 | |
78 | #define __TASK_TRACED 0x0008 | |
5eca1c10 | 79 | /* Used in tsk->exit_state: */ |
92c4bc9f PZ |
80 | #define EXIT_DEAD 0x0010 |
81 | #define EXIT_ZOMBIE 0x0020 | |
5eca1c10 IM |
82 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) |
83 | /* Used in tsk->state again: */ | |
8ef9925b PZ |
84 | #define TASK_PARKED 0x0040 |
85 | #define TASK_DEAD 0x0080 | |
86 | #define TASK_WAKEKILL 0x0100 | |
87 | #define TASK_WAKING 0x0200 | |
92c4bc9f PZ |
88 | #define TASK_NOLOAD 0x0400 |
89 | #define TASK_NEW 0x0800 | |
90 | #define TASK_STATE_MAX 0x1000 | |
5eca1c10 | 91 | |
5eca1c10 IM |
92 | /* Convenience macros for the sake of set_current_state: */ |
93 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) | |
94 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
95 | #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) | |
96 | ||
97 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) | |
98 | ||
99 | /* Convenience macros for the sake of wake_up(): */ | |
100 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
5eca1c10 IM |
101 | |
102 | /* get_task_state(): */ | |
103 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
104 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ | |
8ef9925b PZ |
105 | __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \ |
106 | TASK_PARKED) | |
5eca1c10 IM |
107 | |
108 | #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) | |
109 | ||
110 | #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) | |
111 | ||
112 | #define task_is_stopped_or_traced(task) ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) | |
113 | ||
114 | #define task_contributes_to_load(task) ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ | |
115 | (task->flags & PF_FROZEN) == 0 && \ | |
116 | (task->state & TASK_NOLOAD) == 0) | |
1da177e4 | 117 | |
8eb23b9f PZ |
118 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
119 | ||
b5bf9a90 PZ |
120 | /* |
121 | * Special states are those that do not use the normal wait-loop pattern. See | |
122 | * the comment with set_special_state(). | |
123 | */ | |
124 | #define is_special_task_state(state) \ | |
1cef1150 | 125 | ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD)) |
b5bf9a90 | 126 | |
8eb23b9f PZ |
127 | #define __set_current_state(state_value) \ |
128 | do { \ | |
b5bf9a90 | 129 | WARN_ON_ONCE(is_special_task_state(state_value));\ |
8eb23b9f PZ |
130 | current->task_state_change = _THIS_IP_; \ |
131 | current->state = (state_value); \ | |
132 | } while (0) | |
b5bf9a90 | 133 | |
8eb23b9f PZ |
134 | #define set_current_state(state_value) \ |
135 | do { \ | |
b5bf9a90 | 136 | WARN_ON_ONCE(is_special_task_state(state_value));\ |
8eb23b9f | 137 | current->task_state_change = _THIS_IP_; \ |
a2250238 | 138 | smp_store_mb(current->state, (state_value)); \ |
8eb23b9f PZ |
139 | } while (0) |
140 | ||
b5bf9a90 PZ |
141 | #define set_special_state(state_value) \ |
142 | do { \ | |
143 | unsigned long flags; /* may shadow */ \ | |
144 | WARN_ON_ONCE(!is_special_task_state(state_value)); \ | |
145 | raw_spin_lock_irqsave(¤t->pi_lock, flags); \ | |
146 | current->task_state_change = _THIS_IP_; \ | |
147 | current->state = (state_value); \ | |
148 | raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ | |
149 | } while (0) | |
8eb23b9f | 150 | #else |
498d0c57 AM |
151 | /* |
152 | * set_current_state() includes a barrier so that the write of current->state | |
153 | * is correctly serialised wrt the caller's subsequent test of whether to | |
154 | * actually sleep: | |
155 | * | |
a2250238 | 156 | * for (;;) { |
498d0c57 | 157 | * set_current_state(TASK_UNINTERRUPTIBLE); |
a2250238 PZ |
158 | * if (!need_sleep) |
159 | * break; | |
160 | * | |
161 | * schedule(); | |
162 | * } | |
163 | * __set_current_state(TASK_RUNNING); | |
164 | * | |
165 | * If the caller does not need such serialisation (because, for instance, the | |
166 | * condition test and condition change and wakeup are under the same lock) then | |
167 | * use __set_current_state(). | |
168 | * | |
169 | * The above is typically ordered against the wakeup, which does: | |
170 | * | |
b5bf9a90 PZ |
171 | * need_sleep = false; |
172 | * wake_up_state(p, TASK_UNINTERRUPTIBLE); | |
a2250238 | 173 | * |
7696f991 AP |
174 | * where wake_up_state() executes a full memory barrier before accessing the |
175 | * task state. | |
a2250238 PZ |
176 | * |
177 | * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, | |
178 | * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a | |
179 | * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). | |
498d0c57 | 180 | * |
b5bf9a90 | 181 | * However, with slightly different timing the wakeup TASK_RUNNING store can |
dfcb245e | 182 | * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not |
b5bf9a90 PZ |
183 | * a problem either because that will result in one extra go around the loop |
184 | * and our @cond test will save the day. | |
498d0c57 | 185 | * |
a2250238 | 186 | * Also see the comments of try_to_wake_up(). |
498d0c57 | 187 | */ |
b5bf9a90 PZ |
188 | #define __set_current_state(state_value) \ |
189 | current->state = (state_value) | |
190 | ||
191 | #define set_current_state(state_value) \ | |
192 | smp_store_mb(current->state, (state_value)) | |
193 | ||
194 | /* | |
195 | * set_special_state() should be used for those states when the blocking task | |
196 | * can not use the regular condition based wait-loop. In that case we must | |
197 | * serialize against wakeups such that any possible in-flight TASK_RUNNING stores | |
198 | * will not collide with our state change. | |
199 | */ | |
200 | #define set_special_state(state_value) \ | |
201 | do { \ | |
202 | unsigned long flags; /* may shadow */ \ | |
203 | raw_spin_lock_irqsave(¤t->pi_lock, flags); \ | |
204 | current->state = (state_value); \ | |
205 | raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ | |
206 | } while (0) | |
207 | ||
8eb23b9f PZ |
208 | #endif |
209 | ||
5eca1c10 IM |
210 | /* Task command name length: */ |
211 | #define TASK_COMM_LEN 16 | |
1da177e4 | 212 | |
1da177e4 LT |
213 | extern void scheduler_tick(void); |
214 | ||
5eca1c10 IM |
215 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX |
216 | ||
217 | extern long schedule_timeout(long timeout); | |
218 | extern long schedule_timeout_interruptible(long timeout); | |
219 | extern long schedule_timeout_killable(long timeout); | |
220 | extern long schedule_timeout_uninterruptible(long timeout); | |
221 | extern long schedule_timeout_idle(long timeout); | |
1da177e4 | 222 | asmlinkage void schedule(void); |
c5491ea7 | 223 | extern void schedule_preempt_disabled(void); |
1da177e4 | 224 | |
10ab5643 TH |
225 | extern int __must_check io_schedule_prepare(void); |
226 | extern void io_schedule_finish(int token); | |
9cff8ade | 227 | extern long io_schedule_timeout(long timeout); |
10ab5643 | 228 | extern void io_schedule(void); |
9cff8ade | 229 | |
d37f761d | 230 | /** |
0ba42a59 | 231 | * struct prev_cputime - snapshot of system and user cputime |
d37f761d FW |
232 | * @utime: time spent in user mode |
233 | * @stime: time spent in system mode | |
9d7fb042 | 234 | * @lock: protects the above two fields |
d37f761d | 235 | * |
9d7fb042 PZ |
236 | * Stores previous user/system time values such that we can guarantee |
237 | * monotonicity. | |
d37f761d | 238 | */ |
9d7fb042 PZ |
239 | struct prev_cputime { |
240 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
5eca1c10 IM |
241 | u64 utime; |
242 | u64 stime; | |
243 | raw_spinlock_t lock; | |
9d7fb042 | 244 | #endif |
d37f761d FW |
245 | }; |
246 | ||
f06febc9 FM |
247 | /** |
248 | * struct task_cputime - collected CPU time counts | |
5613fda9 FW |
249 | * @utime: time spent in user mode, in nanoseconds |
250 | * @stime: time spent in kernel mode, in nanoseconds | |
f06febc9 | 251 | * @sum_exec_runtime: total time spent on the CPU, in nanoseconds |
5ce73a4a | 252 | * |
9d7fb042 PZ |
253 | * This structure groups together three kinds of CPU time that are tracked for |
254 | * threads and thread groups. Most things considering CPU time want to group | |
255 | * these counts together and treat all three of them in parallel. | |
f06febc9 FM |
256 | */ |
257 | struct task_cputime { | |
5eca1c10 IM |
258 | u64 utime; |
259 | u64 stime; | |
260 | unsigned long long sum_exec_runtime; | |
f06febc9 | 261 | }; |
9d7fb042 | 262 | |
5eca1c10 IM |
263 | /* Alternate field names when used on cache expirations: */ |
264 | #define virt_exp utime | |
265 | #define prof_exp stime | |
266 | #define sched_exp sum_exec_runtime | |
f06febc9 | 267 | |
bac5b6b6 FW |
268 | enum vtime_state { |
269 | /* Task is sleeping or running in a CPU with VTIME inactive: */ | |
270 | VTIME_INACTIVE = 0, | |
271 | /* Task runs in userspace in a CPU with VTIME active: */ | |
272 | VTIME_USER, | |
273 | /* Task runs in kernelspace in a CPU with VTIME active: */ | |
274 | VTIME_SYS, | |
275 | }; | |
276 | ||
277 | struct vtime { | |
278 | seqcount_t seqcount; | |
279 | unsigned long long starttime; | |
280 | enum vtime_state state; | |
2a42eb95 WL |
281 | u64 utime; |
282 | u64 stime; | |
283 | u64 gtime; | |
bac5b6b6 FW |
284 | }; |
285 | ||
69842cba PB |
286 | /* |
287 | * Utilization clamp constraints. | |
288 | * @UCLAMP_MIN: Minimum utilization | |
289 | * @UCLAMP_MAX: Maximum utilization | |
290 | * @UCLAMP_CNT: Utilization clamp constraints count | |
291 | */ | |
292 | enum uclamp_id { | |
293 | UCLAMP_MIN = 0, | |
294 | UCLAMP_MAX, | |
295 | UCLAMP_CNT | |
296 | }; | |
297 | ||
1da177e4 | 298 | struct sched_info { |
7f5f8e8d | 299 | #ifdef CONFIG_SCHED_INFO |
5eca1c10 IM |
300 | /* Cumulative counters: */ |
301 | ||
302 | /* # of times we have run on this CPU: */ | |
303 | unsigned long pcount; | |
304 | ||
305 | /* Time spent waiting on a runqueue: */ | |
306 | unsigned long long run_delay; | |
307 | ||
308 | /* Timestamps: */ | |
309 | ||
310 | /* When did we last run on a CPU? */ | |
311 | unsigned long long last_arrival; | |
312 | ||
313 | /* When were we last queued to run? */ | |
314 | unsigned long long last_queued; | |
1da177e4 | 315 | |
f6db8347 | 316 | #endif /* CONFIG_SCHED_INFO */ |
7f5f8e8d | 317 | }; |
1da177e4 | 318 | |
6ecdd749 YD |
319 | /* |
320 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
321 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
322 | * | |
323 | * We define a basic fixed point arithmetic range, and then formalize | |
324 | * all these metrics based on that basic range. | |
325 | */ | |
5eca1c10 IM |
326 | # define SCHED_FIXEDPOINT_SHIFT 10 |
327 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
6ecdd749 | 328 | |
69842cba PB |
329 | /* Increase resolution of cpu_capacity calculations */ |
330 | # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT | |
331 | # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT) | |
332 | ||
20b8a59f | 333 | struct load_weight { |
5eca1c10 IM |
334 | unsigned long weight; |
335 | u32 inv_weight; | |
20b8a59f IM |
336 | }; |
337 | ||
7f65ea42 PB |
338 | /** |
339 | * struct util_est - Estimation utilization of FAIR tasks | |
340 | * @enqueued: instantaneous estimated utilization of a task/cpu | |
341 | * @ewma: the Exponential Weighted Moving Average (EWMA) | |
342 | * utilization of a task | |
343 | * | |
344 | * Support data structure to track an Exponential Weighted Moving Average | |
345 | * (EWMA) of a FAIR task's utilization. New samples are added to the moving | |
346 | * average each time a task completes an activation. Sample's weight is chosen | |
347 | * so that the EWMA will be relatively insensitive to transient changes to the | |
348 | * task's workload. | |
349 | * | |
350 | * The enqueued attribute has a slightly different meaning for tasks and cpus: | |
351 | * - task: the task's util_avg at last task dequeue time | |
352 | * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU | |
353 | * Thus, the util_est.enqueued of a task represents the contribution on the | |
354 | * estimated utilization of the CPU where that task is currently enqueued. | |
355 | * | |
356 | * Only for tasks we track a moving average of the past instantaneous | |
357 | * estimated utilization. This allows to absorb sporadic drops in utilization | |
358 | * of an otherwise almost periodic task. | |
359 | */ | |
360 | struct util_est { | |
361 | unsigned int enqueued; | |
362 | unsigned int ewma; | |
363 | #define UTIL_EST_WEIGHT_SHIFT 2 | |
317d359d | 364 | } __attribute__((__aligned__(sizeof(u64)))); |
7f65ea42 | 365 | |
9d89c257 | 366 | /* |
7b595334 YD |
367 | * The load_avg/util_avg accumulates an infinite geometric series |
368 | * (see __update_load_avg() in kernel/sched/fair.c). | |
369 | * | |
370 | * [load_avg definition] | |
371 | * | |
372 | * load_avg = runnable% * scale_load_down(load) | |
373 | * | |
374 | * where runnable% is the time ratio that a sched_entity is runnable. | |
375 | * For cfs_rq, it is the aggregated load_avg of all runnable and | |
9d89c257 | 376 | * blocked sched_entities. |
7b595334 | 377 | * |
7b595334 YD |
378 | * [util_avg definition] |
379 | * | |
380 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
381 | * | |
382 | * where running% is the time ratio that a sched_entity is running on | |
383 | * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable | |
384 | * and blocked sched_entities. | |
385 | * | |
23127296 VG |
386 | * load_avg and util_avg don't direcly factor frequency scaling and CPU |
387 | * capacity scaling. The scaling is done through the rq_clock_pelt that | |
388 | * is used for computing those signals (see update_rq_clock_pelt()) | |
7b595334 | 389 | * |
23127296 VG |
390 | * N.B., the above ratios (runnable% and running%) themselves are in the |
391 | * range of [0, 1]. To do fixed point arithmetics, we therefore scale them | |
392 | * to as large a range as necessary. This is for example reflected by | |
393 | * util_avg's SCHED_CAPACITY_SCALE. | |
7b595334 YD |
394 | * |
395 | * [Overflow issue] | |
396 | * | |
397 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
398 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
399 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
400 | * | |
401 | * For all other cases (including 32-bit kernels), struct load_weight's | |
402 | * weight will overflow first before we do, because: | |
403 | * | |
404 | * Max(load_avg) <= Max(load.weight) | |
405 | * | |
406 | * Then it is the load_weight's responsibility to consider overflow | |
407 | * issues. | |
9d89c257 | 408 | */ |
9d85f21c | 409 | struct sched_avg { |
5eca1c10 IM |
410 | u64 last_update_time; |
411 | u64 load_sum; | |
1ea6c46a | 412 | u64 runnable_load_sum; |
5eca1c10 IM |
413 | u32 util_sum; |
414 | u32 period_contrib; | |
415 | unsigned long load_avg; | |
1ea6c46a | 416 | unsigned long runnable_load_avg; |
5eca1c10 | 417 | unsigned long util_avg; |
7f65ea42 | 418 | struct util_est util_est; |
317d359d | 419 | } ____cacheline_aligned; |
9d85f21c | 420 | |
41acab88 | 421 | struct sched_statistics { |
7f5f8e8d | 422 | #ifdef CONFIG_SCHEDSTATS |
5eca1c10 IM |
423 | u64 wait_start; |
424 | u64 wait_max; | |
425 | u64 wait_count; | |
426 | u64 wait_sum; | |
427 | u64 iowait_count; | |
428 | u64 iowait_sum; | |
429 | ||
430 | u64 sleep_start; | |
431 | u64 sleep_max; | |
432 | s64 sum_sleep_runtime; | |
433 | ||
434 | u64 block_start; | |
435 | u64 block_max; | |
436 | u64 exec_max; | |
437 | u64 slice_max; | |
438 | ||
439 | u64 nr_migrations_cold; | |
440 | u64 nr_failed_migrations_affine; | |
441 | u64 nr_failed_migrations_running; | |
442 | u64 nr_failed_migrations_hot; | |
443 | u64 nr_forced_migrations; | |
444 | ||
445 | u64 nr_wakeups; | |
446 | u64 nr_wakeups_sync; | |
447 | u64 nr_wakeups_migrate; | |
448 | u64 nr_wakeups_local; | |
449 | u64 nr_wakeups_remote; | |
450 | u64 nr_wakeups_affine; | |
451 | u64 nr_wakeups_affine_attempts; | |
452 | u64 nr_wakeups_passive; | |
453 | u64 nr_wakeups_idle; | |
41acab88 | 454 | #endif |
7f5f8e8d | 455 | }; |
41acab88 LDM |
456 | |
457 | struct sched_entity { | |
5eca1c10 IM |
458 | /* For load-balancing: */ |
459 | struct load_weight load; | |
1ea6c46a | 460 | unsigned long runnable_weight; |
5eca1c10 IM |
461 | struct rb_node run_node; |
462 | struct list_head group_node; | |
463 | unsigned int on_rq; | |
41acab88 | 464 | |
5eca1c10 IM |
465 | u64 exec_start; |
466 | u64 sum_exec_runtime; | |
467 | u64 vruntime; | |
468 | u64 prev_sum_exec_runtime; | |
41acab88 | 469 | |
5eca1c10 | 470 | u64 nr_migrations; |
41acab88 | 471 | |
5eca1c10 | 472 | struct sched_statistics statistics; |
94c18227 | 473 | |
20b8a59f | 474 | #ifdef CONFIG_FAIR_GROUP_SCHED |
5eca1c10 IM |
475 | int depth; |
476 | struct sched_entity *parent; | |
20b8a59f | 477 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 478 | struct cfs_rq *cfs_rq; |
20b8a59f | 479 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 480 | struct cfs_rq *my_q; |
20b8a59f | 481 | #endif |
8bd75c77 | 482 | |
141965c7 | 483 | #ifdef CONFIG_SMP |
5a107804 JO |
484 | /* |
485 | * Per entity load average tracking. | |
486 | * | |
487 | * Put into separate cache line so it does not | |
488 | * collide with read-mostly values above. | |
489 | */ | |
317d359d | 490 | struct sched_avg avg; |
9d85f21c | 491 | #endif |
20b8a59f | 492 | }; |
70b97a7f | 493 | |
fa717060 | 494 | struct sched_rt_entity { |
5eca1c10 IM |
495 | struct list_head run_list; |
496 | unsigned long timeout; | |
497 | unsigned long watchdog_stamp; | |
498 | unsigned int time_slice; | |
499 | unsigned short on_rq; | |
500 | unsigned short on_list; | |
501 | ||
502 | struct sched_rt_entity *back; | |
052f1dc7 | 503 | #ifdef CONFIG_RT_GROUP_SCHED |
5eca1c10 | 504 | struct sched_rt_entity *parent; |
6f505b16 | 505 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 506 | struct rt_rq *rt_rq; |
6f505b16 | 507 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 508 | struct rt_rq *my_q; |
6f505b16 | 509 | #endif |
3859a271 | 510 | } __randomize_layout; |
fa717060 | 511 | |
aab03e05 | 512 | struct sched_dl_entity { |
5eca1c10 | 513 | struct rb_node rb_node; |
aab03e05 DF |
514 | |
515 | /* | |
516 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 517 | * during sched_setattr(), they will remain the same until |
518 | * the next sched_setattr(). | |
aab03e05 | 519 | */ |
5eca1c10 IM |
520 | u64 dl_runtime; /* Maximum runtime for each instance */ |
521 | u64 dl_deadline; /* Relative deadline of each instance */ | |
522 | u64 dl_period; /* Separation of two instances (period) */ | |
54d6d303 | 523 | u64 dl_bw; /* dl_runtime / dl_period */ |
3effcb42 | 524 | u64 dl_density; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
525 | |
526 | /* | |
527 | * Actual scheduling parameters. Initialized with the values above, | |
dfcb245e | 528 | * they are continuously updated during task execution. Note that |
aab03e05 DF |
529 | * the remaining runtime could be < 0 in case we are in overrun. |
530 | */ | |
5eca1c10 IM |
531 | s64 runtime; /* Remaining runtime for this instance */ |
532 | u64 deadline; /* Absolute deadline for this instance */ | |
533 | unsigned int flags; /* Specifying the scheduler behaviour */ | |
aab03e05 DF |
534 | |
535 | /* | |
536 | * Some bool flags: | |
537 | * | |
538 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
539 | * task has to wait for a replenishment to be performed at the | |
540 | * next firing of dl_timer. | |
541 | * | |
2d3d891d DF |
542 | * @dl_boosted tells if we are boosted due to DI. If so we are |
543 | * outside bandwidth enforcement mechanism (but only until we | |
5bfd126e JL |
544 | * exit the critical section); |
545 | * | |
5eca1c10 | 546 | * @dl_yielded tells if task gave up the CPU before consuming |
5bfd126e | 547 | * all its available runtime during the last job. |
209a0cbd LA |
548 | * |
549 | * @dl_non_contending tells if the task is inactive while still | |
550 | * contributing to the active utilization. In other words, it | |
551 | * indicates if the inactive timer has been armed and its handler | |
552 | * has not been executed yet. This flag is useful to avoid race | |
553 | * conditions between the inactive timer handler and the wakeup | |
554 | * code. | |
34be3930 JL |
555 | * |
556 | * @dl_overrun tells if the task asked to be informed about runtime | |
557 | * overruns. | |
aab03e05 | 558 | */ |
aa5222e9 DC |
559 | unsigned int dl_throttled : 1; |
560 | unsigned int dl_boosted : 1; | |
561 | unsigned int dl_yielded : 1; | |
562 | unsigned int dl_non_contending : 1; | |
34be3930 | 563 | unsigned int dl_overrun : 1; |
aab03e05 DF |
564 | |
565 | /* | |
566 | * Bandwidth enforcement timer. Each -deadline task has its | |
567 | * own bandwidth to be enforced, thus we need one timer per task. | |
568 | */ | |
5eca1c10 | 569 | struct hrtimer dl_timer; |
209a0cbd LA |
570 | |
571 | /* | |
572 | * Inactive timer, responsible for decreasing the active utilization | |
573 | * at the "0-lag time". When a -deadline task blocks, it contributes | |
574 | * to GRUB's active utilization until the "0-lag time", hence a | |
575 | * timer is needed to decrease the active utilization at the correct | |
576 | * time. | |
577 | */ | |
578 | struct hrtimer inactive_timer; | |
aab03e05 | 579 | }; |
8bd75c77 | 580 | |
69842cba PB |
581 | #ifdef CONFIG_UCLAMP_TASK |
582 | /* Number of utilization clamp buckets (shorter alias) */ | |
583 | #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT | |
584 | ||
585 | /* | |
586 | * Utilization clamp for a scheduling entity | |
587 | * @value: clamp value "assigned" to a se | |
588 | * @bucket_id: bucket index corresponding to the "assigned" value | |
e8f14172 | 589 | * @active: the se is currently refcounted in a rq's bucket |
a509a7cd | 590 | * @user_defined: the requested clamp value comes from user-space |
69842cba PB |
591 | * |
592 | * The bucket_id is the index of the clamp bucket matching the clamp value | |
593 | * which is pre-computed and stored to avoid expensive integer divisions from | |
594 | * the fast path. | |
e8f14172 PB |
595 | * |
596 | * The active bit is set whenever a task has got an "effective" value assigned, | |
597 | * which can be different from the clamp value "requested" from user-space. | |
598 | * This allows to know a task is refcounted in the rq's bucket corresponding | |
599 | * to the "effective" bucket_id. | |
a509a7cd PB |
600 | * |
601 | * The user_defined bit is set whenever a task has got a task-specific clamp | |
602 | * value requested from userspace, i.e. the system defaults apply to this task | |
603 | * just as a restriction. This allows to relax default clamps when a less | |
604 | * restrictive task-specific value has been requested, thus allowing to | |
605 | * implement a "nice" semantic. For example, a task running with a 20% | |
606 | * default boost can still drop its own boosting to 0%. | |
69842cba PB |
607 | */ |
608 | struct uclamp_se { | |
609 | unsigned int value : bits_per(SCHED_CAPACITY_SCALE); | |
610 | unsigned int bucket_id : bits_per(UCLAMP_BUCKETS); | |
e8f14172 | 611 | unsigned int active : 1; |
a509a7cd | 612 | unsigned int user_defined : 1; |
69842cba PB |
613 | }; |
614 | #endif /* CONFIG_UCLAMP_TASK */ | |
615 | ||
1d082fd0 PM |
616 | union rcu_special { |
617 | struct { | |
5eca1c10 IM |
618 | u8 blocked; |
619 | u8 need_qs; | |
05f41571 | 620 | u8 exp_hint; /* Hint for performance. */ |
23634ebc | 621 | u8 deferred_qs; |
8203d6d0 | 622 | } b; /* Bits. */ |
05f41571 | 623 | u32 s; /* Set of bits. */ |
1d082fd0 | 624 | }; |
86848966 | 625 | |
8dc85d54 PZ |
626 | enum perf_event_task_context { |
627 | perf_invalid_context = -1, | |
628 | perf_hw_context = 0, | |
89a1e187 | 629 | perf_sw_context, |
8dc85d54 PZ |
630 | perf_nr_task_contexts, |
631 | }; | |
632 | ||
eb61baf6 IM |
633 | struct wake_q_node { |
634 | struct wake_q_node *next; | |
635 | }; | |
636 | ||
1da177e4 | 637 | struct task_struct { |
c65eacbe AL |
638 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
639 | /* | |
640 | * For reasons of header soup (see current_thread_info()), this | |
641 | * must be the first element of task_struct. | |
642 | */ | |
5eca1c10 | 643 | struct thread_info thread_info; |
c65eacbe | 644 | #endif |
5eca1c10 IM |
645 | /* -1 unrunnable, 0 runnable, >0 stopped: */ |
646 | volatile long state; | |
29e48ce8 KC |
647 | |
648 | /* | |
649 | * This begins the randomizable portion of task_struct. Only | |
650 | * scheduling-critical items should be added above here. | |
651 | */ | |
652 | randomized_struct_fields_start | |
653 | ||
5eca1c10 | 654 | void *stack; |
ec1d2819 | 655 | refcount_t usage; |
5eca1c10 IM |
656 | /* Per task flags (PF_*), defined further below: */ |
657 | unsigned int flags; | |
658 | unsigned int ptrace; | |
1da177e4 | 659 | |
2dd73a4f | 660 | #ifdef CONFIG_SMP |
5eca1c10 IM |
661 | struct llist_node wake_entry; |
662 | int on_cpu; | |
c65eacbe | 663 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 IM |
664 | /* Current CPU: */ |
665 | unsigned int cpu; | |
c65eacbe | 666 | #endif |
5eca1c10 IM |
667 | unsigned int wakee_flips; |
668 | unsigned long wakee_flip_decay_ts; | |
669 | struct task_struct *last_wakee; | |
ac66f547 | 670 | |
32e839dd MG |
671 | /* |
672 | * recent_used_cpu is initially set as the last CPU used by a task | |
673 | * that wakes affine another task. Waker/wakee relationships can | |
674 | * push tasks around a CPU where each wakeup moves to the next one. | |
675 | * Tracking a recently used CPU allows a quick search for a recently | |
676 | * used CPU that may be idle. | |
677 | */ | |
678 | int recent_used_cpu; | |
5eca1c10 | 679 | int wake_cpu; |
2dd73a4f | 680 | #endif |
5eca1c10 IM |
681 | int on_rq; |
682 | ||
683 | int prio; | |
684 | int static_prio; | |
685 | int normal_prio; | |
686 | unsigned int rt_priority; | |
50e645a8 | 687 | |
5eca1c10 IM |
688 | const struct sched_class *sched_class; |
689 | struct sched_entity se; | |
690 | struct sched_rt_entity rt; | |
8323f26c | 691 | #ifdef CONFIG_CGROUP_SCHED |
5eca1c10 | 692 | struct task_group *sched_task_group; |
8323f26c | 693 | #endif |
5eca1c10 | 694 | struct sched_dl_entity dl; |
1da177e4 | 695 | |
69842cba | 696 | #ifdef CONFIG_UCLAMP_TASK |
e8f14172 PB |
697 | /* Clamp values requested for a scheduling entity */ |
698 | struct uclamp_se uclamp_req[UCLAMP_CNT]; | |
699 | /* Effective clamp values used for a scheduling entity */ | |
69842cba PB |
700 | struct uclamp_se uclamp[UCLAMP_CNT]; |
701 | #endif | |
702 | ||
e107be36 | 703 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
5eca1c10 IM |
704 | /* List of struct preempt_notifier: */ |
705 | struct hlist_head preempt_notifiers; | |
e107be36 AK |
706 | #endif |
707 | ||
6c5c9341 | 708 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
5eca1c10 | 709 | unsigned int btrace_seq; |
6c5c9341 | 710 | #endif |
1da177e4 | 711 | |
5eca1c10 IM |
712 | unsigned int policy; |
713 | int nr_cpus_allowed; | |
3bd37062 SAS |
714 | const cpumask_t *cpus_ptr; |
715 | cpumask_t cpus_mask; | |
1da177e4 | 716 | |
a57eb940 | 717 | #ifdef CONFIG_PREEMPT_RCU |
5eca1c10 IM |
718 | int rcu_read_lock_nesting; |
719 | union rcu_special rcu_read_unlock_special; | |
720 | struct list_head rcu_node_entry; | |
721 | struct rcu_node *rcu_blocked_node; | |
28f6569a | 722 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
5eca1c10 | 723 | |
8315f422 | 724 | #ifdef CONFIG_TASKS_RCU |
5eca1c10 | 725 | unsigned long rcu_tasks_nvcsw; |
ccdd29ff PM |
726 | u8 rcu_tasks_holdout; |
727 | u8 rcu_tasks_idx; | |
5eca1c10 | 728 | int rcu_tasks_idle_cpu; |
ccdd29ff | 729 | struct list_head rcu_tasks_holdout_list; |
8315f422 | 730 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 731 | |
5eca1c10 | 732 | struct sched_info sched_info; |
1da177e4 | 733 | |
5eca1c10 | 734 | struct list_head tasks; |
806c09a7 | 735 | #ifdef CONFIG_SMP |
5eca1c10 IM |
736 | struct plist_node pushable_tasks; |
737 | struct rb_node pushable_dl_tasks; | |
806c09a7 | 738 | #endif |
1da177e4 | 739 | |
5eca1c10 IM |
740 | struct mm_struct *mm; |
741 | struct mm_struct *active_mm; | |
314ff785 IM |
742 | |
743 | /* Per-thread vma caching: */ | |
5eca1c10 | 744 | struct vmacache vmacache; |
314ff785 | 745 | |
5eca1c10 IM |
746 | #ifdef SPLIT_RSS_COUNTING |
747 | struct task_rss_stat rss_stat; | |
34e55232 | 748 | #endif |
5eca1c10 IM |
749 | int exit_state; |
750 | int exit_code; | |
751 | int exit_signal; | |
752 | /* The signal sent when the parent dies: */ | |
753 | int pdeath_signal; | |
754 | /* JOBCTL_*, siglock protected: */ | |
755 | unsigned long jobctl; | |
756 | ||
757 | /* Used for emulating ABI behavior of previous Linux versions: */ | |
758 | unsigned int personality; | |
759 | ||
760 | /* Scheduler bits, serialized by scheduler locks: */ | |
761 | unsigned sched_reset_on_fork:1; | |
762 | unsigned sched_contributes_to_load:1; | |
763 | unsigned sched_migrated:1; | |
764 | unsigned sched_remote_wakeup:1; | |
eb414681 JW |
765 | #ifdef CONFIG_PSI |
766 | unsigned sched_psi_wake_requeue:1; | |
767 | #endif | |
768 | ||
5eca1c10 IM |
769 | /* Force alignment to the next boundary: */ |
770 | unsigned :0; | |
771 | ||
772 | /* Unserialized, strictly 'current' */ | |
773 | ||
774 | /* Bit to tell LSMs we're in execve(): */ | |
775 | unsigned in_execve:1; | |
776 | unsigned in_iowait:1; | |
777 | #ifndef TIF_RESTORE_SIGMASK | |
778 | unsigned restore_sigmask:1; | |
7e781418 | 779 | #endif |
626ebc41 | 780 | #ifdef CONFIG_MEMCG |
29ef680a | 781 | unsigned in_user_fault:1; |
127424c8 | 782 | #endif |
ff303e66 | 783 | #ifdef CONFIG_COMPAT_BRK |
5eca1c10 | 784 | unsigned brk_randomized:1; |
ff303e66 | 785 | #endif |
77f88796 TH |
786 | #ifdef CONFIG_CGROUPS |
787 | /* disallow userland-initiated cgroup migration */ | |
788 | unsigned no_cgroup_migration:1; | |
76f969e8 RG |
789 | /* task is frozen/stopped (used by the cgroup freezer) */ |
790 | unsigned frozen:1; | |
77f88796 | 791 | #endif |
d09d8df3 JB |
792 | #ifdef CONFIG_BLK_CGROUP |
793 | /* to be used once the psi infrastructure lands upstream. */ | |
794 | unsigned use_memdelay:1; | |
795 | #endif | |
6f185c29 | 796 | |
5eca1c10 | 797 | unsigned long atomic_flags; /* Flags requiring atomic access. */ |
1d4457f9 | 798 | |
5eca1c10 | 799 | struct restart_block restart_block; |
f56141e3 | 800 | |
5eca1c10 IM |
801 | pid_t pid; |
802 | pid_t tgid; | |
0a425405 | 803 | |
050e9baa | 804 | #ifdef CONFIG_STACKPROTECTOR |
5eca1c10 IM |
805 | /* Canary value for the -fstack-protector GCC feature: */ |
806 | unsigned long stack_canary; | |
1314562a | 807 | #endif |
4d1d61a6 | 808 | /* |
5eca1c10 | 809 | * Pointers to the (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 810 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 811 | * p->real_parent->pid) |
1da177e4 | 812 | */ |
5eca1c10 IM |
813 | |
814 | /* Real parent process: */ | |
815 | struct task_struct __rcu *real_parent; | |
816 | ||
817 | /* Recipient of SIGCHLD, wait4() reports: */ | |
818 | struct task_struct __rcu *parent; | |
819 | ||
1da177e4 | 820 | /* |
5eca1c10 | 821 | * Children/sibling form the list of natural children: |
1da177e4 | 822 | */ |
5eca1c10 IM |
823 | struct list_head children; |
824 | struct list_head sibling; | |
825 | struct task_struct *group_leader; | |
1da177e4 | 826 | |
f470021a | 827 | /* |
5eca1c10 IM |
828 | * 'ptraced' is the list of tasks this task is using ptrace() on. |
829 | * | |
f470021a | 830 | * This includes both natural children and PTRACE_ATTACH targets. |
5eca1c10 | 831 | * 'ptrace_entry' is this task's link on the p->parent->ptraced list. |
f470021a | 832 | */ |
5eca1c10 IM |
833 | struct list_head ptraced; |
834 | struct list_head ptrace_entry; | |
f470021a | 835 | |
1da177e4 | 836 | /* PID/PID hash table linkage. */ |
2c470475 EB |
837 | struct pid *thread_pid; |
838 | struct hlist_node pid_links[PIDTYPE_MAX]; | |
5eca1c10 IM |
839 | struct list_head thread_group; |
840 | struct list_head thread_node; | |
841 | ||
842 | struct completion *vfork_done; | |
1da177e4 | 843 | |
5eca1c10 IM |
844 | /* CLONE_CHILD_SETTID: */ |
845 | int __user *set_child_tid; | |
1da177e4 | 846 | |
5eca1c10 IM |
847 | /* CLONE_CHILD_CLEARTID: */ |
848 | int __user *clear_child_tid; | |
849 | ||
850 | u64 utime; | |
851 | u64 stime; | |
40565b5a | 852 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME |
5eca1c10 IM |
853 | u64 utimescaled; |
854 | u64 stimescaled; | |
40565b5a | 855 | #endif |
5eca1c10 IM |
856 | u64 gtime; |
857 | struct prev_cputime prev_cputime; | |
6a61671b | 858 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
bac5b6b6 | 859 | struct vtime vtime; |
d99ca3b9 | 860 | #endif |
d027d45d FW |
861 | |
862 | #ifdef CONFIG_NO_HZ_FULL | |
5eca1c10 | 863 | atomic_t tick_dep_mask; |
d027d45d | 864 | #endif |
5eca1c10 IM |
865 | /* Context switch counts: */ |
866 | unsigned long nvcsw; | |
867 | unsigned long nivcsw; | |
868 | ||
869 | /* Monotonic time in nsecs: */ | |
870 | u64 start_time; | |
871 | ||
872 | /* Boot based time in nsecs: */ | |
873 | u64 real_start_time; | |
874 | ||
875 | /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */ | |
876 | unsigned long min_flt; | |
877 | unsigned long maj_flt; | |
1da177e4 | 878 | |
b18b6a9c | 879 | #ifdef CONFIG_POSIX_TIMERS |
5eca1c10 IM |
880 | struct task_cputime cputime_expires; |
881 | struct list_head cpu_timers[3]; | |
b18b6a9c | 882 | #endif |
1da177e4 | 883 | |
5eca1c10 IM |
884 | /* Process credentials: */ |
885 | ||
886 | /* Tracer's credentials at attach: */ | |
887 | const struct cred __rcu *ptracer_cred; | |
888 | ||
889 | /* Objective and real subjective task credentials (COW): */ | |
890 | const struct cred __rcu *real_cred; | |
891 | ||
892 | /* Effective (overridable) subjective task credentials (COW): */ | |
893 | const struct cred __rcu *cred; | |
894 | ||
7743c48e DH |
895 | #ifdef CONFIG_KEYS |
896 | /* Cached requested key. */ | |
897 | struct key *cached_requested_key; | |
898 | #endif | |
899 | ||
5eca1c10 IM |
900 | /* |
901 | * executable name, excluding path. | |
902 | * | |
903 | * - normally initialized setup_new_exec() | |
904 | * - access it with [gs]et_task_comm() | |
905 | * - lock it with task_lock() | |
906 | */ | |
907 | char comm[TASK_COMM_LEN]; | |
908 | ||
909 | struct nameidata *nameidata; | |
910 | ||
3d5b6fcc | 911 | #ifdef CONFIG_SYSVIPC |
5eca1c10 IM |
912 | struct sysv_sem sysvsem; |
913 | struct sysv_shm sysvshm; | |
3d5b6fcc | 914 | #endif |
e162b39a | 915 | #ifdef CONFIG_DETECT_HUNG_TASK |
5eca1c10 | 916 | unsigned long last_switch_count; |
a2e51445 | 917 | unsigned long last_switch_time; |
82a1fcb9 | 918 | #endif |
5eca1c10 IM |
919 | /* Filesystem information: */ |
920 | struct fs_struct *fs; | |
921 | ||
922 | /* Open file information: */ | |
923 | struct files_struct *files; | |
924 | ||
925 | /* Namespaces: */ | |
926 | struct nsproxy *nsproxy; | |
927 | ||
928 | /* Signal handlers: */ | |
929 | struct signal_struct *signal; | |
930 | struct sighand_struct *sighand; | |
931 | sigset_t blocked; | |
932 | sigset_t real_blocked; | |
933 | /* Restored if set_restore_sigmask() was used: */ | |
934 | sigset_t saved_sigmask; | |
935 | struct sigpending pending; | |
936 | unsigned long sas_ss_sp; | |
937 | size_t sas_ss_size; | |
938 | unsigned int sas_ss_flags; | |
939 | ||
940 | struct callback_head *task_works; | |
941 | ||
4b7d248b | 942 | #ifdef CONFIG_AUDIT |
bfef93a5 | 943 | #ifdef CONFIG_AUDITSYSCALL |
5f3d544f RGB |
944 | struct audit_context *audit_context; |
945 | #endif | |
5eca1c10 IM |
946 | kuid_t loginuid; |
947 | unsigned int sessionid; | |
bfef93a5 | 948 | #endif |
5eca1c10 IM |
949 | struct seccomp seccomp; |
950 | ||
951 | /* Thread group tracking: */ | |
952 | u32 parent_exec_id; | |
953 | u32 self_exec_id; | |
1da177e4 | 954 | |
5eca1c10 IM |
955 | /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */ |
956 | spinlock_t alloc_lock; | |
1da177e4 | 957 | |
b29739f9 | 958 | /* Protection of the PI data structures: */ |
5eca1c10 | 959 | raw_spinlock_t pi_lock; |
b29739f9 | 960 | |
5eca1c10 | 961 | struct wake_q_node wake_q; |
76751049 | 962 | |
23f78d4a | 963 | #ifdef CONFIG_RT_MUTEXES |
5eca1c10 | 964 | /* PI waiters blocked on a rt_mutex held by this task: */ |
a23ba907 | 965 | struct rb_root_cached pi_waiters; |
e96a7705 XP |
966 | /* Updated under owner's pi_lock and rq lock */ |
967 | struct task_struct *pi_top_task; | |
5eca1c10 IM |
968 | /* Deadlock detection and priority inheritance handling: */ |
969 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
970 | #endif |
971 | ||
408894ee | 972 | #ifdef CONFIG_DEBUG_MUTEXES |
5eca1c10 IM |
973 | /* Mutex deadlock detection: */ |
974 | struct mutex_waiter *blocked_on; | |
408894ee | 975 | #endif |
5eca1c10 | 976 | |
312364f3 DV |
977 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
978 | int non_block_count; | |
979 | #endif | |
980 | ||
de30a2b3 | 981 | #ifdef CONFIG_TRACE_IRQFLAGS |
5eca1c10 IM |
982 | unsigned int irq_events; |
983 | unsigned long hardirq_enable_ip; | |
984 | unsigned long hardirq_disable_ip; | |
985 | unsigned int hardirq_enable_event; | |
986 | unsigned int hardirq_disable_event; | |
987 | int hardirqs_enabled; | |
988 | int hardirq_context; | |
989 | unsigned long softirq_disable_ip; | |
990 | unsigned long softirq_enable_ip; | |
991 | unsigned int softirq_disable_event; | |
992 | unsigned int softirq_enable_event; | |
993 | int softirqs_enabled; | |
994 | int softirq_context; | |
de30a2b3 | 995 | #endif |
5eca1c10 | 996 | |
fbb9ce95 | 997 | #ifdef CONFIG_LOCKDEP |
5eca1c10 IM |
998 | # define MAX_LOCK_DEPTH 48UL |
999 | u64 curr_chain_key; | |
1000 | int lockdep_depth; | |
1001 | unsigned int lockdep_recursion; | |
1002 | struct held_lock held_locks[MAX_LOCK_DEPTH]; | |
fbb9ce95 | 1003 | #endif |
5eca1c10 | 1004 | |
c6d30853 | 1005 | #ifdef CONFIG_UBSAN |
5eca1c10 | 1006 | unsigned int in_ubsan; |
c6d30853 | 1007 | #endif |
408894ee | 1008 | |
5eca1c10 IM |
1009 | /* Journalling filesystem info: */ |
1010 | void *journal_info; | |
1da177e4 | 1011 | |
5eca1c10 IM |
1012 | /* Stacked block device info: */ |
1013 | struct bio_list *bio_list; | |
d89d8796 | 1014 | |
73c10101 | 1015 | #ifdef CONFIG_BLOCK |
5eca1c10 IM |
1016 | /* Stack plugging: */ |
1017 | struct blk_plug *plug; | |
73c10101 JA |
1018 | #endif |
1019 | ||
5eca1c10 IM |
1020 | /* VM state: */ |
1021 | struct reclaim_state *reclaim_state; | |
1022 | ||
1023 | struct backing_dev_info *backing_dev_info; | |
1da177e4 | 1024 | |
5eca1c10 | 1025 | struct io_context *io_context; |
1da177e4 | 1026 | |
5e1f0f09 MG |
1027 | #ifdef CONFIG_COMPACTION |
1028 | struct capture_control *capture_control; | |
1029 | #endif | |
5eca1c10 IM |
1030 | /* Ptrace state: */ |
1031 | unsigned long ptrace_message; | |
ae7795bc | 1032 | kernel_siginfo_t *last_siginfo; |
1da177e4 | 1033 | |
5eca1c10 | 1034 | struct task_io_accounting ioac; |
eb414681 JW |
1035 | #ifdef CONFIG_PSI |
1036 | /* Pressure stall state */ | |
1037 | unsigned int psi_flags; | |
1038 | #endif | |
5eca1c10 IM |
1039 | #ifdef CONFIG_TASK_XACCT |
1040 | /* Accumulated RSS usage: */ | |
1041 | u64 acct_rss_mem1; | |
1042 | /* Accumulated virtual memory usage: */ | |
1043 | u64 acct_vm_mem1; | |
1044 | /* stime + utime since last update: */ | |
1045 | u64 acct_timexpd; | |
1da177e4 LT |
1046 | #endif |
1047 | #ifdef CONFIG_CPUSETS | |
5eca1c10 IM |
1048 | /* Protected by ->alloc_lock: */ |
1049 | nodemask_t mems_allowed; | |
1050 | /* Seqence number to catch updates: */ | |
1051 | seqcount_t mems_allowed_seq; | |
1052 | int cpuset_mem_spread_rotor; | |
1053 | int cpuset_slab_spread_rotor; | |
1da177e4 | 1054 | #endif |
ddbcc7e8 | 1055 | #ifdef CONFIG_CGROUPS |
5eca1c10 IM |
1056 | /* Control Group info protected by css_set_lock: */ |
1057 | struct css_set __rcu *cgroups; | |
1058 | /* cg_list protected by css_set_lock and tsk->alloc_lock: */ | |
1059 | struct list_head cg_list; | |
ddbcc7e8 | 1060 | #endif |
e6d42931 | 1061 | #ifdef CONFIG_X86_CPU_RESCTRL |
0734ded1 | 1062 | u32 closid; |
d6aaba61 | 1063 | u32 rmid; |
e02737d5 | 1064 | #endif |
42b2dd0a | 1065 | #ifdef CONFIG_FUTEX |
5eca1c10 | 1066 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
1067 | #ifdef CONFIG_COMPAT |
1068 | struct compat_robust_list_head __user *compat_robust_list; | |
1069 | #endif | |
5eca1c10 IM |
1070 | struct list_head pi_state_list; |
1071 | struct futex_pi_state *pi_state_cache; | |
c7aceaba | 1072 | #endif |
cdd6c482 | 1073 | #ifdef CONFIG_PERF_EVENTS |
5eca1c10 IM |
1074 | struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; |
1075 | struct mutex perf_event_mutex; | |
1076 | struct list_head perf_event_list; | |
a63eaf34 | 1077 | #endif |
8f47b187 | 1078 | #ifdef CONFIG_DEBUG_PREEMPT |
5eca1c10 | 1079 | unsigned long preempt_disable_ip; |
8f47b187 | 1080 | #endif |
c7aceaba | 1081 | #ifdef CONFIG_NUMA |
5eca1c10 IM |
1082 | /* Protected by alloc_lock: */ |
1083 | struct mempolicy *mempolicy; | |
45816682 | 1084 | short il_prev; |
5eca1c10 | 1085 | short pref_node_fork; |
42b2dd0a | 1086 | #endif |
cbee9f88 | 1087 | #ifdef CONFIG_NUMA_BALANCING |
5eca1c10 IM |
1088 | int numa_scan_seq; |
1089 | unsigned int numa_scan_period; | |
1090 | unsigned int numa_scan_period_max; | |
1091 | int numa_preferred_nid; | |
1092 | unsigned long numa_migrate_retry; | |
1093 | /* Migration stamp: */ | |
1094 | u64 node_stamp; | |
1095 | u64 last_task_numa_placement; | |
1096 | u64 last_sum_exec_runtime; | |
1097 | struct callback_head numa_work; | |
1098 | ||
cb361d8c JH |
1099 | /* |
1100 | * This pointer is only modified for current in syscall and | |
1101 | * pagefault context (and for tasks being destroyed), so it can be read | |
1102 | * from any of the following contexts: | |
1103 | * - RCU read-side critical section | |
1104 | * - current->numa_group from everywhere | |
1105 | * - task's runqueue locked, task not running | |
1106 | */ | |
1107 | struct numa_group __rcu *numa_group; | |
8c8a743c | 1108 | |
745d6147 | 1109 | /* |
44dba3d5 IM |
1110 | * numa_faults is an array split into four regions: |
1111 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
1112 | * in this precise order. | |
1113 | * | |
1114 | * faults_memory: Exponential decaying average of faults on a per-node | |
1115 | * basis. Scheduling placement decisions are made based on these | |
1116 | * counts. The values remain static for the duration of a PTE scan. | |
1117 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
1118 | * hinting fault was incurred. | |
1119 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
1120 | * during the current scan window. When the scan completes, the counts | |
1121 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 1122 | */ |
5eca1c10 IM |
1123 | unsigned long *numa_faults; |
1124 | unsigned long total_numa_faults; | |
745d6147 | 1125 | |
04bb2f94 RR |
1126 | /* |
1127 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
1128 | * scan window were remote/local or failed to migrate. The task scan |
1129 | * period is adapted based on the locality of the faults with different | |
1130 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 1131 | */ |
5eca1c10 | 1132 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 1133 | |
5eca1c10 | 1134 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
1135 | #endif /* CONFIG_NUMA_BALANCING */ |
1136 | ||
d7822b1e MD |
1137 | #ifdef CONFIG_RSEQ |
1138 | struct rseq __user *rseq; | |
d7822b1e MD |
1139 | u32 rseq_sig; |
1140 | /* | |
1141 | * RmW on rseq_event_mask must be performed atomically | |
1142 | * with respect to preemption. | |
1143 | */ | |
1144 | unsigned long rseq_event_mask; | |
1145 | #endif | |
1146 | ||
5eca1c10 | 1147 | struct tlbflush_unmap_batch tlb_ubc; |
72b252ae | 1148 | |
5eca1c10 | 1149 | struct rcu_head rcu; |
b92ce558 | 1150 | |
5eca1c10 IM |
1151 | /* Cache last used pipe for splice(): */ |
1152 | struct pipe_inode_info *splice_pipe; | |
5640f768 | 1153 | |
5eca1c10 | 1154 | struct page_frag task_frag; |
5640f768 | 1155 | |
47913d4e IM |
1156 | #ifdef CONFIG_TASK_DELAY_ACCT |
1157 | struct task_delay_info *delays; | |
f4f154fd | 1158 | #endif |
47913d4e | 1159 | |
f4f154fd | 1160 | #ifdef CONFIG_FAULT_INJECTION |
5eca1c10 | 1161 | int make_it_fail; |
9049f2f6 | 1162 | unsigned int fail_nth; |
ca74e92b | 1163 | #endif |
9d823e8f | 1164 | /* |
5eca1c10 IM |
1165 | * When (nr_dirtied >= nr_dirtied_pause), it's time to call |
1166 | * balance_dirty_pages() for a dirty throttling pause: | |
9d823e8f | 1167 | */ |
5eca1c10 IM |
1168 | int nr_dirtied; |
1169 | int nr_dirtied_pause; | |
1170 | /* Start of a write-and-pause period: */ | |
1171 | unsigned long dirty_paused_when; | |
9d823e8f | 1172 | |
9745512c | 1173 | #ifdef CONFIG_LATENCYTOP |
5eca1c10 IM |
1174 | int latency_record_count; |
1175 | struct latency_record latency_record[LT_SAVECOUNT]; | |
9745512c | 1176 | #endif |
6976675d | 1177 | /* |
5eca1c10 | 1178 | * Time slack values; these are used to round up poll() and |
6976675d AV |
1179 | * select() etc timeout values. These are in nanoseconds. |
1180 | */ | |
5eca1c10 IM |
1181 | u64 timer_slack_ns; |
1182 | u64 default_timer_slack_ns; | |
f8d570a4 | 1183 | |
0b24becc | 1184 | #ifdef CONFIG_KASAN |
5eca1c10 | 1185 | unsigned int kasan_depth; |
0b24becc | 1186 | #endif |
5eca1c10 | 1187 | |
fb52607a | 1188 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
5eca1c10 IM |
1189 | /* Index of current stored address in ret_stack: */ |
1190 | int curr_ret_stack; | |
39eb456d | 1191 | int curr_ret_depth; |
5eca1c10 IM |
1192 | |
1193 | /* Stack of return addresses for return function tracing: */ | |
1194 | struct ftrace_ret_stack *ret_stack; | |
1195 | ||
1196 | /* Timestamp for last schedule: */ | |
1197 | unsigned long long ftrace_timestamp; | |
1198 | ||
f201ae23 FW |
1199 | /* |
1200 | * Number of functions that haven't been traced | |
5eca1c10 | 1201 | * because of depth overrun: |
f201ae23 | 1202 | */ |
5eca1c10 IM |
1203 | atomic_t trace_overrun; |
1204 | ||
1205 | /* Pause tracing: */ | |
1206 | atomic_t tracing_graph_pause; | |
f201ae23 | 1207 | #endif |
5eca1c10 | 1208 | |
ea4e2bc4 | 1209 | #ifdef CONFIG_TRACING |
5eca1c10 IM |
1210 | /* State flags for use by tracers: */ |
1211 | unsigned long trace; | |
1212 | ||
1213 | /* Bitmask and counter of trace recursion: */ | |
1214 | unsigned long trace_recursion; | |
261842b7 | 1215 | #endif /* CONFIG_TRACING */ |
5eca1c10 | 1216 | |
5c9a8750 | 1217 | #ifdef CONFIG_KCOV |
5eca1c10 | 1218 | /* Coverage collection mode enabled for this task (0 if disabled): */ |
0ed557aa | 1219 | unsigned int kcov_mode; |
5eca1c10 IM |
1220 | |
1221 | /* Size of the kcov_area: */ | |
1222 | unsigned int kcov_size; | |
1223 | ||
1224 | /* Buffer for coverage collection: */ | |
1225 | void *kcov_area; | |
1226 | ||
1227 | /* KCOV descriptor wired with this task or NULL: */ | |
1228 | struct kcov *kcov; | |
5c9a8750 | 1229 | #endif |
5eca1c10 | 1230 | |
6f185c29 | 1231 | #ifdef CONFIG_MEMCG |
5eca1c10 IM |
1232 | struct mem_cgroup *memcg_in_oom; |
1233 | gfp_t memcg_oom_gfp_mask; | |
1234 | int memcg_oom_order; | |
b23afb93 | 1235 | |
5eca1c10 IM |
1236 | /* Number of pages to reclaim on returning to userland: */ |
1237 | unsigned int memcg_nr_pages_over_high; | |
d46eb14b SB |
1238 | |
1239 | /* Used by memcontrol for targeted memcg charge: */ | |
1240 | struct mem_cgroup *active_memcg; | |
569b846d | 1241 | #endif |
5eca1c10 | 1242 | |
d09d8df3 JB |
1243 | #ifdef CONFIG_BLK_CGROUP |
1244 | struct request_queue *throttle_queue; | |
1245 | #endif | |
1246 | ||
0326f5a9 | 1247 | #ifdef CONFIG_UPROBES |
5eca1c10 | 1248 | struct uprobe_task *utask; |
0326f5a9 | 1249 | #endif |
cafe5635 | 1250 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
5eca1c10 IM |
1251 | unsigned int sequential_io; |
1252 | unsigned int sequential_io_avg; | |
cafe5635 | 1253 | #endif |
8eb23b9f | 1254 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
5eca1c10 | 1255 | unsigned long task_state_change; |
8eb23b9f | 1256 | #endif |
5eca1c10 | 1257 | int pagefault_disabled; |
03049269 | 1258 | #ifdef CONFIG_MMU |
5eca1c10 | 1259 | struct task_struct *oom_reaper_list; |
03049269 | 1260 | #endif |
ba14a194 | 1261 | #ifdef CONFIG_VMAP_STACK |
5eca1c10 | 1262 | struct vm_struct *stack_vm_area; |
ba14a194 | 1263 | #endif |
68f24b08 | 1264 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 | 1265 | /* A live task holds one reference: */ |
f0b89d39 | 1266 | refcount_t stack_refcount; |
d83a7cb3 JP |
1267 | #endif |
1268 | #ifdef CONFIG_LIVEPATCH | |
1269 | int patch_state; | |
0302e28d | 1270 | #endif |
e4e55b47 TH |
1271 | #ifdef CONFIG_SECURITY |
1272 | /* Used by LSM modules for access restriction: */ | |
1273 | void *security; | |
68f24b08 | 1274 | #endif |
29e48ce8 | 1275 | |
afaef01c AP |
1276 | #ifdef CONFIG_GCC_PLUGIN_STACKLEAK |
1277 | unsigned long lowest_stack; | |
c8d12627 | 1278 | unsigned long prev_lowest_stack; |
afaef01c AP |
1279 | #endif |
1280 | ||
29e48ce8 KC |
1281 | /* |
1282 | * New fields for task_struct should be added above here, so that | |
1283 | * they are included in the randomized portion of task_struct. | |
1284 | */ | |
1285 | randomized_struct_fields_end | |
1286 | ||
5eca1c10 IM |
1287 | /* CPU-specific state of this task: */ |
1288 | struct thread_struct thread; | |
1289 | ||
1290 | /* | |
1291 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1292 | * structure. It *MUST* be at the end of 'task_struct'. | |
1293 | * | |
1294 | * Do not put anything below here! | |
1295 | */ | |
1da177e4 LT |
1296 | }; |
1297 | ||
e868171a | 1298 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 | 1299 | { |
2c470475 | 1300 | return task->thread_pid; |
22c935f4 EB |
1301 | } |
1302 | ||
7af57294 PE |
1303 | /* |
1304 | * the helpers to get the task's different pids as they are seen | |
1305 | * from various namespaces | |
1306 | * | |
1307 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
1308 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
1309 | * current. | |
7af57294 PE |
1310 | * task_xid_nr_ns() : id seen from the ns specified; |
1311 | * | |
7af57294 PE |
1312 | * see also pid_nr() etc in include/linux/pid.h |
1313 | */ | |
5eca1c10 | 1314 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns); |
7af57294 | 1315 | |
e868171a | 1316 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
1317 | { |
1318 | return tsk->pid; | |
1319 | } | |
1320 | ||
5eca1c10 | 1321 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
52ee2dfd ON |
1322 | { |
1323 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
1324 | } | |
7af57294 PE |
1325 | |
1326 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
1327 | { | |
52ee2dfd | 1328 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
1329 | } |
1330 | ||
1331 | ||
e868171a | 1332 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
1333 | { |
1334 | return tsk->tgid; | |
1335 | } | |
1336 | ||
5eca1c10 IM |
1337 | /** |
1338 | * pid_alive - check that a task structure is not stale | |
1339 | * @p: Task structure to be checked. | |
1340 | * | |
1341 | * Test if a process is not yet dead (at most zombie state) | |
1342 | * If pid_alive fails, then pointers within the task structure | |
1343 | * can be stale and must not be dereferenced. | |
1344 | * | |
1345 | * Return: 1 if the process is alive. 0 otherwise. | |
1346 | */ | |
1347 | static inline int pid_alive(const struct task_struct *p) | |
1348 | { | |
2c470475 | 1349 | return p->thread_pid != NULL; |
5eca1c10 | 1350 | } |
7af57294 | 1351 | |
5eca1c10 | 1352 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1353 | { |
52ee2dfd | 1354 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
1355 | } |
1356 | ||
7af57294 PE |
1357 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
1358 | { | |
52ee2dfd | 1359 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
1360 | } |
1361 | ||
1362 | ||
5eca1c10 | 1363 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1364 | { |
52ee2dfd | 1365 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
1366 | } |
1367 | ||
7af57294 PE |
1368 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
1369 | { | |
52ee2dfd | 1370 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
1371 | } |
1372 | ||
dd1c1f2f ON |
1373 | static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
1374 | { | |
6883f81a | 1375 | return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns); |
dd1c1f2f ON |
1376 | } |
1377 | ||
1378 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
1379 | { | |
6883f81a | 1380 | return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL); |
dd1c1f2f ON |
1381 | } |
1382 | ||
1383 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) | |
1384 | { | |
1385 | pid_t pid = 0; | |
1386 | ||
1387 | rcu_read_lock(); | |
1388 | if (pid_alive(tsk)) | |
1389 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
1390 | rcu_read_unlock(); | |
1391 | ||
1392 | return pid; | |
1393 | } | |
1394 | ||
1395 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
1396 | { | |
1397 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
1398 | } | |
1399 | ||
5eca1c10 | 1400 | /* Obsolete, do not use: */ |
1b0f7ffd ON |
1401 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) |
1402 | { | |
1403 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
1404 | } | |
7af57294 | 1405 | |
06eb6184 PZ |
1406 | #define TASK_REPORT_IDLE (TASK_REPORT + 1) |
1407 | #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1) | |
1408 | ||
1d48b080 | 1409 | static inline unsigned int task_state_index(struct task_struct *tsk) |
20435d84 | 1410 | { |
1593baab PZ |
1411 | unsigned int tsk_state = READ_ONCE(tsk->state); |
1412 | unsigned int state = (tsk_state | tsk->exit_state) & TASK_REPORT; | |
20435d84 | 1413 | |
06eb6184 PZ |
1414 | BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX); |
1415 | ||
06eb6184 PZ |
1416 | if (tsk_state == TASK_IDLE) |
1417 | state = TASK_REPORT_IDLE; | |
1418 | ||
1593baab PZ |
1419 | return fls(state); |
1420 | } | |
1421 | ||
1d48b080 | 1422 | static inline char task_index_to_char(unsigned int state) |
1593baab | 1423 | { |
8ef9925b | 1424 | static const char state_char[] = "RSDTtXZPI"; |
1593baab | 1425 | |
06eb6184 | 1426 | BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1); |
20435d84 | 1427 | |
1593baab PZ |
1428 | return state_char[state]; |
1429 | } | |
1430 | ||
1431 | static inline char task_state_to_char(struct task_struct *tsk) | |
1432 | { | |
1d48b080 | 1433 | return task_index_to_char(task_state_index(tsk)); |
20435d84 XX |
1434 | } |
1435 | ||
f400e198 | 1436 | /** |
570f5241 SS |
1437 | * is_global_init - check if a task structure is init. Since init |
1438 | * is free to have sub-threads we need to check tgid. | |
3260259f HK |
1439 | * @tsk: Task structure to be checked. |
1440 | * | |
1441 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
1442 | * |
1443 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 1444 | */ |
e868171a | 1445 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 1446 | { |
570f5241 | 1447 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 1448 | } |
b460cbc5 | 1449 | |
9ec52099 CLG |
1450 | extern struct pid *cad_pid; |
1451 | ||
1da177e4 LT |
1452 | /* |
1453 | * Per process flags | |
1454 | */ | |
5eca1c10 IM |
1455 | #define PF_IDLE 0x00000002 /* I am an IDLE thread */ |
1456 | #define PF_EXITING 0x00000004 /* Getting shut down */ | |
1457 | #define PF_EXITPIDONE 0x00000008 /* PI exit done on shut down */ | |
1458 | #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ | |
1459 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ | |
1460 | #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */ | |
1461 | #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */ | |
1462 | #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */ | |
1463 | #define PF_DUMPCORE 0x00000200 /* Dumped core */ | |
1464 | #define PF_SIGNALED 0x00000400 /* Killed by a signal */ | |
1465 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
1466 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */ | |
1467 | #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */ | |
1468 | #define PF_USED_ASYNC 0x00004000 /* Used async_schedule*(), used by module init */ | |
1469 | #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */ | |
1470 | #define PF_FROZEN 0x00010000 /* Frozen for system suspend */ | |
7dea19f9 MH |
1471 | #define PF_KSWAPD 0x00020000 /* I am kswapd */ |
1472 | #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */ | |
1473 | #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */ | |
5eca1c10 IM |
1474 | #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ |
1475 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ | |
1476 | #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */ | |
1477 | #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ | |
eb414681 | 1478 | #define PF_MEMSTALL 0x01000000 /* Stalled due to lack of memory */ |
73ab1cb2 | 1479 | #define PF_UMH 0x02000000 /* I'm an Usermodehelper process */ |
3bd37062 | 1480 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */ |
5eca1c10 | 1481 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ |
d7fefcc8 | 1482 | #define PF_MEMALLOC_NOCMA 0x10000000 /* All allocation request will have _GFP_MOVABLE cleared */ |
5eca1c10 IM |
1483 | #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ |
1484 | #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */ | |
1da177e4 LT |
1485 | |
1486 | /* | |
1487 | * Only the _current_ task can read/write to tsk->flags, but other | |
1488 | * tasks can access tsk->flags in readonly mode for example | |
1489 | * with tsk_used_math (like during threaded core dumping). | |
1490 | * There is however an exception to this rule during ptrace | |
1491 | * or during fork: the ptracer task is allowed to write to the | |
1492 | * child->flags of its traced child (same goes for fork, the parent | |
1493 | * can write to the child->flags), because we're guaranteed the | |
1494 | * child is not running and in turn not changing child->flags | |
1495 | * at the same time the parent does it. | |
1496 | */ | |
5eca1c10 IM |
1497 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) |
1498 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
1499 | #define clear_used_math() clear_stopped_child_used_math(current) | |
1500 | #define set_used_math() set_stopped_child_used_math(current) | |
1501 | ||
1da177e4 LT |
1502 | #define conditional_stopped_child_used_math(condition, child) \ |
1503 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
5eca1c10 IM |
1504 | |
1505 | #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current) | |
1506 | ||
1da177e4 LT |
1507 | #define copy_to_stopped_child_used_math(child) \ |
1508 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
5eca1c10 | 1509 | |
1da177e4 | 1510 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ |
5eca1c10 IM |
1511 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) |
1512 | #define used_math() tsk_used_math(current) | |
1da177e4 | 1513 | |
62ec05dd TG |
1514 | static inline bool is_percpu_thread(void) |
1515 | { | |
1516 | #ifdef CONFIG_SMP | |
1517 | return (current->flags & PF_NO_SETAFFINITY) && | |
1518 | (current->nr_cpus_allowed == 1); | |
1519 | #else | |
1520 | return true; | |
1521 | #endif | |
1522 | } | |
1523 | ||
1d4457f9 | 1524 | /* Per-process atomic flags. */ |
5eca1c10 IM |
1525 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
1526 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ | |
1527 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
356e4bff TG |
1528 | #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */ |
1529 | #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/ | |
9137bb27 TG |
1530 | #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */ |
1531 | #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */ | |
71368af9 | 1532 | #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */ |
1d4457f9 | 1533 | |
e0e5070b ZL |
1534 | #define TASK_PFA_TEST(name, func) \ |
1535 | static inline bool task_##func(struct task_struct *p) \ | |
1536 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1537 | |
e0e5070b ZL |
1538 | #define TASK_PFA_SET(name, func) \ |
1539 | static inline void task_set_##func(struct task_struct *p) \ | |
1540 | { set_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1541 | |
e0e5070b ZL |
1542 | #define TASK_PFA_CLEAR(name, func) \ |
1543 | static inline void task_clear_##func(struct task_struct *p) \ | |
1544 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
1545 | ||
1546 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
1547 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 1548 | |
2ad654bc ZL |
1549 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
1550 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
1551 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
1552 | ||
1553 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
1554 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
1555 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 1556 | |
356e4bff TG |
1557 | TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable) |
1558 | TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1559 | TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1560 | ||
71368af9 WL |
1561 | TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec) |
1562 | TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec) | |
1563 | TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec) | |
1564 | ||
356e4bff TG |
1565 | TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) |
1566 | TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) | |
1567 | ||
9137bb27 TG |
1568 | TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable) |
1569 | TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable) | |
1570 | TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable) | |
1571 | ||
1572 | TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) | |
1573 | TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) | |
1574 | ||
5eca1c10 | 1575 | static inline void |
717a94b5 | 1576 | current_restore_flags(unsigned long orig_flags, unsigned long flags) |
907aed48 | 1577 | { |
717a94b5 N |
1578 | current->flags &= ~flags; |
1579 | current->flags |= orig_flags & flags; | |
907aed48 MG |
1580 | } |
1581 | ||
5eca1c10 IM |
1582 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial); |
1583 | extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_cpus_allowed); | |
1da177e4 | 1584 | #ifdef CONFIG_SMP |
5eca1c10 IM |
1585 | extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); |
1586 | extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask); | |
1da177e4 | 1587 | #else |
5eca1c10 | 1588 | static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) |
1e1b6c51 KM |
1589 | { |
1590 | } | |
5eca1c10 | 1591 | static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) |
1da177e4 | 1592 | { |
96f874e2 | 1593 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
1594 | return -EINVAL; |
1595 | return 0; | |
1596 | } | |
1597 | #endif | |
e0ad9556 | 1598 | |
fa93384f | 1599 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
1600 | extern void set_user_nice(struct task_struct *p, long nice); |
1601 | extern int task_prio(const struct task_struct *p); | |
5eca1c10 | 1602 | |
d0ea0268 DY |
1603 | /** |
1604 | * task_nice - return the nice value of a given task. | |
1605 | * @p: the task in question. | |
1606 | * | |
1607 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
1608 | */ | |
1609 | static inline int task_nice(const struct task_struct *p) | |
1610 | { | |
1611 | return PRIO_TO_NICE((p)->static_prio); | |
1612 | } | |
5eca1c10 | 1613 | |
36c8b586 IM |
1614 | extern int can_nice(const struct task_struct *p, const int nice); |
1615 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 1616 | extern int idle_cpu(int cpu); |
943d355d | 1617 | extern int available_idle_cpu(int cpu); |
5eca1c10 IM |
1618 | extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *); |
1619 | extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *); | |
1620 | extern int sched_setattr(struct task_struct *, const struct sched_attr *); | |
794a56eb | 1621 | extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *); |
36c8b586 | 1622 | extern struct task_struct *idle_task(int cpu); |
5eca1c10 | 1623 | |
c4f30608 PM |
1624 | /** |
1625 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 1626 | * @p: the task in question. |
e69f6186 YB |
1627 | * |
1628 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 1629 | */ |
7061ca3b | 1630 | static inline bool is_idle_task(const struct task_struct *p) |
c4f30608 | 1631 | { |
c1de45ca | 1632 | return !!(p->flags & PF_IDLE); |
c4f30608 | 1633 | } |
5eca1c10 | 1634 | |
36c8b586 | 1635 | extern struct task_struct *curr_task(int cpu); |
a458ae2e | 1636 | extern void ia64_set_curr_task(int cpu, struct task_struct *p); |
1da177e4 LT |
1637 | |
1638 | void yield(void); | |
1639 | ||
1da177e4 | 1640 | union thread_union { |
0500871f DH |
1641 | #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK |
1642 | struct task_struct task; | |
1643 | #endif | |
c65eacbe | 1644 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1da177e4 | 1645 | struct thread_info thread_info; |
c65eacbe | 1646 | #endif |
1da177e4 LT |
1647 | unsigned long stack[THREAD_SIZE/sizeof(long)]; |
1648 | }; | |
1649 | ||
0500871f DH |
1650 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1651 | extern struct thread_info init_thread_info; | |
1652 | #endif | |
1653 | ||
1654 | extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)]; | |
1655 | ||
f3ac6067 IM |
1656 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1657 | static inline struct thread_info *task_thread_info(struct task_struct *task) | |
1658 | { | |
1659 | return &task->thread_info; | |
1660 | } | |
1661 | #elif !defined(__HAVE_THREAD_FUNCTIONS) | |
1662 | # define task_thread_info(task) ((struct thread_info *)(task)->stack) | |
1663 | #endif | |
1664 | ||
198fe21b PE |
1665 | /* |
1666 | * find a task by one of its numerical ids | |
1667 | * | |
198fe21b PE |
1668 | * find_task_by_pid_ns(): |
1669 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
1670 | * find_task_by_vpid(): |
1671 | * finds a task by its virtual pid | |
198fe21b | 1672 | * |
e49859e7 | 1673 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
1674 | */ |
1675 | ||
228ebcbe | 1676 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
5eca1c10 | 1677 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns); |
198fe21b | 1678 | |
2ee08260 MR |
1679 | /* |
1680 | * find a task by its virtual pid and get the task struct | |
1681 | */ | |
1682 | extern struct task_struct *find_get_task_by_vpid(pid_t nr); | |
1683 | ||
b3c97528 HH |
1684 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
1685 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 1686 | extern void wake_up_new_task(struct task_struct *tsk); |
5eca1c10 | 1687 | |
1da177e4 | 1688 | #ifdef CONFIG_SMP |
5eca1c10 | 1689 | extern void kick_process(struct task_struct *tsk); |
1da177e4 | 1690 | #else |
5eca1c10 | 1691 | static inline void kick_process(struct task_struct *tsk) { } |
1da177e4 | 1692 | #endif |
1da177e4 | 1693 | |
82b89778 | 1694 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
5eca1c10 | 1695 | |
82b89778 AH |
1696 | static inline void set_task_comm(struct task_struct *tsk, const char *from) |
1697 | { | |
1698 | __set_task_comm(tsk, from, false); | |
1699 | } | |
5eca1c10 | 1700 | |
3756f640 AB |
1701 | extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk); |
1702 | #define get_task_comm(buf, tsk) ({ \ | |
1703 | BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \ | |
1704 | __get_task_comm(buf, sizeof(buf), tsk); \ | |
1705 | }) | |
1da177e4 LT |
1706 | |
1707 | #ifdef CONFIG_SMP | |
317f3941 | 1708 | void scheduler_ipi(void); |
85ba2d86 | 1709 | extern unsigned long wait_task_inactive(struct task_struct *, long match_state); |
1da177e4 | 1710 | #else |
184748cc | 1711 | static inline void scheduler_ipi(void) { } |
5eca1c10 | 1712 | static inline unsigned long wait_task_inactive(struct task_struct *p, long match_state) |
85ba2d86 RM |
1713 | { |
1714 | return 1; | |
1715 | } | |
1da177e4 LT |
1716 | #endif |
1717 | ||
5eca1c10 IM |
1718 | /* |
1719 | * Set thread flags in other task's structures. | |
1720 | * See asm/thread_info.h for TIF_xxxx flags available: | |
1da177e4 LT |
1721 | */ |
1722 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1723 | { | |
a1261f54 | 1724 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1725 | } |
1726 | ||
1727 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1728 | { | |
a1261f54 | 1729 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1730 | } |
1731 | ||
93ee37c2 DM |
1732 | static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag, |
1733 | bool value) | |
1734 | { | |
1735 | update_ti_thread_flag(task_thread_info(tsk), flag, value); | |
1736 | } | |
1737 | ||
1da177e4 LT |
1738 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) |
1739 | { | |
a1261f54 | 1740 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1741 | } |
1742 | ||
1743 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1744 | { | |
a1261f54 | 1745 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1746 | } |
1747 | ||
1748 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
1749 | { | |
a1261f54 | 1750 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
1751 | } |
1752 | ||
1753 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
1754 | { | |
1755 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1756 | } | |
1757 | ||
1758 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
1759 | { | |
1760 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
1761 | } | |
1762 | ||
8ae121ac GH |
1763 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
1764 | { | |
1765 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
1766 | } | |
1767 | ||
1da177e4 LT |
1768 | /* |
1769 | * cond_resched() and cond_resched_lock(): latency reduction via | |
1770 | * explicit rescheduling in places that are safe. The return | |
1771 | * value indicates whether a reschedule was done in fact. | |
1772 | * cond_resched_lock() will drop the spinlock before scheduling, | |
1da177e4 | 1773 | */ |
35a773a0 | 1774 | #ifndef CONFIG_PREEMPT |
c3921ab7 | 1775 | extern int _cond_resched(void); |
35a773a0 PZ |
1776 | #else |
1777 | static inline int _cond_resched(void) { return 0; } | |
1778 | #endif | |
6f80bd98 | 1779 | |
613afbf8 | 1780 | #define cond_resched() ({ \ |
3427445a | 1781 | ___might_sleep(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
1782 | _cond_resched(); \ |
1783 | }) | |
6f80bd98 | 1784 | |
613afbf8 FW |
1785 | extern int __cond_resched_lock(spinlock_t *lock); |
1786 | ||
1787 | #define cond_resched_lock(lock) ({ \ | |
3427445a | 1788 | ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ |
613afbf8 FW |
1789 | __cond_resched_lock(lock); \ |
1790 | }) | |
1791 | ||
f6f3c437 SH |
1792 | static inline void cond_resched_rcu(void) |
1793 | { | |
1794 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
1795 | rcu_read_unlock(); | |
1796 | cond_resched(); | |
1797 | rcu_read_lock(); | |
1798 | #endif | |
1799 | } | |
1800 | ||
1da177e4 LT |
1801 | /* |
1802 | * Does a critical section need to be broken due to another | |
95c354fe NP |
1803 | * task waiting?: (technically does not depend on CONFIG_PREEMPT, |
1804 | * but a general need for low latency) | |
1da177e4 | 1805 | */ |
95c354fe | 1806 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 1807 | { |
95c354fe NP |
1808 | #ifdef CONFIG_PREEMPT |
1809 | return spin_is_contended(lock); | |
1810 | #else | |
1da177e4 | 1811 | return 0; |
95c354fe | 1812 | #endif |
1da177e4 LT |
1813 | } |
1814 | ||
75f93fed PZ |
1815 | static __always_inline bool need_resched(void) |
1816 | { | |
1817 | return unlikely(tif_need_resched()); | |
1818 | } | |
1819 | ||
1da177e4 LT |
1820 | /* |
1821 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
1822 | */ | |
1823 | #ifdef CONFIG_SMP | |
1824 | ||
1825 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1826 | { | |
c65eacbe | 1827 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
c546951d | 1828 | return READ_ONCE(p->cpu); |
c65eacbe | 1829 | #else |
c546951d | 1830 | return READ_ONCE(task_thread_info(p)->cpu); |
c65eacbe | 1831 | #endif |
1da177e4 LT |
1832 | } |
1833 | ||
c65cc870 | 1834 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
1835 | |
1836 | #else | |
1837 | ||
1838 | static inline unsigned int task_cpu(const struct task_struct *p) | |
1839 | { | |
1840 | return 0; | |
1841 | } | |
1842 | ||
1843 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
1844 | { | |
1845 | } | |
1846 | ||
1847 | #endif /* CONFIG_SMP */ | |
1848 | ||
d9345c65 PX |
1849 | /* |
1850 | * In order to reduce various lock holder preemption latencies provide an | |
1851 | * interface to see if a vCPU is currently running or not. | |
1852 | * | |
1853 | * This allows us to terminate optimistic spin loops and block, analogous to | |
1854 | * the native optimistic spin heuristic of testing if the lock owner task is | |
1855 | * running or not. | |
1856 | */ | |
1857 | #ifndef vcpu_is_preempted | |
1858 | # define vcpu_is_preempted(cpu) false | |
1859 | #endif | |
1860 | ||
96f874e2 RR |
1861 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
1862 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 1863 | |
82455257 DH |
1864 | #ifndef TASK_SIZE_OF |
1865 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
1866 | #endif | |
1867 | ||
d7822b1e MD |
1868 | #ifdef CONFIG_RSEQ |
1869 | ||
1870 | /* | |
1871 | * Map the event mask on the user-space ABI enum rseq_cs_flags | |
1872 | * for direct mask checks. | |
1873 | */ | |
1874 | enum rseq_event_mask_bits { | |
1875 | RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT, | |
1876 | RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT, | |
1877 | RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT, | |
1878 | }; | |
1879 | ||
1880 | enum rseq_event_mask { | |
1881 | RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT), | |
1882 | RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT), | |
1883 | RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT), | |
1884 | }; | |
1885 | ||
1886 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
1887 | { | |
1888 | if (t->rseq) | |
1889 | set_tsk_thread_flag(t, TIF_NOTIFY_RESUME); | |
1890 | } | |
1891 | ||
784e0300 | 1892 | void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs); |
d7822b1e | 1893 | |
784e0300 WD |
1894 | static inline void rseq_handle_notify_resume(struct ksignal *ksig, |
1895 | struct pt_regs *regs) | |
d7822b1e MD |
1896 | { |
1897 | if (current->rseq) | |
784e0300 | 1898 | __rseq_handle_notify_resume(ksig, regs); |
d7822b1e MD |
1899 | } |
1900 | ||
784e0300 WD |
1901 | static inline void rseq_signal_deliver(struct ksignal *ksig, |
1902 | struct pt_regs *regs) | |
d7822b1e MD |
1903 | { |
1904 | preempt_disable(); | |
1905 | __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask); | |
1906 | preempt_enable(); | |
784e0300 | 1907 | rseq_handle_notify_resume(ksig, regs); |
d7822b1e MD |
1908 | } |
1909 | ||
1910 | /* rseq_preempt() requires preemption to be disabled. */ | |
1911 | static inline void rseq_preempt(struct task_struct *t) | |
1912 | { | |
1913 | __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask); | |
1914 | rseq_set_notify_resume(t); | |
1915 | } | |
1916 | ||
1917 | /* rseq_migrate() requires preemption to be disabled. */ | |
1918 | static inline void rseq_migrate(struct task_struct *t) | |
1919 | { | |
1920 | __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask); | |
1921 | rseq_set_notify_resume(t); | |
1922 | } | |
1923 | ||
1924 | /* | |
1925 | * If parent process has a registered restartable sequences area, the | |
9a789fcf | 1926 | * child inherits. Only applies when forking a process, not a thread. |
d7822b1e MD |
1927 | */ |
1928 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
1929 | { | |
1930 | if (clone_flags & CLONE_THREAD) { | |
1931 | t->rseq = NULL; | |
d7822b1e MD |
1932 | t->rseq_sig = 0; |
1933 | t->rseq_event_mask = 0; | |
1934 | } else { | |
1935 | t->rseq = current->rseq; | |
d7822b1e MD |
1936 | t->rseq_sig = current->rseq_sig; |
1937 | t->rseq_event_mask = current->rseq_event_mask; | |
d7822b1e MD |
1938 | } |
1939 | } | |
1940 | ||
1941 | static inline void rseq_execve(struct task_struct *t) | |
1942 | { | |
1943 | t->rseq = NULL; | |
d7822b1e MD |
1944 | t->rseq_sig = 0; |
1945 | t->rseq_event_mask = 0; | |
1946 | } | |
1947 | ||
1948 | #else | |
1949 | ||
1950 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
1951 | { | |
1952 | } | |
784e0300 WD |
1953 | static inline void rseq_handle_notify_resume(struct ksignal *ksig, |
1954 | struct pt_regs *regs) | |
d7822b1e MD |
1955 | { |
1956 | } | |
784e0300 WD |
1957 | static inline void rseq_signal_deliver(struct ksignal *ksig, |
1958 | struct pt_regs *regs) | |
d7822b1e MD |
1959 | { |
1960 | } | |
1961 | static inline void rseq_preempt(struct task_struct *t) | |
1962 | { | |
1963 | } | |
1964 | static inline void rseq_migrate(struct task_struct *t) | |
1965 | { | |
1966 | } | |
1967 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
1968 | { | |
1969 | } | |
1970 | static inline void rseq_execve(struct task_struct *t) | |
1971 | { | |
1972 | } | |
1973 | ||
1974 | #endif | |
1975 | ||
73ab1cb2 TY |
1976 | void __exit_umh(struct task_struct *tsk); |
1977 | ||
1978 | static inline void exit_umh(struct task_struct *tsk) | |
1979 | { | |
1980 | if (unlikely(tsk->flags & PF_UMH)) | |
1981 | __exit_umh(tsk); | |
1982 | } | |
1983 | ||
d7822b1e MD |
1984 | #ifdef CONFIG_DEBUG_RSEQ |
1985 | ||
1986 | void rseq_syscall(struct pt_regs *regs); | |
1987 | ||
1988 | #else | |
1989 | ||
1990 | static inline void rseq_syscall(struct pt_regs *regs) | |
1991 | { | |
1992 | } | |
1993 | ||
1994 | #endif | |
1995 | ||
3c93a0c0 QY |
1996 | const struct sched_avg *sched_trace_cfs_rq_avg(struct cfs_rq *cfs_rq); |
1997 | char *sched_trace_cfs_rq_path(struct cfs_rq *cfs_rq, char *str, int len); | |
1998 | int sched_trace_cfs_rq_cpu(struct cfs_rq *cfs_rq); | |
1999 | ||
2000 | const struct sched_avg *sched_trace_rq_avg_rt(struct rq *rq); | |
2001 | const struct sched_avg *sched_trace_rq_avg_dl(struct rq *rq); | |
2002 | const struct sched_avg *sched_trace_rq_avg_irq(struct rq *rq); | |
2003 | ||
2004 | int sched_trace_rq_cpu(struct rq *rq); | |
2005 | ||
2006 | const struct cpumask *sched_trace_rd_span(struct root_domain *rd); | |
2007 | ||
1da177e4 | 2008 | #endif |