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> |
f80be457 | 17 | #include <linux/kmsan_types.h> |
5eca1c10 IM |
18 | #include <linux/mutex.h> |
19 | #include <linux/plist.h> | |
20 | #include <linux/hrtimer.h> | |
0584df9c | 21 | #include <linux/irqflags.h> |
1da177e4 | 22 | #include <linux/seccomp.h> |
5eca1c10 | 23 | #include <linux/nodemask.h> |
b68070e1 | 24 | #include <linux/rcupdate.h> |
ec1d2819 | 25 | #include <linux/refcount.h> |
a3b6714e | 26 | #include <linux/resource.h> |
9745512c | 27 | #include <linux/latencytop.h> |
5eca1c10 | 28 | #include <linux/sched/prio.h> |
9eacb5c7 | 29 | #include <linux/sched/types.h> |
5eca1c10 | 30 | #include <linux/signal_types.h> |
1446e1df | 31 | #include <linux/syscall_user_dispatch.h> |
5eca1c10 IM |
32 | #include <linux/mm_types_task.h> |
33 | #include <linux/task_io_accounting.h> | |
2b69942f | 34 | #include <linux/posix-timers.h> |
d7822b1e | 35 | #include <linux/rseq.h> |
0cd39f46 | 36 | #include <linux/seqlock.h> |
dfd402a4 | 37 | #include <linux/kcsan.h> |
102227b9 | 38 | #include <linux/rv.h> |
e3ff7c60 | 39 | #include <linux/livepatch_sched.h> |
5fbda3ec | 40 | #include <asm/kmap_size.h> |
a3b6714e | 41 | |
5eca1c10 | 42 | /* task_struct member predeclarations (sorted alphabetically): */ |
c7af7877 | 43 | struct audit_context; |
c7af7877 | 44 | struct backing_dev_info; |
bddd87c7 | 45 | struct bio_list; |
73c10101 | 46 | struct blk_plug; |
a10787e6 | 47 | struct bpf_local_storage; |
c7603cfa | 48 | struct bpf_run_ctx; |
3c93a0c0 | 49 | struct capture_control; |
c7af7877 | 50 | struct cfs_rq; |
c7af7877 IM |
51 | struct fs_struct; |
52 | struct futex_pi_state; | |
53 | struct io_context; | |
1875dc5b | 54 | struct io_uring_task; |
c7af7877 | 55 | struct mempolicy; |
89076bc3 | 56 | struct nameidata; |
c7af7877 IM |
57 | struct nsproxy; |
58 | struct perf_event_context; | |
59 | struct pid_namespace; | |
60 | struct pipe_inode_info; | |
61 | struct rcu_node; | |
62 | struct reclaim_state; | |
63 | struct robust_list_head; | |
3c93a0c0 QY |
64 | struct root_domain; |
65 | struct rq; | |
c7af7877 IM |
66 | struct sched_attr; |
67 | struct sched_param; | |
43ae34cb | 68 | struct seq_file; |
c7af7877 IM |
69 | struct sighand_struct; |
70 | struct signal_struct; | |
71 | struct task_delay_info; | |
4cf86d77 | 72 | struct task_group; |
fd593511 | 73 | struct user_event_mm; |
1da177e4 | 74 | |
4a8342d2 LT |
75 | /* |
76 | * Task state bitmask. NOTE! These bits are also | |
77 | * encoded in fs/proc/array.c: get_task_state(). | |
78 | * | |
79 | * We have two separate sets of flags: task->state | |
80 | * is about runnability, while task->exit_state are | |
81 | * about the task exiting. Confusing, but this way | |
82 | * modifying one set can't modify the other one by | |
83 | * mistake. | |
84 | */ | |
5eca1c10 IM |
85 | |
86 | /* Used in tsk->state: */ | |
9963e444 PZ |
87 | #define TASK_RUNNING 0x00000000 |
88 | #define TASK_INTERRUPTIBLE 0x00000001 | |
89 | #define TASK_UNINTERRUPTIBLE 0x00000002 | |
90 | #define __TASK_STOPPED 0x00000004 | |
91 | #define __TASK_TRACED 0x00000008 | |
5eca1c10 | 92 | /* Used in tsk->exit_state: */ |
9963e444 PZ |
93 | #define EXIT_DEAD 0x00000010 |
94 | #define EXIT_ZOMBIE 0x00000020 | |
5eca1c10 IM |
95 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) |
96 | /* Used in tsk->state again: */ | |
9963e444 PZ |
97 | #define TASK_PARKED 0x00000040 |
98 | #define TASK_DEAD 0x00000080 | |
99 | #define TASK_WAKEKILL 0x00000100 | |
100 | #define TASK_WAKING 0x00000200 | |
101 | #define TASK_NOLOAD 0x00000400 | |
102 | #define TASK_NEW 0x00000800 | |
9963e444 | 103 | #define TASK_RTLOCK_WAIT 0x00001000 |
f5d39b02 PZ |
104 | #define TASK_FREEZABLE 0x00002000 |
105 | #define __TASK_FREEZABLE_UNSAFE (0x00004000 * IS_ENABLED(CONFIG_LOCKDEP)) | |
106 | #define TASK_FROZEN 0x00008000 | |
107 | #define TASK_STATE_MAX 0x00010000 | |
5eca1c10 | 108 | |
f9fc8cad PZ |
109 | #define TASK_ANY (TASK_STATE_MAX-1) |
110 | ||
f5d39b02 PZ |
111 | /* |
112 | * DO NOT ADD ANY NEW USERS ! | |
113 | */ | |
114 | #define TASK_FREEZABLE_UNSAFE (TASK_FREEZABLE | __TASK_FREEZABLE_UNSAFE) | |
5eca1c10 | 115 | |
5eca1c10 IM |
116 | /* Convenience macros for the sake of set_current_state: */ |
117 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) | |
118 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
2500ad1c | 119 | #define TASK_TRACED __TASK_TRACED |
5eca1c10 IM |
120 | |
121 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) | |
122 | ||
123 | /* Convenience macros for the sake of wake_up(): */ | |
124 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
5eca1c10 IM |
125 | |
126 | /* get_task_state(): */ | |
127 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
128 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ | |
8ef9925b PZ |
129 | __TASK_TRACED | EXIT_DEAD | EXIT_ZOMBIE | \ |
130 | TASK_PARKED) | |
5eca1c10 | 131 | |
2f064a59 | 132 | #define task_is_running(task) (READ_ONCE((task)->__state) == TASK_RUNNING) |
5eca1c10 | 133 | |
31cae1ea PZ |
134 | #define task_is_traced(task) ((READ_ONCE(task->jobctl) & JOBCTL_TRACED) != 0) |
135 | #define task_is_stopped(task) ((READ_ONCE(task->jobctl) & JOBCTL_STOPPED) != 0) | |
136 | #define task_is_stopped_or_traced(task) ((READ_ONCE(task->jobctl) & (JOBCTL_STOPPED | JOBCTL_TRACED)) != 0) | |
5eca1c10 | 137 | |
b5bf9a90 PZ |
138 | /* |
139 | * Special states are those that do not use the normal wait-loop pattern. See | |
140 | * the comment with set_special_state(). | |
141 | */ | |
142 | #define is_special_task_state(state) \ | |
1cef1150 | 143 | ((state) & (__TASK_STOPPED | __TASK_TRACED | TASK_PARKED | TASK_DEAD)) |
b5bf9a90 | 144 | |
85019c16 TG |
145 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
146 | # define debug_normal_state_change(state_value) \ | |
147 | do { \ | |
148 | WARN_ON_ONCE(is_special_task_state(state_value)); \ | |
149 | current->task_state_change = _THIS_IP_; \ | |
8eb23b9f PZ |
150 | } while (0) |
151 | ||
85019c16 | 152 | # define debug_special_state_change(state_value) \ |
b5bf9a90 | 153 | do { \ |
b5bf9a90 | 154 | WARN_ON_ONCE(!is_special_task_state(state_value)); \ |
b5bf9a90 | 155 | current->task_state_change = _THIS_IP_; \ |
b5bf9a90 | 156 | } while (0) |
85019c16 | 157 | |
5f220be2 TG |
158 | # define debug_rtlock_wait_set_state() \ |
159 | do { \ | |
160 | current->saved_state_change = current->task_state_change;\ | |
161 | current->task_state_change = _THIS_IP_; \ | |
162 | } while (0) | |
163 | ||
164 | # define debug_rtlock_wait_restore_state() \ | |
165 | do { \ | |
166 | current->task_state_change = current->saved_state_change;\ | |
167 | } while (0) | |
168 | ||
8eb23b9f | 169 | #else |
85019c16 TG |
170 | # define debug_normal_state_change(cond) do { } while (0) |
171 | # define debug_special_state_change(cond) do { } while (0) | |
5f220be2 TG |
172 | # define debug_rtlock_wait_set_state() do { } while (0) |
173 | # define debug_rtlock_wait_restore_state() do { } while (0) | |
85019c16 TG |
174 | #endif |
175 | ||
498d0c57 AM |
176 | /* |
177 | * set_current_state() includes a barrier so that the write of current->state | |
178 | * is correctly serialised wrt the caller's subsequent test of whether to | |
179 | * actually sleep: | |
180 | * | |
a2250238 | 181 | * for (;;) { |
498d0c57 | 182 | * set_current_state(TASK_UNINTERRUPTIBLE); |
58877d34 PZ |
183 | * if (CONDITION) |
184 | * break; | |
a2250238 PZ |
185 | * |
186 | * schedule(); | |
187 | * } | |
188 | * __set_current_state(TASK_RUNNING); | |
189 | * | |
190 | * If the caller does not need such serialisation (because, for instance, the | |
58877d34 | 191 | * CONDITION test and condition change and wakeup are under the same lock) then |
a2250238 PZ |
192 | * use __set_current_state(). |
193 | * | |
194 | * The above is typically ordered against the wakeup, which does: | |
195 | * | |
58877d34 | 196 | * CONDITION = 1; |
b5bf9a90 | 197 | * wake_up_state(p, TASK_UNINTERRUPTIBLE); |
a2250238 | 198 | * |
58877d34 PZ |
199 | * where wake_up_state()/try_to_wake_up() executes a full memory barrier before |
200 | * accessing p->state. | |
a2250238 PZ |
201 | * |
202 | * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, | |
203 | * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a | |
204 | * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). | |
498d0c57 | 205 | * |
b5bf9a90 | 206 | * However, with slightly different timing the wakeup TASK_RUNNING store can |
dfcb245e | 207 | * also collide with the TASK_UNINTERRUPTIBLE store. Losing that store is not |
b5bf9a90 PZ |
208 | * a problem either because that will result in one extra go around the loop |
209 | * and our @cond test will save the day. | |
498d0c57 | 210 | * |
a2250238 | 211 | * Also see the comments of try_to_wake_up(). |
498d0c57 | 212 | */ |
b5bf9a90 | 213 | #define __set_current_state(state_value) \ |
85019c16 TG |
214 | do { \ |
215 | debug_normal_state_change((state_value)); \ | |
216 | WRITE_ONCE(current->__state, (state_value)); \ | |
217 | } while (0) | |
b5bf9a90 PZ |
218 | |
219 | #define set_current_state(state_value) \ | |
85019c16 TG |
220 | do { \ |
221 | debug_normal_state_change((state_value)); \ | |
222 | smp_store_mb(current->__state, (state_value)); \ | |
223 | } while (0) | |
b5bf9a90 PZ |
224 | |
225 | /* | |
226 | * set_special_state() should be used for those states when the blocking task | |
227 | * can not use the regular condition based wait-loop. In that case we must | |
85019c16 TG |
228 | * serialize against wakeups such that any possible in-flight TASK_RUNNING |
229 | * stores will not collide with our state change. | |
b5bf9a90 PZ |
230 | */ |
231 | #define set_special_state(state_value) \ | |
232 | do { \ | |
233 | unsigned long flags; /* may shadow */ \ | |
85019c16 | 234 | \ |
b5bf9a90 | 235 | raw_spin_lock_irqsave(¤t->pi_lock, flags); \ |
85019c16 | 236 | debug_special_state_change((state_value)); \ |
2f064a59 | 237 | WRITE_ONCE(current->__state, (state_value)); \ |
b5bf9a90 PZ |
238 | raw_spin_unlock_irqrestore(¤t->pi_lock, flags); \ |
239 | } while (0) | |
240 | ||
5f220be2 TG |
241 | /* |
242 | * PREEMPT_RT specific variants for "sleeping" spin/rwlocks | |
243 | * | |
244 | * RT's spin/rwlock substitutions are state preserving. The state of the | |
245 | * task when blocking on the lock is saved in task_struct::saved_state and | |
246 | * restored after the lock has been acquired. These operations are | |
247 | * serialized by task_struct::pi_lock against try_to_wake_up(). Any non RT | |
248 | * lock related wakeups while the task is blocked on the lock are | |
249 | * redirected to operate on task_struct::saved_state to ensure that these | |
250 | * are not dropped. On restore task_struct::saved_state is set to | |
251 | * TASK_RUNNING so any wakeup attempt redirected to saved_state will fail. | |
252 | * | |
253 | * The lock operation looks like this: | |
254 | * | |
255 | * current_save_and_set_rtlock_wait_state(); | |
256 | * for (;;) { | |
257 | * if (try_lock()) | |
258 | * break; | |
259 | * raw_spin_unlock_irq(&lock->wait_lock); | |
260 | * schedule_rtlock(); | |
261 | * raw_spin_lock_irq(&lock->wait_lock); | |
262 | * set_current_state(TASK_RTLOCK_WAIT); | |
263 | * } | |
264 | * current_restore_rtlock_saved_state(); | |
265 | */ | |
266 | #define current_save_and_set_rtlock_wait_state() \ | |
267 | do { \ | |
268 | lockdep_assert_irqs_disabled(); \ | |
269 | raw_spin_lock(¤t->pi_lock); \ | |
270 | current->saved_state = current->__state; \ | |
271 | debug_rtlock_wait_set_state(); \ | |
272 | WRITE_ONCE(current->__state, TASK_RTLOCK_WAIT); \ | |
273 | raw_spin_unlock(¤t->pi_lock); \ | |
274 | } while (0); | |
275 | ||
276 | #define current_restore_rtlock_saved_state() \ | |
277 | do { \ | |
278 | lockdep_assert_irqs_disabled(); \ | |
279 | raw_spin_lock(¤t->pi_lock); \ | |
280 | debug_rtlock_wait_restore_state(); \ | |
281 | WRITE_ONCE(current->__state, current->saved_state); \ | |
282 | current->saved_state = TASK_RUNNING; \ | |
283 | raw_spin_unlock(¤t->pi_lock); \ | |
284 | } while (0); | |
8eb23b9f | 285 | |
2f064a59 | 286 | #define get_current_state() READ_ONCE(current->__state) |
d6c23bb3 | 287 | |
3087c61e YS |
288 | /* |
289 | * Define the task command name length as enum, then it can be visible to | |
290 | * BPF programs. | |
291 | */ | |
292 | enum { | |
293 | TASK_COMM_LEN = 16, | |
294 | }; | |
1da177e4 | 295 | |
1da177e4 LT |
296 | extern void scheduler_tick(void); |
297 | ||
5eca1c10 IM |
298 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX |
299 | ||
300 | extern long schedule_timeout(long timeout); | |
301 | extern long schedule_timeout_interruptible(long timeout); | |
302 | extern long schedule_timeout_killable(long timeout); | |
303 | extern long schedule_timeout_uninterruptible(long timeout); | |
304 | extern long schedule_timeout_idle(long timeout); | |
1da177e4 | 305 | asmlinkage void schedule(void); |
c5491ea7 | 306 | extern void schedule_preempt_disabled(void); |
19c95f26 | 307 | asmlinkage void preempt_schedule_irq(void); |
6991436c TG |
308 | #ifdef CONFIG_PREEMPT_RT |
309 | extern void schedule_rtlock(void); | |
310 | #endif | |
1da177e4 | 311 | |
10ab5643 TH |
312 | extern int __must_check io_schedule_prepare(void); |
313 | extern void io_schedule_finish(int token); | |
9cff8ade | 314 | extern long io_schedule_timeout(long timeout); |
10ab5643 | 315 | extern void io_schedule(void); |
9cff8ade | 316 | |
d37f761d | 317 | /** |
0ba42a59 | 318 | * struct prev_cputime - snapshot of system and user cputime |
d37f761d FW |
319 | * @utime: time spent in user mode |
320 | * @stime: time spent in system mode | |
9d7fb042 | 321 | * @lock: protects the above two fields |
d37f761d | 322 | * |
9d7fb042 PZ |
323 | * Stores previous user/system time values such that we can guarantee |
324 | * monotonicity. | |
d37f761d | 325 | */ |
9d7fb042 PZ |
326 | struct prev_cputime { |
327 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
5eca1c10 IM |
328 | u64 utime; |
329 | u64 stime; | |
330 | raw_spinlock_t lock; | |
9d7fb042 | 331 | #endif |
d37f761d FW |
332 | }; |
333 | ||
bac5b6b6 FW |
334 | enum vtime_state { |
335 | /* Task is sleeping or running in a CPU with VTIME inactive: */ | |
336 | VTIME_INACTIVE = 0, | |
14faf6fc FW |
337 | /* Task is idle */ |
338 | VTIME_IDLE, | |
bac5b6b6 FW |
339 | /* Task runs in kernelspace in a CPU with VTIME active: */ |
340 | VTIME_SYS, | |
14faf6fc FW |
341 | /* Task runs in userspace in a CPU with VTIME active: */ |
342 | VTIME_USER, | |
e6d5bf3e FW |
343 | /* Task runs as guests in a CPU with VTIME active: */ |
344 | VTIME_GUEST, | |
bac5b6b6 FW |
345 | }; |
346 | ||
347 | struct vtime { | |
348 | seqcount_t seqcount; | |
349 | unsigned long long starttime; | |
350 | enum vtime_state state; | |
802f4a82 | 351 | unsigned int cpu; |
2a42eb95 WL |
352 | u64 utime; |
353 | u64 stime; | |
354 | u64 gtime; | |
bac5b6b6 FW |
355 | }; |
356 | ||
69842cba PB |
357 | /* |
358 | * Utilization clamp constraints. | |
359 | * @UCLAMP_MIN: Minimum utilization | |
360 | * @UCLAMP_MAX: Maximum utilization | |
361 | * @UCLAMP_CNT: Utilization clamp constraints count | |
362 | */ | |
363 | enum uclamp_id { | |
364 | UCLAMP_MIN = 0, | |
365 | UCLAMP_MAX, | |
366 | UCLAMP_CNT | |
367 | }; | |
368 | ||
f9a25f77 MP |
369 | #ifdef CONFIG_SMP |
370 | extern struct root_domain def_root_domain; | |
371 | extern struct mutex sched_domains_mutex; | |
372 | #endif | |
373 | ||
1da177e4 | 374 | struct sched_info { |
7f5f8e8d | 375 | #ifdef CONFIG_SCHED_INFO |
5eca1c10 IM |
376 | /* Cumulative counters: */ |
377 | ||
378 | /* # of times we have run on this CPU: */ | |
379 | unsigned long pcount; | |
380 | ||
381 | /* Time spent waiting on a runqueue: */ | |
382 | unsigned long long run_delay; | |
383 | ||
384 | /* Timestamps: */ | |
385 | ||
386 | /* When did we last run on a CPU? */ | |
387 | unsigned long long last_arrival; | |
388 | ||
389 | /* When were we last queued to run? */ | |
390 | unsigned long long last_queued; | |
1da177e4 | 391 | |
f6db8347 | 392 | #endif /* CONFIG_SCHED_INFO */ |
7f5f8e8d | 393 | }; |
1da177e4 | 394 | |
6ecdd749 YD |
395 | /* |
396 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
397 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
398 | * | |
399 | * We define a basic fixed point arithmetic range, and then formalize | |
400 | * all these metrics based on that basic range. | |
401 | */ | |
5eca1c10 IM |
402 | # define SCHED_FIXEDPOINT_SHIFT 10 |
403 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
6ecdd749 | 404 | |
69842cba PB |
405 | /* Increase resolution of cpu_capacity calculations */ |
406 | # define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT | |
407 | # define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT) | |
408 | ||
20b8a59f | 409 | struct load_weight { |
5eca1c10 IM |
410 | unsigned long weight; |
411 | u32 inv_weight; | |
20b8a59f IM |
412 | }; |
413 | ||
7f65ea42 PB |
414 | /** |
415 | * struct util_est - Estimation utilization of FAIR tasks | |
416 | * @enqueued: instantaneous estimated utilization of a task/cpu | |
417 | * @ewma: the Exponential Weighted Moving Average (EWMA) | |
418 | * utilization of a task | |
419 | * | |
420 | * Support data structure to track an Exponential Weighted Moving Average | |
421 | * (EWMA) of a FAIR task's utilization. New samples are added to the moving | |
422 | * average each time a task completes an activation. Sample's weight is chosen | |
423 | * so that the EWMA will be relatively insensitive to transient changes to the | |
424 | * task's workload. | |
425 | * | |
426 | * The enqueued attribute has a slightly different meaning for tasks and cpus: | |
427 | * - task: the task's util_avg at last task dequeue time | |
428 | * - cfs_rq: the sum of util_est.enqueued for each RUNNABLE task on that CPU | |
429 | * Thus, the util_est.enqueued of a task represents the contribution on the | |
430 | * estimated utilization of the CPU where that task is currently enqueued. | |
431 | * | |
432 | * Only for tasks we track a moving average of the past instantaneous | |
433 | * estimated utilization. This allows to absorb sporadic drops in utilization | |
434 | * of an otherwise almost periodic task. | |
68d7a190 DE |
435 | * |
436 | * The UTIL_AVG_UNCHANGED flag is used to synchronize util_est with util_avg | |
437 | * updates. When a task is dequeued, its util_est should not be updated if its | |
438 | * util_avg has not been updated in the meantime. | |
439 | * This information is mapped into the MSB bit of util_est.enqueued at dequeue | |
440 | * time. Since max value of util_est.enqueued for a task is 1024 (PELT util_avg | |
441 | * for a task) it is safe to use MSB. | |
7f65ea42 PB |
442 | */ |
443 | struct util_est { | |
444 | unsigned int enqueued; | |
445 | unsigned int ewma; | |
446 | #define UTIL_EST_WEIGHT_SHIFT 2 | |
68d7a190 | 447 | #define UTIL_AVG_UNCHANGED 0x80000000 |
317d359d | 448 | } __attribute__((__aligned__(sizeof(u64)))); |
7f65ea42 | 449 | |
9d89c257 | 450 | /* |
9f683953 | 451 | * The load/runnable/util_avg accumulates an infinite geometric series |
0dacee1b | 452 | * (see __update_load_avg_cfs_rq() in kernel/sched/pelt.c). |
7b595334 YD |
453 | * |
454 | * [load_avg definition] | |
455 | * | |
456 | * load_avg = runnable% * scale_load_down(load) | |
457 | * | |
9f683953 VG |
458 | * [runnable_avg definition] |
459 | * | |
460 | * runnable_avg = runnable% * SCHED_CAPACITY_SCALE | |
7b595334 | 461 | * |
7b595334 YD |
462 | * [util_avg definition] |
463 | * | |
464 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
465 | * | |
9f683953 VG |
466 | * where runnable% is the time ratio that a sched_entity is runnable and |
467 | * running% the time ratio that a sched_entity is running. | |
468 | * | |
469 | * For cfs_rq, they are the aggregated values of all runnable and blocked | |
470 | * sched_entities. | |
7b595334 | 471 | * |
c1b7b8d4 | 472 | * The load/runnable/util_avg doesn't directly factor frequency scaling and CPU |
9f683953 VG |
473 | * capacity scaling. The scaling is done through the rq_clock_pelt that is used |
474 | * for computing those signals (see update_rq_clock_pelt()) | |
7b595334 | 475 | * |
23127296 VG |
476 | * N.B., the above ratios (runnable% and running%) themselves are in the |
477 | * range of [0, 1]. To do fixed point arithmetics, we therefore scale them | |
478 | * to as large a range as necessary. This is for example reflected by | |
479 | * util_avg's SCHED_CAPACITY_SCALE. | |
7b595334 YD |
480 | * |
481 | * [Overflow issue] | |
482 | * | |
483 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
484 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
485 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
486 | * | |
487 | * For all other cases (including 32-bit kernels), struct load_weight's | |
488 | * weight will overflow first before we do, because: | |
489 | * | |
490 | * Max(load_avg) <= Max(load.weight) | |
491 | * | |
492 | * Then it is the load_weight's responsibility to consider overflow | |
493 | * issues. | |
9d89c257 | 494 | */ |
9d85f21c | 495 | struct sched_avg { |
5eca1c10 IM |
496 | u64 last_update_time; |
497 | u64 load_sum; | |
9f683953 | 498 | u64 runnable_sum; |
5eca1c10 IM |
499 | u32 util_sum; |
500 | u32 period_contrib; | |
501 | unsigned long load_avg; | |
9f683953 | 502 | unsigned long runnable_avg; |
5eca1c10 | 503 | unsigned long util_avg; |
7f65ea42 | 504 | struct util_est util_est; |
317d359d | 505 | } ____cacheline_aligned; |
9d85f21c | 506 | |
41acab88 | 507 | struct sched_statistics { |
7f5f8e8d | 508 | #ifdef CONFIG_SCHEDSTATS |
5eca1c10 IM |
509 | u64 wait_start; |
510 | u64 wait_max; | |
511 | u64 wait_count; | |
512 | u64 wait_sum; | |
513 | u64 iowait_count; | |
514 | u64 iowait_sum; | |
515 | ||
516 | u64 sleep_start; | |
517 | u64 sleep_max; | |
518 | s64 sum_sleep_runtime; | |
519 | ||
520 | u64 block_start; | |
521 | u64 block_max; | |
847fc0cd YS |
522 | s64 sum_block_runtime; |
523 | ||
5eca1c10 IM |
524 | u64 exec_max; |
525 | u64 slice_max; | |
526 | ||
527 | u64 nr_migrations_cold; | |
528 | u64 nr_failed_migrations_affine; | |
529 | u64 nr_failed_migrations_running; | |
530 | u64 nr_failed_migrations_hot; | |
531 | u64 nr_forced_migrations; | |
532 | ||
533 | u64 nr_wakeups; | |
534 | u64 nr_wakeups_sync; | |
535 | u64 nr_wakeups_migrate; | |
536 | u64 nr_wakeups_local; | |
537 | u64 nr_wakeups_remote; | |
538 | u64 nr_wakeups_affine; | |
539 | u64 nr_wakeups_affine_attempts; | |
540 | u64 nr_wakeups_passive; | |
541 | u64 nr_wakeups_idle; | |
4feee7d1 JD |
542 | |
543 | #ifdef CONFIG_SCHED_CORE | |
544 | u64 core_forceidle_sum; | |
41acab88 | 545 | #endif |
4feee7d1 | 546 | #endif /* CONFIG_SCHEDSTATS */ |
ceeadb83 | 547 | } ____cacheline_aligned; |
41acab88 LDM |
548 | |
549 | struct sched_entity { | |
5eca1c10 IM |
550 | /* For load-balancing: */ |
551 | struct load_weight load; | |
552 | struct rb_node run_node; | |
553 | struct list_head group_node; | |
554 | unsigned int on_rq; | |
41acab88 | 555 | |
5eca1c10 IM |
556 | u64 exec_start; |
557 | u64 sum_exec_runtime; | |
558 | u64 vruntime; | |
559 | u64 prev_sum_exec_runtime; | |
41acab88 | 560 | |
5eca1c10 | 561 | u64 nr_migrations; |
41acab88 | 562 | |
20b8a59f | 563 | #ifdef CONFIG_FAIR_GROUP_SCHED |
5eca1c10 IM |
564 | int depth; |
565 | struct sched_entity *parent; | |
20b8a59f | 566 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 567 | struct cfs_rq *cfs_rq; |
20b8a59f | 568 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 569 | struct cfs_rq *my_q; |
9f683953 VG |
570 | /* cached value of my_q->h_nr_running */ |
571 | unsigned long runnable_weight; | |
20b8a59f | 572 | #endif |
8bd75c77 | 573 | |
141965c7 | 574 | #ifdef CONFIG_SMP |
5a107804 JO |
575 | /* |
576 | * Per entity load average tracking. | |
577 | * | |
578 | * Put into separate cache line so it does not | |
579 | * collide with read-mostly values above. | |
580 | */ | |
317d359d | 581 | struct sched_avg avg; |
9d85f21c | 582 | #endif |
20b8a59f | 583 | }; |
70b97a7f | 584 | |
fa717060 | 585 | struct sched_rt_entity { |
5eca1c10 IM |
586 | struct list_head run_list; |
587 | unsigned long timeout; | |
588 | unsigned long watchdog_stamp; | |
589 | unsigned int time_slice; | |
590 | unsigned short on_rq; | |
591 | unsigned short on_list; | |
592 | ||
593 | struct sched_rt_entity *back; | |
052f1dc7 | 594 | #ifdef CONFIG_RT_GROUP_SCHED |
5eca1c10 | 595 | struct sched_rt_entity *parent; |
6f505b16 | 596 | /* rq on which this entity is (to be) queued: */ |
5eca1c10 | 597 | struct rt_rq *rt_rq; |
6f505b16 | 598 | /* rq "owned" by this entity/group: */ |
5eca1c10 | 599 | struct rt_rq *my_q; |
6f505b16 | 600 | #endif |
3859a271 | 601 | } __randomize_layout; |
fa717060 | 602 | |
aab03e05 | 603 | struct sched_dl_entity { |
5eca1c10 | 604 | struct rb_node rb_node; |
aab03e05 DF |
605 | |
606 | /* | |
607 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 608 | * during sched_setattr(), they will remain the same until |
609 | * the next sched_setattr(). | |
aab03e05 | 610 | */ |
5eca1c10 IM |
611 | u64 dl_runtime; /* Maximum runtime for each instance */ |
612 | u64 dl_deadline; /* Relative deadline of each instance */ | |
613 | u64 dl_period; /* Separation of two instances (period) */ | |
54d6d303 | 614 | u64 dl_bw; /* dl_runtime / dl_period */ |
3effcb42 | 615 | u64 dl_density; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
616 | |
617 | /* | |
618 | * Actual scheduling parameters. Initialized with the values above, | |
dfcb245e | 619 | * they are continuously updated during task execution. Note that |
aab03e05 DF |
620 | * the remaining runtime could be < 0 in case we are in overrun. |
621 | */ | |
5eca1c10 IM |
622 | s64 runtime; /* Remaining runtime for this instance */ |
623 | u64 deadline; /* Absolute deadline for this instance */ | |
624 | unsigned int flags; /* Specifying the scheduler behaviour */ | |
aab03e05 DF |
625 | |
626 | /* | |
627 | * Some bool flags: | |
628 | * | |
629 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
630 | * task has to wait for a replenishment to be performed at the | |
631 | * next firing of dl_timer. | |
632 | * | |
5eca1c10 | 633 | * @dl_yielded tells if task gave up the CPU before consuming |
5bfd126e | 634 | * all its available runtime during the last job. |
209a0cbd LA |
635 | * |
636 | * @dl_non_contending tells if the task is inactive while still | |
637 | * contributing to the active utilization. In other words, it | |
638 | * indicates if the inactive timer has been armed and its handler | |
639 | * has not been executed yet. This flag is useful to avoid race | |
640 | * conditions between the inactive timer handler and the wakeup | |
641 | * code. | |
34be3930 JL |
642 | * |
643 | * @dl_overrun tells if the task asked to be informed about runtime | |
644 | * overruns. | |
aab03e05 | 645 | */ |
aa5222e9 | 646 | unsigned int dl_throttled : 1; |
aa5222e9 DC |
647 | unsigned int dl_yielded : 1; |
648 | unsigned int dl_non_contending : 1; | |
34be3930 | 649 | unsigned int dl_overrun : 1; |
aab03e05 DF |
650 | |
651 | /* | |
652 | * Bandwidth enforcement timer. Each -deadline task has its | |
653 | * own bandwidth to be enforced, thus we need one timer per task. | |
654 | */ | |
5eca1c10 | 655 | struct hrtimer dl_timer; |
209a0cbd LA |
656 | |
657 | /* | |
658 | * Inactive timer, responsible for decreasing the active utilization | |
659 | * at the "0-lag time". When a -deadline task blocks, it contributes | |
660 | * to GRUB's active utilization until the "0-lag time", hence a | |
661 | * timer is needed to decrease the active utilization at the correct | |
662 | * time. | |
663 | */ | |
664 | struct hrtimer inactive_timer; | |
2279f540 JL |
665 | |
666 | #ifdef CONFIG_RT_MUTEXES | |
667 | /* | |
668 | * Priority Inheritance. When a DEADLINE scheduling entity is boosted | |
669 | * pi_se points to the donor, otherwise points to the dl_se it belongs | |
670 | * to (the original one/itself). | |
671 | */ | |
672 | struct sched_dl_entity *pi_se; | |
673 | #endif | |
aab03e05 | 674 | }; |
8bd75c77 | 675 | |
69842cba PB |
676 | #ifdef CONFIG_UCLAMP_TASK |
677 | /* Number of utilization clamp buckets (shorter alias) */ | |
678 | #define UCLAMP_BUCKETS CONFIG_UCLAMP_BUCKETS_COUNT | |
679 | ||
680 | /* | |
681 | * Utilization clamp for a scheduling entity | |
682 | * @value: clamp value "assigned" to a se | |
683 | * @bucket_id: bucket index corresponding to the "assigned" value | |
e8f14172 | 684 | * @active: the se is currently refcounted in a rq's bucket |
a509a7cd | 685 | * @user_defined: the requested clamp value comes from user-space |
69842cba PB |
686 | * |
687 | * The bucket_id is the index of the clamp bucket matching the clamp value | |
688 | * which is pre-computed and stored to avoid expensive integer divisions from | |
689 | * the fast path. | |
e8f14172 PB |
690 | * |
691 | * The active bit is set whenever a task has got an "effective" value assigned, | |
692 | * which can be different from the clamp value "requested" from user-space. | |
693 | * This allows to know a task is refcounted in the rq's bucket corresponding | |
694 | * to the "effective" bucket_id. | |
a509a7cd PB |
695 | * |
696 | * The user_defined bit is set whenever a task has got a task-specific clamp | |
697 | * value requested from userspace, i.e. the system defaults apply to this task | |
698 | * just as a restriction. This allows to relax default clamps when a less | |
699 | * restrictive task-specific value has been requested, thus allowing to | |
700 | * implement a "nice" semantic. For example, a task running with a 20% | |
701 | * default boost can still drop its own boosting to 0%. | |
69842cba PB |
702 | */ |
703 | struct uclamp_se { | |
704 | unsigned int value : bits_per(SCHED_CAPACITY_SCALE); | |
705 | unsigned int bucket_id : bits_per(UCLAMP_BUCKETS); | |
e8f14172 | 706 | unsigned int active : 1; |
a509a7cd | 707 | unsigned int user_defined : 1; |
69842cba PB |
708 | }; |
709 | #endif /* CONFIG_UCLAMP_TASK */ | |
710 | ||
1d082fd0 PM |
711 | union rcu_special { |
712 | struct { | |
5eca1c10 IM |
713 | u8 blocked; |
714 | u8 need_qs; | |
05f41571 | 715 | u8 exp_hint; /* Hint for performance. */ |
276c4104 | 716 | u8 need_mb; /* Readers need smp_mb(). */ |
8203d6d0 | 717 | } b; /* Bits. */ |
05f41571 | 718 | u32 s; /* Set of bits. */ |
1d082fd0 | 719 | }; |
86848966 | 720 | |
8dc85d54 PZ |
721 | enum perf_event_task_context { |
722 | perf_invalid_context = -1, | |
723 | perf_hw_context = 0, | |
89a1e187 | 724 | perf_sw_context, |
8dc85d54 PZ |
725 | perf_nr_task_contexts, |
726 | }; | |
727 | ||
eb61baf6 IM |
728 | struct wake_q_node { |
729 | struct wake_q_node *next; | |
730 | }; | |
731 | ||
5fbda3ec TG |
732 | struct kmap_ctrl { |
733 | #ifdef CONFIG_KMAP_LOCAL | |
734 | int idx; | |
735 | pte_t pteval[KM_MAX_IDX]; | |
736 | #endif | |
737 | }; | |
738 | ||
1da177e4 | 739 | struct task_struct { |
c65eacbe AL |
740 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
741 | /* | |
742 | * For reasons of header soup (see current_thread_info()), this | |
743 | * must be the first element of task_struct. | |
744 | */ | |
5eca1c10 | 745 | struct thread_info thread_info; |
c65eacbe | 746 | #endif |
2f064a59 | 747 | unsigned int __state; |
29e48ce8 | 748 | |
5f220be2 TG |
749 | #ifdef CONFIG_PREEMPT_RT |
750 | /* saved state for "spinlock sleepers" */ | |
751 | unsigned int saved_state; | |
752 | #endif | |
753 | ||
29e48ce8 KC |
754 | /* |
755 | * This begins the randomizable portion of task_struct. Only | |
756 | * scheduling-critical items should be added above here. | |
757 | */ | |
758 | randomized_struct_fields_start | |
759 | ||
5eca1c10 | 760 | void *stack; |
ec1d2819 | 761 | refcount_t usage; |
5eca1c10 IM |
762 | /* Per task flags (PF_*), defined further below: */ |
763 | unsigned int flags; | |
764 | unsigned int ptrace; | |
1da177e4 | 765 | |
2dd73a4f | 766 | #ifdef CONFIG_SMP |
5eca1c10 | 767 | int on_cpu; |
8c4890d1 | 768 | struct __call_single_node wake_entry; |
5eca1c10 IM |
769 | unsigned int wakee_flips; |
770 | unsigned long wakee_flip_decay_ts; | |
771 | struct task_struct *last_wakee; | |
ac66f547 | 772 | |
32e839dd MG |
773 | /* |
774 | * recent_used_cpu is initially set as the last CPU used by a task | |
775 | * that wakes affine another task. Waker/wakee relationships can | |
776 | * push tasks around a CPU where each wakeup moves to the next one. | |
777 | * Tracking a recently used CPU allows a quick search for a recently | |
778 | * used CPU that may be idle. | |
779 | */ | |
780 | int recent_used_cpu; | |
5eca1c10 | 781 | int wake_cpu; |
2dd73a4f | 782 | #endif |
5eca1c10 IM |
783 | int on_rq; |
784 | ||
785 | int prio; | |
786 | int static_prio; | |
787 | int normal_prio; | |
788 | unsigned int rt_priority; | |
50e645a8 | 789 | |
5eca1c10 IM |
790 | struct sched_entity se; |
791 | struct sched_rt_entity rt; | |
8a311c74 | 792 | struct sched_dl_entity dl; |
804bccba | 793 | const struct sched_class *sched_class; |
8a311c74 PZ |
794 | |
795 | #ifdef CONFIG_SCHED_CORE | |
796 | struct rb_node core_node; | |
797 | unsigned long core_cookie; | |
d2dfa17b | 798 | unsigned int core_occupation; |
8a311c74 PZ |
799 | #endif |
800 | ||
8323f26c | 801 | #ifdef CONFIG_CGROUP_SCHED |
5eca1c10 | 802 | struct task_group *sched_task_group; |
8323f26c | 803 | #endif |
1da177e4 | 804 | |
69842cba | 805 | #ifdef CONFIG_UCLAMP_TASK |
13685c4a QY |
806 | /* |
807 | * Clamp values requested for a scheduling entity. | |
808 | * Must be updated with task_rq_lock() held. | |
809 | */ | |
e8f14172 | 810 | struct uclamp_se uclamp_req[UCLAMP_CNT]; |
13685c4a QY |
811 | /* |
812 | * Effective clamp values used for a scheduling entity. | |
813 | * Must be updated with task_rq_lock() held. | |
814 | */ | |
69842cba PB |
815 | struct uclamp_se uclamp[UCLAMP_CNT]; |
816 | #endif | |
817 | ||
ceeadb83 YS |
818 | struct sched_statistics stats; |
819 | ||
e107be36 | 820 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
5eca1c10 IM |
821 | /* List of struct preempt_notifier: */ |
822 | struct hlist_head preempt_notifiers; | |
e107be36 AK |
823 | #endif |
824 | ||
6c5c9341 | 825 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
5eca1c10 | 826 | unsigned int btrace_seq; |
6c5c9341 | 827 | #endif |
1da177e4 | 828 | |
5eca1c10 IM |
829 | unsigned int policy; |
830 | int nr_cpus_allowed; | |
3bd37062 | 831 | const cpumask_t *cpus_ptr; |
b90ca8ba | 832 | cpumask_t *user_cpus_ptr; |
3bd37062 | 833 | cpumask_t cpus_mask; |
6d337eab | 834 | void *migration_pending; |
74d862b6 | 835 | #ifdef CONFIG_SMP |
a7c81556 | 836 | unsigned short migration_disabled; |
af449901 | 837 | #endif |
a7c81556 | 838 | unsigned short migration_flags; |
1da177e4 | 839 | |
a57eb940 | 840 | #ifdef CONFIG_PREEMPT_RCU |
5eca1c10 IM |
841 | int rcu_read_lock_nesting; |
842 | union rcu_special rcu_read_unlock_special; | |
843 | struct list_head rcu_node_entry; | |
844 | struct rcu_node *rcu_blocked_node; | |
28f6569a | 845 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
5eca1c10 | 846 | |
8315f422 | 847 | #ifdef CONFIG_TASKS_RCU |
5eca1c10 | 848 | unsigned long rcu_tasks_nvcsw; |
ccdd29ff PM |
849 | u8 rcu_tasks_holdout; |
850 | u8 rcu_tasks_idx; | |
5eca1c10 | 851 | int rcu_tasks_idle_cpu; |
ccdd29ff | 852 | struct list_head rcu_tasks_holdout_list; |
8315f422 | 853 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 854 | |
d5f177d3 PM |
855 | #ifdef CONFIG_TASKS_TRACE_RCU |
856 | int trc_reader_nesting; | |
857 | int trc_ipi_to_cpu; | |
276c4104 | 858 | union rcu_special trc_reader_special; |
d5f177d3 | 859 | struct list_head trc_holdout_list; |
434c9eef PM |
860 | struct list_head trc_blkd_node; |
861 | int trc_blkd_cpu; | |
d5f177d3 PM |
862 | #endif /* #ifdef CONFIG_TASKS_TRACE_RCU */ |
863 | ||
5eca1c10 | 864 | struct sched_info sched_info; |
1da177e4 | 865 | |
5eca1c10 | 866 | struct list_head tasks; |
806c09a7 | 867 | #ifdef CONFIG_SMP |
5eca1c10 IM |
868 | struct plist_node pushable_tasks; |
869 | struct rb_node pushable_dl_tasks; | |
806c09a7 | 870 | #endif |
1da177e4 | 871 | |
5eca1c10 IM |
872 | struct mm_struct *mm; |
873 | struct mm_struct *active_mm; | |
314ff785 | 874 | |
5eca1c10 IM |
875 | int exit_state; |
876 | int exit_code; | |
877 | int exit_signal; | |
878 | /* The signal sent when the parent dies: */ | |
879 | int pdeath_signal; | |
880 | /* JOBCTL_*, siglock protected: */ | |
881 | unsigned long jobctl; | |
882 | ||
883 | /* Used for emulating ABI behavior of previous Linux versions: */ | |
884 | unsigned int personality; | |
885 | ||
886 | /* Scheduler bits, serialized by scheduler locks: */ | |
887 | unsigned sched_reset_on_fork:1; | |
888 | unsigned sched_contributes_to_load:1; | |
889 | unsigned sched_migrated:1; | |
eb414681 | 890 | |
5eca1c10 IM |
891 | /* Force alignment to the next boundary: */ |
892 | unsigned :0; | |
893 | ||
894 | /* Unserialized, strictly 'current' */ | |
895 | ||
f97bb527 PZ |
896 | /* |
897 | * This field must not be in the scheduler word above due to wakelist | |
898 | * queueing no longer being serialized by p->on_cpu. However: | |
899 | * | |
900 | * p->XXX = X; ttwu() | |
901 | * schedule() if (p->on_rq && ..) // false | |
902 | * smp_mb__after_spinlock(); if (smp_load_acquire(&p->on_cpu) && //true | |
903 | * deactivate_task() ttwu_queue_wakelist()) | |
904 | * p->on_rq = 0; p->sched_remote_wakeup = Y; | |
905 | * | |
906 | * guarantees all stores of 'current' are visible before | |
907 | * ->sched_remote_wakeup gets used, so it can be in this word. | |
908 | */ | |
909 | unsigned sched_remote_wakeup:1; | |
910 | ||
5eca1c10 IM |
911 | /* Bit to tell LSMs we're in execve(): */ |
912 | unsigned in_execve:1; | |
913 | unsigned in_iowait:1; | |
914 | #ifndef TIF_RESTORE_SIGMASK | |
915 | unsigned restore_sigmask:1; | |
7e781418 | 916 | #endif |
626ebc41 | 917 | #ifdef CONFIG_MEMCG |
29ef680a | 918 | unsigned in_user_fault:1; |
127424c8 | 919 | #endif |
ec1c86b2 YZ |
920 | #ifdef CONFIG_LRU_GEN |
921 | /* whether the LRU algorithm may apply to this access */ | |
922 | unsigned in_lru_fault:1; | |
923 | #endif | |
ff303e66 | 924 | #ifdef CONFIG_COMPAT_BRK |
5eca1c10 | 925 | unsigned brk_randomized:1; |
ff303e66 | 926 | #endif |
77f88796 TH |
927 | #ifdef CONFIG_CGROUPS |
928 | /* disallow userland-initiated cgroup migration */ | |
929 | unsigned no_cgroup_migration:1; | |
76f969e8 RG |
930 | /* task is frozen/stopped (used by the cgroup freezer) */ |
931 | unsigned frozen:1; | |
77f88796 | 932 | #endif |
d09d8df3 | 933 | #ifdef CONFIG_BLK_CGROUP |
d09d8df3 JB |
934 | unsigned use_memdelay:1; |
935 | #endif | |
1066d1b6 YS |
936 | #ifdef CONFIG_PSI |
937 | /* Stalled due to lack of memory */ | |
938 | unsigned in_memstall:1; | |
939 | #endif | |
8e9b16c4 ST |
940 | #ifdef CONFIG_PAGE_OWNER |
941 | /* Used by page_owner=on to detect recursion in page tracking. */ | |
942 | unsigned in_page_owner:1; | |
943 | #endif | |
b542e383 TG |
944 | #ifdef CONFIG_EVENTFD |
945 | /* Recursion prevention for eventfd_signal() */ | |
9f0deaa1 | 946 | unsigned in_eventfd:1; |
b542e383 | 947 | #endif |
a3d29e82 PZ |
948 | #ifdef CONFIG_IOMMU_SVA |
949 | unsigned pasid_activated:1; | |
950 | #endif | |
b041b525 TL |
951 | #ifdef CONFIG_CPU_SUP_INTEL |
952 | unsigned reported_split_lock:1; | |
953 | #endif | |
aa1cf99b YY |
954 | #ifdef CONFIG_TASK_DELAY_ACCT |
955 | /* delay due to memory thrashing */ | |
956 | unsigned in_thrashing:1; | |
957 | #endif | |
6f185c29 | 958 | |
5eca1c10 | 959 | unsigned long atomic_flags; /* Flags requiring atomic access. */ |
1d4457f9 | 960 | |
5eca1c10 | 961 | struct restart_block restart_block; |
f56141e3 | 962 | |
5eca1c10 IM |
963 | pid_t pid; |
964 | pid_t tgid; | |
0a425405 | 965 | |
050e9baa | 966 | #ifdef CONFIG_STACKPROTECTOR |
5eca1c10 IM |
967 | /* Canary value for the -fstack-protector GCC feature: */ |
968 | unsigned long stack_canary; | |
1314562a | 969 | #endif |
4d1d61a6 | 970 | /* |
5eca1c10 | 971 | * Pointers to the (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 972 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 973 | * p->real_parent->pid) |
1da177e4 | 974 | */ |
5eca1c10 IM |
975 | |
976 | /* Real parent process: */ | |
977 | struct task_struct __rcu *real_parent; | |
978 | ||
979 | /* Recipient of SIGCHLD, wait4() reports: */ | |
980 | struct task_struct __rcu *parent; | |
981 | ||
1da177e4 | 982 | /* |
5eca1c10 | 983 | * Children/sibling form the list of natural children: |
1da177e4 | 984 | */ |
5eca1c10 IM |
985 | struct list_head children; |
986 | struct list_head sibling; | |
987 | struct task_struct *group_leader; | |
1da177e4 | 988 | |
f470021a | 989 | /* |
5eca1c10 IM |
990 | * 'ptraced' is the list of tasks this task is using ptrace() on. |
991 | * | |
f470021a | 992 | * This includes both natural children and PTRACE_ATTACH targets. |
5eca1c10 | 993 | * 'ptrace_entry' is this task's link on the p->parent->ptraced list. |
f470021a | 994 | */ |
5eca1c10 IM |
995 | struct list_head ptraced; |
996 | struct list_head ptrace_entry; | |
f470021a | 997 | |
1da177e4 | 998 | /* PID/PID hash table linkage. */ |
2c470475 EB |
999 | struct pid *thread_pid; |
1000 | struct hlist_node pid_links[PIDTYPE_MAX]; | |
5eca1c10 IM |
1001 | struct list_head thread_group; |
1002 | struct list_head thread_node; | |
1003 | ||
1004 | struct completion *vfork_done; | |
1da177e4 | 1005 | |
5eca1c10 IM |
1006 | /* CLONE_CHILD_SETTID: */ |
1007 | int __user *set_child_tid; | |
1da177e4 | 1008 | |
5eca1c10 IM |
1009 | /* CLONE_CHILD_CLEARTID: */ |
1010 | int __user *clear_child_tid; | |
1011 | ||
e32cf5df EB |
1012 | /* PF_KTHREAD | PF_IO_WORKER */ |
1013 | void *worker_private; | |
3bfe6106 | 1014 | |
5eca1c10 IM |
1015 | u64 utime; |
1016 | u64 stime; | |
40565b5a | 1017 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME |
5eca1c10 IM |
1018 | u64 utimescaled; |
1019 | u64 stimescaled; | |
40565b5a | 1020 | #endif |
5eca1c10 IM |
1021 | u64 gtime; |
1022 | struct prev_cputime prev_cputime; | |
6a61671b | 1023 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
bac5b6b6 | 1024 | struct vtime vtime; |
d99ca3b9 | 1025 | #endif |
d027d45d FW |
1026 | |
1027 | #ifdef CONFIG_NO_HZ_FULL | |
5eca1c10 | 1028 | atomic_t tick_dep_mask; |
d027d45d | 1029 | #endif |
5eca1c10 IM |
1030 | /* Context switch counts: */ |
1031 | unsigned long nvcsw; | |
1032 | unsigned long nivcsw; | |
1033 | ||
1034 | /* Monotonic time in nsecs: */ | |
1035 | u64 start_time; | |
1036 | ||
1037 | /* Boot based time in nsecs: */ | |
cf25e24d | 1038 | u64 start_boottime; |
5eca1c10 IM |
1039 | |
1040 | /* MM fault and swap info: this can arguably be seen as either mm-specific or thread-specific: */ | |
1041 | unsigned long min_flt; | |
1042 | unsigned long maj_flt; | |
1da177e4 | 1043 | |
2b69942f TG |
1044 | /* Empty if CONFIG_POSIX_CPUTIMERS=n */ |
1045 | struct posix_cputimers posix_cputimers; | |
1da177e4 | 1046 | |
1fb497dd TG |
1047 | #ifdef CONFIG_POSIX_CPU_TIMERS_TASK_WORK |
1048 | struct posix_cputimers_work posix_cputimers_work; | |
1049 | #endif | |
1050 | ||
5eca1c10 IM |
1051 | /* Process credentials: */ |
1052 | ||
1053 | /* Tracer's credentials at attach: */ | |
1054 | const struct cred __rcu *ptracer_cred; | |
1055 | ||
1056 | /* Objective and real subjective task credentials (COW): */ | |
1057 | const struct cred __rcu *real_cred; | |
1058 | ||
1059 | /* Effective (overridable) subjective task credentials (COW): */ | |
1060 | const struct cred __rcu *cred; | |
1061 | ||
7743c48e DH |
1062 | #ifdef CONFIG_KEYS |
1063 | /* Cached requested key. */ | |
1064 | struct key *cached_requested_key; | |
1065 | #endif | |
1066 | ||
5eca1c10 IM |
1067 | /* |
1068 | * executable name, excluding path. | |
1069 | * | |
1070 | * - normally initialized setup_new_exec() | |
1071 | * - access it with [gs]et_task_comm() | |
1072 | * - lock it with task_lock() | |
1073 | */ | |
1074 | char comm[TASK_COMM_LEN]; | |
1075 | ||
1076 | struct nameidata *nameidata; | |
1077 | ||
3d5b6fcc | 1078 | #ifdef CONFIG_SYSVIPC |
5eca1c10 IM |
1079 | struct sysv_sem sysvsem; |
1080 | struct sysv_shm sysvshm; | |
3d5b6fcc | 1081 | #endif |
e162b39a | 1082 | #ifdef CONFIG_DETECT_HUNG_TASK |
5eca1c10 | 1083 | unsigned long last_switch_count; |
a2e51445 | 1084 | unsigned long last_switch_time; |
82a1fcb9 | 1085 | #endif |
5eca1c10 IM |
1086 | /* Filesystem information: */ |
1087 | struct fs_struct *fs; | |
1088 | ||
1089 | /* Open file information: */ | |
1090 | struct files_struct *files; | |
1091 | ||
0f212204 JA |
1092 | #ifdef CONFIG_IO_URING |
1093 | struct io_uring_task *io_uring; | |
1094 | #endif | |
1095 | ||
5eca1c10 IM |
1096 | /* Namespaces: */ |
1097 | struct nsproxy *nsproxy; | |
1098 | ||
1099 | /* Signal handlers: */ | |
1100 | struct signal_struct *signal; | |
913292c9 | 1101 | struct sighand_struct __rcu *sighand; |
5eca1c10 IM |
1102 | sigset_t blocked; |
1103 | sigset_t real_blocked; | |
1104 | /* Restored if set_restore_sigmask() was used: */ | |
1105 | sigset_t saved_sigmask; | |
1106 | struct sigpending pending; | |
1107 | unsigned long sas_ss_sp; | |
1108 | size_t sas_ss_size; | |
1109 | unsigned int sas_ss_flags; | |
1110 | ||
1111 | struct callback_head *task_works; | |
1112 | ||
4b7d248b | 1113 | #ifdef CONFIG_AUDIT |
bfef93a5 | 1114 | #ifdef CONFIG_AUDITSYSCALL |
5f3d544f RGB |
1115 | struct audit_context *audit_context; |
1116 | #endif | |
5eca1c10 IM |
1117 | kuid_t loginuid; |
1118 | unsigned int sessionid; | |
bfef93a5 | 1119 | #endif |
5eca1c10 | 1120 | struct seccomp seccomp; |
1446e1df | 1121 | struct syscall_user_dispatch syscall_dispatch; |
5eca1c10 IM |
1122 | |
1123 | /* Thread group tracking: */ | |
d1e7fd64 EB |
1124 | u64 parent_exec_id; |
1125 | u64 self_exec_id; | |
1da177e4 | 1126 | |
5eca1c10 IM |
1127 | /* Protection against (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, mempolicy: */ |
1128 | spinlock_t alloc_lock; | |
1da177e4 | 1129 | |
b29739f9 | 1130 | /* Protection of the PI data structures: */ |
5eca1c10 | 1131 | raw_spinlock_t pi_lock; |
b29739f9 | 1132 | |
5eca1c10 | 1133 | struct wake_q_node wake_q; |
76751049 | 1134 | |
23f78d4a | 1135 | #ifdef CONFIG_RT_MUTEXES |
5eca1c10 | 1136 | /* PI waiters blocked on a rt_mutex held by this task: */ |
a23ba907 | 1137 | struct rb_root_cached pi_waiters; |
e96a7705 XP |
1138 | /* Updated under owner's pi_lock and rq lock */ |
1139 | struct task_struct *pi_top_task; | |
5eca1c10 IM |
1140 | /* Deadlock detection and priority inheritance handling: */ |
1141 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
1142 | #endif |
1143 | ||
408894ee | 1144 | #ifdef CONFIG_DEBUG_MUTEXES |
5eca1c10 IM |
1145 | /* Mutex deadlock detection: */ |
1146 | struct mutex_waiter *blocked_on; | |
408894ee | 1147 | #endif |
5eca1c10 | 1148 | |
312364f3 DV |
1149 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
1150 | int non_block_count; | |
1151 | #endif | |
1152 | ||
de30a2b3 | 1153 | #ifdef CONFIG_TRACE_IRQFLAGS |
0584df9c | 1154 | struct irqtrace_events irqtrace; |
de8f5e4f | 1155 | unsigned int hardirq_threaded; |
c86e9b98 | 1156 | u64 hardirq_chain_key; |
5eca1c10 IM |
1157 | int softirqs_enabled; |
1158 | int softirq_context; | |
40db1739 | 1159 | int irq_config; |
de30a2b3 | 1160 | #endif |
728b478d TG |
1161 | #ifdef CONFIG_PREEMPT_RT |
1162 | int softirq_disable_cnt; | |
1163 | #endif | |
5eca1c10 | 1164 | |
fbb9ce95 | 1165 | #ifdef CONFIG_LOCKDEP |
5eca1c10 IM |
1166 | # define MAX_LOCK_DEPTH 48UL |
1167 | u64 curr_chain_key; | |
1168 | int lockdep_depth; | |
1169 | unsigned int lockdep_recursion; | |
1170 | struct held_lock held_locks[MAX_LOCK_DEPTH]; | |
fbb9ce95 | 1171 | #endif |
5eca1c10 | 1172 | |
5cf53f3c | 1173 | #if defined(CONFIG_UBSAN) && !defined(CONFIG_UBSAN_TRAP) |
5eca1c10 | 1174 | unsigned int in_ubsan; |
c6d30853 | 1175 | #endif |
408894ee | 1176 | |
5eca1c10 IM |
1177 | /* Journalling filesystem info: */ |
1178 | void *journal_info; | |
1da177e4 | 1179 | |
5eca1c10 IM |
1180 | /* Stacked block device info: */ |
1181 | struct bio_list *bio_list; | |
d89d8796 | 1182 | |
5eca1c10 IM |
1183 | /* Stack plugging: */ |
1184 | struct blk_plug *plug; | |
73c10101 | 1185 | |
5eca1c10 IM |
1186 | /* VM state: */ |
1187 | struct reclaim_state *reclaim_state; | |
1188 | ||
1189 | struct backing_dev_info *backing_dev_info; | |
1da177e4 | 1190 | |
5eca1c10 | 1191 | struct io_context *io_context; |
1da177e4 | 1192 | |
5e1f0f09 MG |
1193 | #ifdef CONFIG_COMPACTION |
1194 | struct capture_control *capture_control; | |
1195 | #endif | |
5eca1c10 IM |
1196 | /* Ptrace state: */ |
1197 | unsigned long ptrace_message; | |
ae7795bc | 1198 | kernel_siginfo_t *last_siginfo; |
1da177e4 | 1199 | |
5eca1c10 | 1200 | struct task_io_accounting ioac; |
eb414681 JW |
1201 | #ifdef CONFIG_PSI |
1202 | /* Pressure stall state */ | |
1203 | unsigned int psi_flags; | |
1204 | #endif | |
5eca1c10 IM |
1205 | #ifdef CONFIG_TASK_XACCT |
1206 | /* Accumulated RSS usage: */ | |
1207 | u64 acct_rss_mem1; | |
1208 | /* Accumulated virtual memory usage: */ | |
1209 | u64 acct_vm_mem1; | |
1210 | /* stime + utime since last update: */ | |
1211 | u64 acct_timexpd; | |
1da177e4 LT |
1212 | #endif |
1213 | #ifdef CONFIG_CPUSETS | |
5eca1c10 IM |
1214 | /* Protected by ->alloc_lock: */ |
1215 | nodemask_t mems_allowed; | |
3b03706f | 1216 | /* Sequence number to catch updates: */ |
b7505861 | 1217 | seqcount_spinlock_t mems_allowed_seq; |
5eca1c10 IM |
1218 | int cpuset_mem_spread_rotor; |
1219 | int cpuset_slab_spread_rotor; | |
1da177e4 | 1220 | #endif |
ddbcc7e8 | 1221 | #ifdef CONFIG_CGROUPS |
5eca1c10 IM |
1222 | /* Control Group info protected by css_set_lock: */ |
1223 | struct css_set __rcu *cgroups; | |
1224 | /* cg_list protected by css_set_lock and tsk->alloc_lock: */ | |
1225 | struct list_head cg_list; | |
ddbcc7e8 | 1226 | #endif |
e6d42931 | 1227 | #ifdef CONFIG_X86_CPU_RESCTRL |
0734ded1 | 1228 | u32 closid; |
d6aaba61 | 1229 | u32 rmid; |
e02737d5 | 1230 | #endif |
42b2dd0a | 1231 | #ifdef CONFIG_FUTEX |
5eca1c10 | 1232 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
1233 | #ifdef CONFIG_COMPAT |
1234 | struct compat_robust_list_head __user *compat_robust_list; | |
1235 | #endif | |
5eca1c10 IM |
1236 | struct list_head pi_state_list; |
1237 | struct futex_pi_state *pi_state_cache; | |
3f186d97 | 1238 | struct mutex futex_exit_mutex; |
3d4775df | 1239 | unsigned int futex_state; |
c7aceaba | 1240 | #endif |
cdd6c482 | 1241 | #ifdef CONFIG_PERF_EVENTS |
bd275681 | 1242 | struct perf_event_context *perf_event_ctxp; |
5eca1c10 IM |
1243 | struct mutex perf_event_mutex; |
1244 | struct list_head perf_event_list; | |
a63eaf34 | 1245 | #endif |
8f47b187 | 1246 | #ifdef CONFIG_DEBUG_PREEMPT |
5eca1c10 | 1247 | unsigned long preempt_disable_ip; |
8f47b187 | 1248 | #endif |
c7aceaba | 1249 | #ifdef CONFIG_NUMA |
5eca1c10 IM |
1250 | /* Protected by alloc_lock: */ |
1251 | struct mempolicy *mempolicy; | |
45816682 | 1252 | short il_prev; |
5eca1c10 | 1253 | short pref_node_fork; |
42b2dd0a | 1254 | #endif |
cbee9f88 | 1255 | #ifdef CONFIG_NUMA_BALANCING |
5eca1c10 IM |
1256 | int numa_scan_seq; |
1257 | unsigned int numa_scan_period; | |
1258 | unsigned int numa_scan_period_max; | |
1259 | int numa_preferred_nid; | |
1260 | unsigned long numa_migrate_retry; | |
1261 | /* Migration stamp: */ | |
1262 | u64 node_stamp; | |
1263 | u64 last_task_numa_placement; | |
1264 | u64 last_sum_exec_runtime; | |
1265 | struct callback_head numa_work; | |
1266 | ||
cb361d8c JH |
1267 | /* |
1268 | * This pointer is only modified for current in syscall and | |
1269 | * pagefault context (and for tasks being destroyed), so it can be read | |
1270 | * from any of the following contexts: | |
1271 | * - RCU read-side critical section | |
1272 | * - current->numa_group from everywhere | |
1273 | * - task's runqueue locked, task not running | |
1274 | */ | |
1275 | struct numa_group __rcu *numa_group; | |
8c8a743c | 1276 | |
745d6147 | 1277 | /* |
44dba3d5 IM |
1278 | * numa_faults is an array split into four regions: |
1279 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
1280 | * in this precise order. | |
1281 | * | |
1282 | * faults_memory: Exponential decaying average of faults on a per-node | |
1283 | * basis. Scheduling placement decisions are made based on these | |
1284 | * counts. The values remain static for the duration of a PTE scan. | |
1285 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
1286 | * hinting fault was incurred. | |
1287 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
1288 | * during the current scan window. When the scan completes, the counts | |
1289 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 1290 | */ |
5eca1c10 IM |
1291 | unsigned long *numa_faults; |
1292 | unsigned long total_numa_faults; | |
745d6147 | 1293 | |
04bb2f94 RR |
1294 | /* |
1295 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
1296 | * scan window were remote/local or failed to migrate. The task scan |
1297 | * period is adapted based on the locality of the faults with different | |
1298 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 1299 | */ |
5eca1c10 | 1300 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 1301 | |
5eca1c10 | 1302 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
1303 | #endif /* CONFIG_NUMA_BALANCING */ |
1304 | ||
d7822b1e MD |
1305 | #ifdef CONFIG_RSEQ |
1306 | struct rseq __user *rseq; | |
ee3e3ac0 | 1307 | u32 rseq_len; |
d7822b1e MD |
1308 | u32 rseq_sig; |
1309 | /* | |
1310 | * RmW on rseq_event_mask must be performed atomically | |
1311 | * with respect to preemption. | |
1312 | */ | |
1313 | unsigned long rseq_event_mask; | |
1314 | #endif | |
1315 | ||
af7f588d MD |
1316 | #ifdef CONFIG_SCHED_MM_CID |
1317 | int mm_cid; /* Current cid in mm */ | |
223baf9d MD |
1318 | int last_mm_cid; /* Most recent cid in mm */ |
1319 | int migrate_from_cpu; | |
af7f588d | 1320 | int mm_cid_active; /* Whether cid bitmap is active */ |
223baf9d | 1321 | struct callback_head cid_work; |
af7f588d MD |
1322 | #endif |
1323 | ||
5eca1c10 | 1324 | struct tlbflush_unmap_batch tlb_ubc; |
72b252ae | 1325 | |
5eca1c10 IM |
1326 | /* Cache last used pipe for splice(): */ |
1327 | struct pipe_inode_info *splice_pipe; | |
5640f768 | 1328 | |
5eca1c10 | 1329 | struct page_frag task_frag; |
5640f768 | 1330 | |
47913d4e IM |
1331 | #ifdef CONFIG_TASK_DELAY_ACCT |
1332 | struct task_delay_info *delays; | |
f4f154fd | 1333 | #endif |
47913d4e | 1334 | |
f4f154fd | 1335 | #ifdef CONFIG_FAULT_INJECTION |
5eca1c10 | 1336 | int make_it_fail; |
9049f2f6 | 1337 | unsigned int fail_nth; |
ca74e92b | 1338 | #endif |
9d823e8f | 1339 | /* |
5eca1c10 IM |
1340 | * When (nr_dirtied >= nr_dirtied_pause), it's time to call |
1341 | * balance_dirty_pages() for a dirty throttling pause: | |
9d823e8f | 1342 | */ |
5eca1c10 IM |
1343 | int nr_dirtied; |
1344 | int nr_dirtied_pause; | |
1345 | /* Start of a write-and-pause period: */ | |
1346 | unsigned long dirty_paused_when; | |
9d823e8f | 1347 | |
9745512c | 1348 | #ifdef CONFIG_LATENCYTOP |
5eca1c10 IM |
1349 | int latency_record_count; |
1350 | struct latency_record latency_record[LT_SAVECOUNT]; | |
9745512c | 1351 | #endif |
6976675d | 1352 | /* |
5eca1c10 | 1353 | * Time slack values; these are used to round up poll() and |
6976675d AV |
1354 | * select() etc timeout values. These are in nanoseconds. |
1355 | */ | |
5eca1c10 IM |
1356 | u64 timer_slack_ns; |
1357 | u64 default_timer_slack_ns; | |
f8d570a4 | 1358 | |
d73b4936 | 1359 | #if defined(CONFIG_KASAN_GENERIC) || defined(CONFIG_KASAN_SW_TAGS) |
5eca1c10 | 1360 | unsigned int kasan_depth; |
0b24becc | 1361 | #endif |
92c209ac | 1362 | |
dfd402a4 ME |
1363 | #ifdef CONFIG_KCSAN |
1364 | struct kcsan_ctx kcsan_ctx; | |
92c209ac ME |
1365 | #ifdef CONFIG_TRACE_IRQFLAGS |
1366 | struct irqtrace_events kcsan_save_irqtrace; | |
1367 | #endif | |
69562e49 ME |
1368 | #ifdef CONFIG_KCSAN_WEAK_MEMORY |
1369 | int kcsan_stack_depth; | |
1370 | #endif | |
dfd402a4 | 1371 | #endif |
5eca1c10 | 1372 | |
f80be457 AP |
1373 | #ifdef CONFIG_KMSAN |
1374 | struct kmsan_ctx kmsan_ctx; | |
1375 | #endif | |
1376 | ||
393824f6 PA |
1377 | #if IS_ENABLED(CONFIG_KUNIT) |
1378 | struct kunit *kunit_test; | |
1379 | #endif | |
1380 | ||
fb52607a | 1381 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
5eca1c10 IM |
1382 | /* Index of current stored address in ret_stack: */ |
1383 | int curr_ret_stack; | |
39eb456d | 1384 | int curr_ret_depth; |
5eca1c10 IM |
1385 | |
1386 | /* Stack of return addresses for return function tracing: */ | |
1387 | struct ftrace_ret_stack *ret_stack; | |
1388 | ||
1389 | /* Timestamp for last schedule: */ | |
1390 | unsigned long long ftrace_timestamp; | |
1391 | ||
f201ae23 FW |
1392 | /* |
1393 | * Number of functions that haven't been traced | |
5eca1c10 | 1394 | * because of depth overrun: |
f201ae23 | 1395 | */ |
5eca1c10 IM |
1396 | atomic_t trace_overrun; |
1397 | ||
1398 | /* Pause tracing: */ | |
1399 | atomic_t tracing_graph_pause; | |
f201ae23 | 1400 | #endif |
5eca1c10 | 1401 | |
ea4e2bc4 | 1402 | #ifdef CONFIG_TRACING |
5eca1c10 IM |
1403 | /* Bitmask and counter of trace recursion: */ |
1404 | unsigned long trace_recursion; | |
261842b7 | 1405 | #endif /* CONFIG_TRACING */ |
5eca1c10 | 1406 | |
5c9a8750 | 1407 | #ifdef CONFIG_KCOV |
eec028c9 AK |
1408 | /* See kernel/kcov.c for more details. */ |
1409 | ||
5eca1c10 | 1410 | /* Coverage collection mode enabled for this task (0 if disabled): */ |
0ed557aa | 1411 | unsigned int kcov_mode; |
5eca1c10 IM |
1412 | |
1413 | /* Size of the kcov_area: */ | |
1414 | unsigned int kcov_size; | |
1415 | ||
1416 | /* Buffer for coverage collection: */ | |
1417 | void *kcov_area; | |
1418 | ||
1419 | /* KCOV descriptor wired with this task or NULL: */ | |
1420 | struct kcov *kcov; | |
eec028c9 AK |
1421 | |
1422 | /* KCOV common handle for remote coverage collection: */ | |
1423 | u64 kcov_handle; | |
1424 | ||
1425 | /* KCOV sequence number: */ | |
1426 | int kcov_sequence; | |
5ff3b30a AK |
1427 | |
1428 | /* Collect coverage from softirq context: */ | |
1429 | unsigned int kcov_softirq; | |
5c9a8750 | 1430 | #endif |
5eca1c10 | 1431 | |
6f185c29 | 1432 | #ifdef CONFIG_MEMCG |
5eca1c10 IM |
1433 | struct mem_cgroup *memcg_in_oom; |
1434 | gfp_t memcg_oom_gfp_mask; | |
1435 | int memcg_oom_order; | |
b23afb93 | 1436 | |
5eca1c10 IM |
1437 | /* Number of pages to reclaim on returning to userland: */ |
1438 | unsigned int memcg_nr_pages_over_high; | |
d46eb14b SB |
1439 | |
1440 | /* Used by memcontrol for targeted memcg charge: */ | |
1441 | struct mem_cgroup *active_memcg; | |
569b846d | 1442 | #endif |
5eca1c10 | 1443 | |
d09d8df3 | 1444 | #ifdef CONFIG_BLK_CGROUP |
f05837ed | 1445 | struct gendisk *throttle_disk; |
d09d8df3 JB |
1446 | #endif |
1447 | ||
0326f5a9 | 1448 | #ifdef CONFIG_UPROBES |
5eca1c10 | 1449 | struct uprobe_task *utask; |
0326f5a9 | 1450 | #endif |
cafe5635 | 1451 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
5eca1c10 IM |
1452 | unsigned int sequential_io; |
1453 | unsigned int sequential_io_avg; | |
cafe5635 | 1454 | #endif |
5fbda3ec | 1455 | struct kmap_ctrl kmap_ctrl; |
8eb23b9f | 1456 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
5eca1c10 | 1457 | unsigned long task_state_change; |
5f220be2 TG |
1458 | # ifdef CONFIG_PREEMPT_RT |
1459 | unsigned long saved_state_change; | |
1460 | # endif | |
8eb23b9f | 1461 | #endif |
22df776a DV |
1462 | struct rcu_head rcu; |
1463 | refcount_t rcu_users; | |
5eca1c10 | 1464 | int pagefault_disabled; |
03049269 | 1465 | #ifdef CONFIG_MMU |
5eca1c10 | 1466 | struct task_struct *oom_reaper_list; |
e4a38402 | 1467 | struct timer_list oom_reaper_timer; |
03049269 | 1468 | #endif |
ba14a194 | 1469 | #ifdef CONFIG_VMAP_STACK |
5eca1c10 | 1470 | struct vm_struct *stack_vm_area; |
ba14a194 | 1471 | #endif |
68f24b08 | 1472 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
5eca1c10 | 1473 | /* A live task holds one reference: */ |
f0b89d39 | 1474 | refcount_t stack_refcount; |
d83a7cb3 JP |
1475 | #endif |
1476 | #ifdef CONFIG_LIVEPATCH | |
1477 | int patch_state; | |
0302e28d | 1478 | #endif |
e4e55b47 TH |
1479 | #ifdef CONFIG_SECURITY |
1480 | /* Used by LSM modules for access restriction: */ | |
1481 | void *security; | |
68f24b08 | 1482 | #endif |
a10787e6 SL |
1483 | #ifdef CONFIG_BPF_SYSCALL |
1484 | /* Used by BPF task local storage */ | |
1485 | struct bpf_local_storage __rcu *bpf_storage; | |
c7603cfa AN |
1486 | /* Used for BPF run context */ |
1487 | struct bpf_run_ctx *bpf_ctx; | |
a10787e6 | 1488 | #endif |
29e48ce8 | 1489 | |
afaef01c AP |
1490 | #ifdef CONFIG_GCC_PLUGIN_STACKLEAK |
1491 | unsigned long lowest_stack; | |
c8d12627 | 1492 | unsigned long prev_lowest_stack; |
afaef01c AP |
1493 | #endif |
1494 | ||
5567d11c | 1495 | #ifdef CONFIG_X86_MCE |
c0ab7ffc TL |
1496 | void __user *mce_vaddr; |
1497 | __u64 mce_kflags; | |
5567d11c | 1498 | u64 mce_addr; |
17fae129 TL |
1499 | __u64 mce_ripv : 1, |
1500 | mce_whole_page : 1, | |
1501 | __mce_reserved : 62; | |
5567d11c | 1502 | struct callback_head mce_kill_me; |
81065b35 | 1503 | int mce_count; |
5567d11c PZ |
1504 | #endif |
1505 | ||
d741bf41 PZ |
1506 | #ifdef CONFIG_KRETPROBES |
1507 | struct llist_head kretprobe_instances; | |
1508 | #endif | |
54ecbe6f MH |
1509 | #ifdef CONFIG_RETHOOK |
1510 | struct llist_head rethooks; | |
1511 | #endif | |
d741bf41 | 1512 | |
58e106e7 BS |
1513 | #ifdef CONFIG_ARCH_HAS_PARANOID_L1D_FLUSH |
1514 | /* | |
1515 | * If L1D flush is supported on mm context switch | |
1516 | * then we use this callback head to queue kill work | |
1517 | * to kill tasks that are not running on SMT disabled | |
1518 | * cores | |
1519 | */ | |
1520 | struct callback_head l1d_flush_kill; | |
1521 | #endif | |
1522 | ||
102227b9 DBO |
1523 | #ifdef CONFIG_RV |
1524 | /* | |
1525 | * Per-task RV monitor. Nowadays fixed in RV_PER_TASK_MONITORS. | |
1526 | * If we find justification for more monitors, we can think | |
1527 | * about adding more or developing a dynamic method. So far, | |
1528 | * none of these are justified. | |
1529 | */ | |
1530 | union rv_task_monitor rv[RV_PER_TASK_MONITORS]; | |
1531 | #endif | |
1532 | ||
fd593511 BB |
1533 | #ifdef CONFIG_USER_EVENTS |
1534 | struct user_event_mm *user_event_mm; | |
1535 | #endif | |
1536 | ||
29e48ce8 KC |
1537 | /* |
1538 | * New fields for task_struct should be added above here, so that | |
1539 | * they are included in the randomized portion of task_struct. | |
1540 | */ | |
1541 | randomized_struct_fields_end | |
1542 | ||
5eca1c10 IM |
1543 | /* CPU-specific state of this task: */ |
1544 | struct thread_struct thread; | |
1545 | ||
1546 | /* | |
1547 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1548 | * structure. It *MUST* be at the end of 'task_struct'. | |
1549 | * | |
1550 | * Do not put anything below here! | |
1551 | */ | |
1da177e4 LT |
1552 | }; |
1553 | ||
e868171a | 1554 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 | 1555 | { |
2c470475 | 1556 | return task->thread_pid; |
22c935f4 EB |
1557 | } |
1558 | ||
7af57294 PE |
1559 | /* |
1560 | * the helpers to get the task's different pids as they are seen | |
1561 | * from various namespaces | |
1562 | * | |
1563 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
1564 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
1565 | * current. | |
7af57294 PE |
1566 | * task_xid_nr_ns() : id seen from the ns specified; |
1567 | * | |
7af57294 PE |
1568 | * see also pid_nr() etc in include/linux/pid.h |
1569 | */ | |
5eca1c10 | 1570 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, struct pid_namespace *ns); |
7af57294 | 1571 | |
e868171a | 1572 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
1573 | { |
1574 | return tsk->pid; | |
1575 | } | |
1576 | ||
5eca1c10 | 1577 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
52ee2dfd ON |
1578 | { |
1579 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
1580 | } | |
7af57294 PE |
1581 | |
1582 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
1583 | { | |
52ee2dfd | 1584 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
1585 | } |
1586 | ||
1587 | ||
e868171a | 1588 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
1589 | { |
1590 | return tsk->tgid; | |
1591 | } | |
1592 | ||
5eca1c10 IM |
1593 | /** |
1594 | * pid_alive - check that a task structure is not stale | |
1595 | * @p: Task structure to be checked. | |
1596 | * | |
1597 | * Test if a process is not yet dead (at most zombie state) | |
1598 | * If pid_alive fails, then pointers within the task structure | |
1599 | * can be stale and must not be dereferenced. | |
1600 | * | |
1601 | * Return: 1 if the process is alive. 0 otherwise. | |
1602 | */ | |
1603 | static inline int pid_alive(const struct task_struct *p) | |
1604 | { | |
2c470475 | 1605 | return p->thread_pid != NULL; |
5eca1c10 | 1606 | } |
7af57294 | 1607 | |
5eca1c10 | 1608 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1609 | { |
52ee2dfd | 1610 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
1611 | } |
1612 | ||
7af57294 PE |
1613 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
1614 | { | |
52ee2dfd | 1615 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
1616 | } |
1617 | ||
1618 | ||
5eca1c10 | 1619 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
7af57294 | 1620 | { |
52ee2dfd | 1621 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
1622 | } |
1623 | ||
7af57294 PE |
1624 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
1625 | { | |
52ee2dfd | 1626 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
1627 | } |
1628 | ||
dd1c1f2f ON |
1629 | static inline pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns) |
1630 | { | |
6883f81a | 1631 | return __task_pid_nr_ns(tsk, PIDTYPE_TGID, ns); |
dd1c1f2f ON |
1632 | } |
1633 | ||
1634 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
1635 | { | |
6883f81a | 1636 | return __task_pid_nr_ns(tsk, PIDTYPE_TGID, NULL); |
dd1c1f2f ON |
1637 | } |
1638 | ||
1639 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) | |
1640 | { | |
1641 | pid_t pid = 0; | |
1642 | ||
1643 | rcu_read_lock(); | |
1644 | if (pid_alive(tsk)) | |
1645 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
1646 | rcu_read_unlock(); | |
1647 | ||
1648 | return pid; | |
1649 | } | |
1650 | ||
1651 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
1652 | { | |
1653 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
1654 | } | |
1655 | ||
5eca1c10 | 1656 | /* Obsolete, do not use: */ |
1b0f7ffd ON |
1657 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) |
1658 | { | |
1659 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
1660 | } | |
7af57294 | 1661 | |
06eb6184 PZ |
1662 | #define TASK_REPORT_IDLE (TASK_REPORT + 1) |
1663 | #define TASK_REPORT_MAX (TASK_REPORT_IDLE << 1) | |
1664 | ||
fa2c3254 VS |
1665 | static inline unsigned int __task_state_index(unsigned int tsk_state, |
1666 | unsigned int tsk_exit_state) | |
20435d84 | 1667 | { |
fa2c3254 | 1668 | unsigned int state = (tsk_state | tsk_exit_state) & TASK_REPORT; |
20435d84 | 1669 | |
06eb6184 PZ |
1670 | BUILD_BUG_ON_NOT_POWER_OF_2(TASK_REPORT_MAX); |
1671 | ||
06eb6184 PZ |
1672 | if (tsk_state == TASK_IDLE) |
1673 | state = TASK_REPORT_IDLE; | |
1674 | ||
25795ef6 VS |
1675 | /* |
1676 | * We're lying here, but rather than expose a completely new task state | |
1677 | * to userspace, we can make this appear as if the task has gone through | |
1678 | * a regular rt_mutex_lock() call. | |
1679 | */ | |
1680 | if (tsk_state == TASK_RTLOCK_WAIT) | |
1681 | state = TASK_UNINTERRUPTIBLE; | |
1682 | ||
1593baab PZ |
1683 | return fls(state); |
1684 | } | |
1685 | ||
fa2c3254 VS |
1686 | static inline unsigned int task_state_index(struct task_struct *tsk) |
1687 | { | |
1688 | return __task_state_index(READ_ONCE(tsk->__state), tsk->exit_state); | |
1689 | } | |
1690 | ||
1d48b080 | 1691 | static inline char task_index_to_char(unsigned int state) |
1593baab | 1692 | { |
8ef9925b | 1693 | static const char state_char[] = "RSDTtXZPI"; |
1593baab | 1694 | |
06eb6184 | 1695 | BUILD_BUG_ON(1 + ilog2(TASK_REPORT_MAX) != sizeof(state_char) - 1); |
20435d84 | 1696 | |
1593baab PZ |
1697 | return state_char[state]; |
1698 | } | |
1699 | ||
1700 | static inline char task_state_to_char(struct task_struct *tsk) | |
1701 | { | |
1d48b080 | 1702 | return task_index_to_char(task_state_index(tsk)); |
20435d84 XX |
1703 | } |
1704 | ||
f400e198 | 1705 | /** |
570f5241 SS |
1706 | * is_global_init - check if a task structure is init. Since init |
1707 | * is free to have sub-threads we need to check tgid. | |
3260259f H |
1708 | * @tsk: Task structure to be checked. |
1709 | * | |
1710 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
1711 | * |
1712 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 1713 | */ |
e868171a | 1714 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 1715 | { |
570f5241 | 1716 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 1717 | } |
b460cbc5 | 1718 | |
9ec52099 CLG |
1719 | extern struct pid *cad_pid; |
1720 | ||
1da177e4 LT |
1721 | /* |
1722 | * Per process flags | |
1723 | */ | |
01ccf592 | 1724 | #define PF_VCPU 0x00000001 /* I'm a virtual CPU */ |
5eca1c10 IM |
1725 | #define PF_IDLE 0x00000002 /* I am an IDLE thread */ |
1726 | #define PF_EXITING 0x00000004 /* Getting shut down */ | |
92307383 | 1727 | #define PF_POSTCOREDUMP 0x00000008 /* Coredumps should ignore this task */ |
01ccf592 | 1728 | #define PF_IO_WORKER 0x00000010 /* Task is an IO worker */ |
5eca1c10 IM |
1729 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ |
1730 | #define PF_FORKNOEXEC 0x00000040 /* Forked but didn't exec */ | |
1731 | #define PF_MCE_PROCESS 0x00000080 /* Process policy on mce errors */ | |
1732 | #define PF_SUPERPRIV 0x00000100 /* Used super-user privileges */ | |
1733 | #define PF_DUMPCORE 0x00000200 /* Dumped core */ | |
1734 | #define PF_SIGNALED 0x00000400 /* Killed by a signal */ | |
1735 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
1736 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user() noticed that RLIMIT_NPROC was exceeded */ | |
1737 | #define PF_USED_MATH 0x00002000 /* If unset the fpu must be initialized before use */ | |
54e6842d | 1738 | #define PF_USER_WORKER 0x00004000 /* Kernel thread cloned from userspace thread */ |
5eca1c10 | 1739 | #define PF_NOFREEZE 0x00008000 /* This thread should not be frozen */ |
fb04563d | 1740 | #define PF__HOLE__00010000 0x00010000 |
7dea19f9 MH |
1741 | #define PF_KSWAPD 0x00020000 /* I am kswapd */ |
1742 | #define PF_MEMALLOC_NOFS 0x00040000 /* All allocation requests will inherit GFP_NOFS */ | |
1743 | #define PF_MEMALLOC_NOIO 0x00080000 /* All allocation requests will inherit GFP_NOIO */ | |
a37b0715 N |
1744 | #define PF_LOCAL_THROTTLE 0x00100000 /* Throttle writes only against the bdi I write to, |
1745 | * I am cleaning dirty pages from some other bdi. */ | |
5eca1c10 IM |
1746 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ |
1747 | #define PF_RANDOMIZE 0x00400000 /* Randomize virtual address space */ | |
fb04563d PZ |
1748 | #define PF__HOLE__00800000 0x00800000 |
1749 | #define PF__HOLE__01000000 0x01000000 | |
1750 | #define PF__HOLE__02000000 0x02000000 | |
3bd37062 | 1751 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_mask */ |
5eca1c10 | 1752 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ |
1a08ae36 | 1753 | #define PF_MEMALLOC_PIN 0x10000000 /* Allocation context constrained to zones which allow long term pinning. */ |
fb04563d PZ |
1754 | #define PF__HOLE__20000000 0x20000000 |
1755 | #define PF__HOLE__40000000 0x40000000 | |
5eca1c10 | 1756 | #define PF_SUSPEND_TASK 0x80000000 /* This thread called freeze_processes() and should not be frozen */ |
1da177e4 LT |
1757 | |
1758 | /* | |
1759 | * Only the _current_ task can read/write to tsk->flags, but other | |
1760 | * tasks can access tsk->flags in readonly mode for example | |
1761 | * with tsk_used_math (like during threaded core dumping). | |
1762 | * There is however an exception to this rule during ptrace | |
1763 | * or during fork: the ptracer task is allowed to write to the | |
1764 | * child->flags of its traced child (same goes for fork, the parent | |
1765 | * can write to the child->flags), because we're guaranteed the | |
1766 | * child is not running and in turn not changing child->flags | |
1767 | * at the same time the parent does it. | |
1768 | */ | |
5eca1c10 IM |
1769 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) |
1770 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
1771 | #define clear_used_math() clear_stopped_child_used_math(current) | |
1772 | #define set_used_math() set_stopped_child_used_math(current) | |
1773 | ||
1da177e4 LT |
1774 | #define conditional_stopped_child_used_math(condition, child) \ |
1775 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
5eca1c10 IM |
1776 | |
1777 | #define conditional_used_math(condition) conditional_stopped_child_used_math(condition, current) | |
1778 | ||
1da177e4 LT |
1779 | #define copy_to_stopped_child_used_math(child) \ |
1780 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
5eca1c10 | 1781 | |
1da177e4 | 1782 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ |
5eca1c10 IM |
1783 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) |
1784 | #define used_math() tsk_used_math(current) | |
1da177e4 | 1785 | |
83d40a61 | 1786 | static __always_inline bool is_percpu_thread(void) |
62ec05dd TG |
1787 | { |
1788 | #ifdef CONFIG_SMP | |
1789 | return (current->flags & PF_NO_SETAFFINITY) && | |
1790 | (current->nr_cpus_allowed == 1); | |
1791 | #else | |
1792 | return true; | |
1793 | #endif | |
1794 | } | |
1795 | ||
1d4457f9 | 1796 | /* Per-process atomic flags. */ |
5eca1c10 IM |
1797 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
1798 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ | |
1799 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
356e4bff TG |
1800 | #define PFA_SPEC_SSB_DISABLE 3 /* Speculative Store Bypass disabled */ |
1801 | #define PFA_SPEC_SSB_FORCE_DISABLE 4 /* Speculative Store Bypass force disabled*/ | |
9137bb27 TG |
1802 | #define PFA_SPEC_IB_DISABLE 5 /* Indirect branch speculation restricted */ |
1803 | #define PFA_SPEC_IB_FORCE_DISABLE 6 /* Indirect branch speculation permanently restricted */ | |
71368af9 | 1804 | #define PFA_SPEC_SSB_NOEXEC 7 /* Speculative Store Bypass clear on execve() */ |
1d4457f9 | 1805 | |
e0e5070b ZL |
1806 | #define TASK_PFA_TEST(name, func) \ |
1807 | static inline bool task_##func(struct task_struct *p) \ | |
1808 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1809 | |
e0e5070b ZL |
1810 | #define TASK_PFA_SET(name, func) \ |
1811 | static inline void task_set_##func(struct task_struct *p) \ | |
1812 | { set_bit(PFA_##name, &p->atomic_flags); } | |
5eca1c10 | 1813 | |
e0e5070b ZL |
1814 | #define TASK_PFA_CLEAR(name, func) \ |
1815 | static inline void task_clear_##func(struct task_struct *p) \ | |
1816 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
1817 | ||
1818 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
1819 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 1820 | |
2ad654bc ZL |
1821 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
1822 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
1823 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
1824 | ||
1825 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
1826 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
1827 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 1828 | |
356e4bff TG |
1829 | TASK_PFA_TEST(SPEC_SSB_DISABLE, spec_ssb_disable) |
1830 | TASK_PFA_SET(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1831 | TASK_PFA_CLEAR(SPEC_SSB_DISABLE, spec_ssb_disable) | |
1832 | ||
71368af9 WL |
1833 | TASK_PFA_TEST(SPEC_SSB_NOEXEC, spec_ssb_noexec) |
1834 | TASK_PFA_SET(SPEC_SSB_NOEXEC, spec_ssb_noexec) | |
1835 | TASK_PFA_CLEAR(SPEC_SSB_NOEXEC, spec_ssb_noexec) | |
1836 | ||
356e4bff TG |
1837 | TASK_PFA_TEST(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) |
1838 | TASK_PFA_SET(SPEC_SSB_FORCE_DISABLE, spec_ssb_force_disable) | |
1839 | ||
9137bb27 TG |
1840 | TASK_PFA_TEST(SPEC_IB_DISABLE, spec_ib_disable) |
1841 | TASK_PFA_SET(SPEC_IB_DISABLE, spec_ib_disable) | |
1842 | TASK_PFA_CLEAR(SPEC_IB_DISABLE, spec_ib_disable) | |
1843 | ||
1844 | TASK_PFA_TEST(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) | |
1845 | TASK_PFA_SET(SPEC_IB_FORCE_DISABLE, spec_ib_force_disable) | |
1846 | ||
5eca1c10 | 1847 | static inline void |
717a94b5 | 1848 | current_restore_flags(unsigned long orig_flags, unsigned long flags) |
907aed48 | 1849 | { |
717a94b5 N |
1850 | current->flags &= ~flags; |
1851 | current->flags |= orig_flags & flags; | |
907aed48 MG |
1852 | } |
1853 | ||
5eca1c10 | 1854 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, const struct cpumask *trial); |
b6e8d40d | 1855 | extern int task_can_attach(struct task_struct *p, const struct cpumask *cs_effective_cpus); |
1da177e4 | 1856 | #ifdef CONFIG_SMP |
5eca1c10 IM |
1857 | extern void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask); |
1858 | extern int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask); | |
b90ca8ba WD |
1859 | extern int dup_user_cpus_ptr(struct task_struct *dst, struct task_struct *src, int node); |
1860 | extern void release_user_cpus_ptr(struct task_struct *p); | |
234b8ab6 | 1861 | extern int dl_task_check_affinity(struct task_struct *p, const struct cpumask *mask); |
07ec77a1 WD |
1862 | extern void force_compatible_cpus_allowed_ptr(struct task_struct *p); |
1863 | extern void relax_compatible_cpus_allowed_ptr(struct task_struct *p); | |
1da177e4 | 1864 | #else |
5eca1c10 | 1865 | static inline void do_set_cpus_allowed(struct task_struct *p, const struct cpumask *new_mask) |
1e1b6c51 KM |
1866 | { |
1867 | } | |
5eca1c10 | 1868 | static inline int set_cpus_allowed_ptr(struct task_struct *p, const struct cpumask *new_mask) |
1da177e4 | 1869 | { |
96f874e2 | 1870 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
1871 | return -EINVAL; |
1872 | return 0; | |
1873 | } | |
b90ca8ba WD |
1874 | static inline int dup_user_cpus_ptr(struct task_struct *dst, struct task_struct *src, int node) |
1875 | { | |
1876 | if (src->user_cpus_ptr) | |
1877 | return -EINVAL; | |
1878 | return 0; | |
1879 | } | |
1880 | static inline void release_user_cpus_ptr(struct task_struct *p) | |
1881 | { | |
1882 | WARN_ON(p->user_cpus_ptr); | |
1883 | } | |
234b8ab6 WD |
1884 | |
1885 | static inline int dl_task_check_affinity(struct task_struct *p, const struct cpumask *mask) | |
1886 | { | |
1887 | return 0; | |
1888 | } | |
1da177e4 | 1889 | #endif |
e0ad9556 | 1890 | |
fa93384f | 1891 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
1892 | extern void set_user_nice(struct task_struct *p, long nice); |
1893 | extern int task_prio(const struct task_struct *p); | |
5eca1c10 | 1894 | |
d0ea0268 DY |
1895 | /** |
1896 | * task_nice - return the nice value of a given task. | |
1897 | * @p: the task in question. | |
1898 | * | |
1899 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
1900 | */ | |
1901 | static inline int task_nice(const struct task_struct *p) | |
1902 | { | |
1903 | return PRIO_TO_NICE((p)->static_prio); | |
1904 | } | |
5eca1c10 | 1905 | |
36c8b586 IM |
1906 | extern int can_nice(const struct task_struct *p, const int nice); |
1907 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 1908 | extern int idle_cpu(int cpu); |
943d355d | 1909 | extern int available_idle_cpu(int cpu); |
5eca1c10 IM |
1910 | extern int sched_setscheduler(struct task_struct *, int, const struct sched_param *); |
1911 | extern int sched_setscheduler_nocheck(struct task_struct *, int, const struct sched_param *); | |
8b700983 PZ |
1912 | extern void sched_set_fifo(struct task_struct *p); |
1913 | extern void sched_set_fifo_low(struct task_struct *p); | |
1914 | extern void sched_set_normal(struct task_struct *p, int nice); | |
5eca1c10 | 1915 | extern int sched_setattr(struct task_struct *, const struct sched_attr *); |
794a56eb | 1916 | extern int sched_setattr_nocheck(struct task_struct *, const struct sched_attr *); |
36c8b586 | 1917 | extern struct task_struct *idle_task(int cpu); |
5eca1c10 | 1918 | |
c4f30608 PM |
1919 | /** |
1920 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 1921 | * @p: the task in question. |
e69f6186 YB |
1922 | * |
1923 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 1924 | */ |
c94a88f3 | 1925 | static __always_inline bool is_idle_task(const struct task_struct *p) |
c4f30608 | 1926 | { |
c1de45ca | 1927 | return !!(p->flags & PF_IDLE); |
c4f30608 | 1928 | } |
5eca1c10 | 1929 | |
36c8b586 | 1930 | extern struct task_struct *curr_task(int cpu); |
a458ae2e | 1931 | extern void ia64_set_curr_task(int cpu, struct task_struct *p); |
1da177e4 LT |
1932 | |
1933 | void yield(void); | |
1934 | ||
1da177e4 | 1935 | union thread_union { |
0500871f DH |
1936 | #ifndef CONFIG_ARCH_TASK_STRUCT_ON_STACK |
1937 | struct task_struct task; | |
1938 | #endif | |
c65eacbe | 1939 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1da177e4 | 1940 | struct thread_info thread_info; |
c65eacbe | 1941 | #endif |
1da177e4 LT |
1942 | unsigned long stack[THREAD_SIZE/sizeof(long)]; |
1943 | }; | |
1944 | ||
0500871f DH |
1945 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1946 | extern struct thread_info init_thread_info; | |
1947 | #endif | |
1948 | ||
1949 | extern unsigned long init_stack[THREAD_SIZE / sizeof(unsigned long)]; | |
1950 | ||
f3ac6067 | 1951 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
bcf9033e | 1952 | # define task_thread_info(task) (&(task)->thread_info) |
f3ac6067 IM |
1953 | #elif !defined(__HAVE_THREAD_FUNCTIONS) |
1954 | # define task_thread_info(task) ((struct thread_info *)(task)->stack) | |
1955 | #endif | |
1956 | ||
198fe21b PE |
1957 | /* |
1958 | * find a task by one of its numerical ids | |
1959 | * | |
198fe21b PE |
1960 | * find_task_by_pid_ns(): |
1961 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
1962 | * find_task_by_vpid(): |
1963 | * finds a task by its virtual pid | |
198fe21b | 1964 | * |
e49859e7 | 1965 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
1966 | */ |
1967 | ||
228ebcbe | 1968 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
5eca1c10 | 1969 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, struct pid_namespace *ns); |
198fe21b | 1970 | |
2ee08260 MR |
1971 | /* |
1972 | * find a task by its virtual pid and get the task struct | |
1973 | */ | |
1974 | extern struct task_struct *find_get_task_by_vpid(pid_t nr); | |
1975 | ||
b3c97528 HH |
1976 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
1977 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 1978 | extern void wake_up_new_task(struct task_struct *tsk); |
5eca1c10 | 1979 | |
1da177e4 | 1980 | #ifdef CONFIG_SMP |
5eca1c10 | 1981 | extern void kick_process(struct task_struct *tsk); |
1da177e4 | 1982 | #else |
5eca1c10 | 1983 | static inline void kick_process(struct task_struct *tsk) { } |
1da177e4 | 1984 | #endif |
1da177e4 | 1985 | |
82b89778 | 1986 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
5eca1c10 | 1987 | |
82b89778 AH |
1988 | static inline void set_task_comm(struct task_struct *tsk, const char *from) |
1989 | { | |
1990 | __set_task_comm(tsk, from, false); | |
1991 | } | |
5eca1c10 | 1992 | |
3756f640 AB |
1993 | extern char *__get_task_comm(char *to, size_t len, struct task_struct *tsk); |
1994 | #define get_task_comm(buf, tsk) ({ \ | |
1995 | BUILD_BUG_ON(sizeof(buf) != TASK_COMM_LEN); \ | |
1996 | __get_task_comm(buf, sizeof(buf), tsk); \ | |
1997 | }) | |
1da177e4 LT |
1998 | |
1999 | #ifdef CONFIG_SMP | |
2a0a24eb TG |
2000 | static __always_inline void scheduler_ipi(void) |
2001 | { | |
2002 | /* | |
2003 | * Fold TIF_NEED_RESCHED into the preempt_count; anybody setting | |
2004 | * TIF_NEED_RESCHED remotely (for the first time) will also send | |
2005 | * this IPI. | |
2006 | */ | |
2007 | preempt_fold_need_resched(); | |
2008 | } | |
2f064a59 | 2009 | extern unsigned long wait_task_inactive(struct task_struct *, unsigned int match_state); |
1da177e4 | 2010 | #else |
184748cc | 2011 | static inline void scheduler_ipi(void) { } |
2f064a59 | 2012 | static inline unsigned long wait_task_inactive(struct task_struct *p, unsigned int match_state) |
85ba2d86 RM |
2013 | { |
2014 | return 1; | |
2015 | } | |
1da177e4 LT |
2016 | #endif |
2017 | ||
5eca1c10 IM |
2018 | /* |
2019 | * Set thread flags in other task's structures. | |
2020 | * See asm/thread_info.h for TIF_xxxx flags available: | |
1da177e4 LT |
2021 | */ |
2022 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
2023 | { | |
a1261f54 | 2024 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
2025 | } |
2026 | ||
2027 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
2028 | { | |
a1261f54 | 2029 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
2030 | } |
2031 | ||
93ee37c2 DM |
2032 | static inline void update_tsk_thread_flag(struct task_struct *tsk, int flag, |
2033 | bool value) | |
2034 | { | |
2035 | update_ti_thread_flag(task_thread_info(tsk), flag, value); | |
2036 | } | |
2037 | ||
1da177e4 LT |
2038 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) |
2039 | { | |
a1261f54 | 2040 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
2041 | } |
2042 | ||
2043 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
2044 | { | |
a1261f54 | 2045 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
2046 | } |
2047 | ||
2048 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
2049 | { | |
a1261f54 | 2050 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
2051 | } |
2052 | ||
2053 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
2054 | { | |
2055 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
2056 | } | |
2057 | ||
2058 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
2059 | { | |
2060 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
2061 | } | |
2062 | ||
8ae121ac GH |
2063 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
2064 | { | |
2065 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
2066 | } | |
2067 | ||
1da177e4 LT |
2068 | /* |
2069 | * cond_resched() and cond_resched_lock(): latency reduction via | |
2070 | * explicit rescheduling in places that are safe. The return | |
2071 | * value indicates whether a reschedule was done in fact. | |
2072 | * cond_resched_lock() will drop the spinlock before scheduling, | |
1da177e4 | 2073 | */ |
b965f1dd PZI |
2074 | #if !defined(CONFIG_PREEMPTION) || defined(CONFIG_PREEMPT_DYNAMIC) |
2075 | extern int __cond_resched(void); | |
2076 | ||
99cf983c | 2077 | #if defined(CONFIG_PREEMPT_DYNAMIC) && defined(CONFIG_HAVE_PREEMPT_DYNAMIC_CALL) |
b965f1dd | 2078 | |
e3ff7c60 JP |
2079 | void sched_dynamic_klp_enable(void); |
2080 | void sched_dynamic_klp_disable(void); | |
2081 | ||
b965f1dd PZI |
2082 | DECLARE_STATIC_CALL(cond_resched, __cond_resched); |
2083 | ||
2084 | static __always_inline int _cond_resched(void) | |
2085 | { | |
ef72661e | 2086 | return static_call_mod(cond_resched)(); |
b965f1dd PZI |
2087 | } |
2088 | ||
99cf983c | 2089 | #elif defined(CONFIG_PREEMPT_DYNAMIC) && defined(CONFIG_HAVE_PREEMPT_DYNAMIC_KEY) |
e3ff7c60 | 2090 | |
99cf983c MR |
2091 | extern int dynamic_cond_resched(void); |
2092 | ||
2093 | static __always_inline int _cond_resched(void) | |
2094 | { | |
2095 | return dynamic_cond_resched(); | |
2096 | } | |
2097 | ||
e3ff7c60 | 2098 | #else /* !CONFIG_PREEMPTION */ |
b965f1dd PZI |
2099 | |
2100 | static inline int _cond_resched(void) | |
2101 | { | |
e3ff7c60 | 2102 | klp_sched_try_switch(); |
b965f1dd PZI |
2103 | return __cond_resched(); |
2104 | } | |
2105 | ||
e3ff7c60 | 2106 | #endif /* PREEMPT_DYNAMIC && CONFIG_HAVE_PREEMPT_DYNAMIC_CALL */ |
b965f1dd | 2107 | |
e3ff7c60 | 2108 | #else /* CONFIG_PREEMPTION && !CONFIG_PREEMPT_DYNAMIC */ |
b965f1dd | 2109 | |
e3ff7c60 JP |
2110 | static inline int _cond_resched(void) |
2111 | { | |
2112 | klp_sched_try_switch(); | |
2113 | return 0; | |
2114 | } | |
b965f1dd | 2115 | |
e3ff7c60 | 2116 | #endif /* !CONFIG_PREEMPTION || CONFIG_PREEMPT_DYNAMIC */ |
6f80bd98 | 2117 | |
613afbf8 | 2118 | #define cond_resched() ({ \ |
874f670e | 2119 | __might_resched(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
2120 | _cond_resched(); \ |
2121 | }) | |
6f80bd98 | 2122 | |
613afbf8 | 2123 | extern int __cond_resched_lock(spinlock_t *lock); |
f3d4b4b1 BG |
2124 | extern int __cond_resched_rwlock_read(rwlock_t *lock); |
2125 | extern int __cond_resched_rwlock_write(rwlock_t *lock); | |
613afbf8 | 2126 | |
50e081b9 TG |
2127 | #define MIGHT_RESCHED_RCU_SHIFT 8 |
2128 | #define MIGHT_RESCHED_PREEMPT_MASK ((1U << MIGHT_RESCHED_RCU_SHIFT) - 1) | |
2129 | ||
3e9cc688 TG |
2130 | #ifndef CONFIG_PREEMPT_RT |
2131 | /* | |
2132 | * Non RT kernels have an elevated preempt count due to the held lock, | |
2133 | * but are not allowed to be inside a RCU read side critical section | |
2134 | */ | |
2135 | # define PREEMPT_LOCK_RESCHED_OFFSETS PREEMPT_LOCK_OFFSET | |
2136 | #else | |
2137 | /* | |
2138 | * spin/rw_lock() on RT implies rcu_read_lock(). The might_sleep() check in | |
2139 | * cond_resched*lock() has to take that into account because it checks for | |
2140 | * preempt_count() and rcu_preempt_depth(). | |
2141 | */ | |
2142 | # define PREEMPT_LOCK_RESCHED_OFFSETS \ | |
2143 | (PREEMPT_LOCK_OFFSET + (1U << MIGHT_RESCHED_RCU_SHIFT)) | |
2144 | #endif | |
2145 | ||
2146 | #define cond_resched_lock(lock) ({ \ | |
2147 | __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \ | |
2148 | __cond_resched_lock(lock); \ | |
613afbf8 FW |
2149 | }) |
2150 | ||
3e9cc688 TG |
2151 | #define cond_resched_rwlock_read(lock) ({ \ |
2152 | __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \ | |
2153 | __cond_resched_rwlock_read(lock); \ | |
f3d4b4b1 BG |
2154 | }) |
2155 | ||
3e9cc688 TG |
2156 | #define cond_resched_rwlock_write(lock) ({ \ |
2157 | __might_resched(__FILE__, __LINE__, PREEMPT_LOCK_RESCHED_OFFSETS); \ | |
2158 | __cond_resched_rwlock_write(lock); \ | |
f3d4b4b1 BG |
2159 | }) |
2160 | ||
f6f3c437 SH |
2161 | static inline void cond_resched_rcu(void) |
2162 | { | |
2163 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
2164 | rcu_read_unlock(); | |
2165 | cond_resched(); | |
2166 | rcu_read_lock(); | |
2167 | #endif | |
2168 | } | |
2169 | ||
cfe43f47 VS |
2170 | #ifdef CONFIG_PREEMPT_DYNAMIC |
2171 | ||
2172 | extern bool preempt_model_none(void); | |
2173 | extern bool preempt_model_voluntary(void); | |
2174 | extern bool preempt_model_full(void); | |
2175 | ||
2176 | #else | |
2177 | ||
2178 | static inline bool preempt_model_none(void) | |
2179 | { | |
2180 | return IS_ENABLED(CONFIG_PREEMPT_NONE); | |
2181 | } | |
2182 | static inline bool preempt_model_voluntary(void) | |
2183 | { | |
2184 | return IS_ENABLED(CONFIG_PREEMPT_VOLUNTARY); | |
2185 | } | |
2186 | static inline bool preempt_model_full(void) | |
2187 | { | |
2188 | return IS_ENABLED(CONFIG_PREEMPT); | |
2189 | } | |
2190 | ||
2191 | #endif | |
2192 | ||
2193 | static inline bool preempt_model_rt(void) | |
2194 | { | |
2195 | return IS_ENABLED(CONFIG_PREEMPT_RT); | |
2196 | } | |
2197 | ||
2198 | /* | |
2199 | * Does the preemption model allow non-cooperative preemption? | |
2200 | * | |
2201 | * For !CONFIG_PREEMPT_DYNAMIC kernels this is an exact match with | |
2202 | * CONFIG_PREEMPTION; for CONFIG_PREEMPT_DYNAMIC this doesn't work as the | |
2203 | * kernel is *built* with CONFIG_PREEMPTION=y but may run with e.g. the | |
2204 | * PREEMPT_NONE model. | |
2205 | */ | |
2206 | static inline bool preempt_model_preemptible(void) | |
2207 | { | |
2208 | return preempt_model_full() || preempt_model_rt(); | |
2209 | } | |
2210 | ||
1da177e4 LT |
2211 | /* |
2212 | * Does a critical section need to be broken due to another | |
c1a280b6 | 2213 | * task waiting?: (technically does not depend on CONFIG_PREEMPTION, |
95c354fe | 2214 | * but a general need for low latency) |
1da177e4 | 2215 | */ |
95c354fe | 2216 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 2217 | { |
c1a280b6 | 2218 | #ifdef CONFIG_PREEMPTION |
95c354fe NP |
2219 | return spin_is_contended(lock); |
2220 | #else | |
1da177e4 | 2221 | return 0; |
95c354fe | 2222 | #endif |
1da177e4 LT |
2223 | } |
2224 | ||
a09a689a BG |
2225 | /* |
2226 | * Check if a rwlock is contended. | |
2227 | * Returns non-zero if there is another task waiting on the rwlock. | |
2228 | * Returns zero if the lock is not contended or the system / underlying | |
2229 | * rwlock implementation does not support contention detection. | |
2230 | * Technically does not depend on CONFIG_PREEMPTION, but a general need | |
2231 | * for low latency. | |
2232 | */ | |
2233 | static inline int rwlock_needbreak(rwlock_t *lock) | |
2234 | { | |
2235 | #ifdef CONFIG_PREEMPTION | |
2236 | return rwlock_is_contended(lock); | |
2237 | #else | |
2238 | return 0; | |
2239 | #endif | |
2240 | } | |
2241 | ||
75f93fed PZ |
2242 | static __always_inline bool need_resched(void) |
2243 | { | |
2244 | return unlikely(tif_need_resched()); | |
2245 | } | |
2246 | ||
1da177e4 LT |
2247 | /* |
2248 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
2249 | */ | |
2250 | #ifdef CONFIG_SMP | |
2251 | ||
2252 | static inline unsigned int task_cpu(const struct task_struct *p) | |
2253 | { | |
c546951d | 2254 | return READ_ONCE(task_thread_info(p)->cpu); |
1da177e4 LT |
2255 | } |
2256 | ||
c65cc870 | 2257 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
2258 | |
2259 | #else | |
2260 | ||
2261 | static inline unsigned int task_cpu(const struct task_struct *p) | |
2262 | { | |
2263 | return 0; | |
2264 | } | |
2265 | ||
2266 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
2267 | { | |
2268 | } | |
2269 | ||
2270 | #endif /* CONFIG_SMP */ | |
2271 | ||
a1dfb631 | 2272 | extern bool sched_task_on_rq(struct task_struct *p); |
42a20f86 | 2273 | extern unsigned long get_wchan(struct task_struct *p); |
e386b672 | 2274 | extern struct task_struct *cpu_curr_snapshot(int cpu); |
a1dfb631 | 2275 | |
d9345c65 PX |
2276 | /* |
2277 | * In order to reduce various lock holder preemption latencies provide an | |
2278 | * interface to see if a vCPU is currently running or not. | |
2279 | * | |
2280 | * This allows us to terminate optimistic spin loops and block, analogous to | |
2281 | * the native optimistic spin heuristic of testing if the lock owner task is | |
2282 | * running or not. | |
2283 | */ | |
2284 | #ifndef vcpu_is_preempted | |
42fd8baa QC |
2285 | static inline bool vcpu_is_preempted(int cpu) |
2286 | { | |
2287 | return false; | |
2288 | } | |
d9345c65 PX |
2289 | #endif |
2290 | ||
96f874e2 RR |
2291 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
2292 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 2293 | |
82455257 DH |
2294 | #ifndef TASK_SIZE_OF |
2295 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
2296 | #endif | |
2297 | ||
a5418be9 | 2298 | #ifdef CONFIG_SMP |
c0bed69d KW |
2299 | static inline bool owner_on_cpu(struct task_struct *owner) |
2300 | { | |
2301 | /* | |
2302 | * As lock holder preemption issue, we both skip spinning if | |
2303 | * task is not on cpu or its cpu is preempted | |
2304 | */ | |
4cf75fd4 | 2305 | return READ_ONCE(owner->on_cpu) && !vcpu_is_preempted(task_cpu(owner)); |
c0bed69d KW |
2306 | } |
2307 | ||
a5418be9 | 2308 | /* Returns effective CPU energy utilization, as seen by the scheduler */ |
bb447999 | 2309 | unsigned long sched_cpu_util(int cpu); |
a5418be9 VK |
2310 | #endif /* CONFIG_SMP */ |
2311 | ||
d7822b1e MD |
2312 | #ifdef CONFIG_RSEQ |
2313 | ||
2314 | /* | |
2315 | * Map the event mask on the user-space ABI enum rseq_cs_flags | |
2316 | * for direct mask checks. | |
2317 | */ | |
2318 | enum rseq_event_mask_bits { | |
2319 | RSEQ_EVENT_PREEMPT_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT_BIT, | |
2320 | RSEQ_EVENT_SIGNAL_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL_BIT, | |
2321 | RSEQ_EVENT_MIGRATE_BIT = RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE_BIT, | |
2322 | }; | |
2323 | ||
2324 | enum rseq_event_mask { | |
2325 | RSEQ_EVENT_PREEMPT = (1U << RSEQ_EVENT_PREEMPT_BIT), | |
2326 | RSEQ_EVENT_SIGNAL = (1U << RSEQ_EVENT_SIGNAL_BIT), | |
2327 | RSEQ_EVENT_MIGRATE = (1U << RSEQ_EVENT_MIGRATE_BIT), | |
2328 | }; | |
2329 | ||
2330 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
2331 | { | |
2332 | if (t->rseq) | |
2333 | set_tsk_thread_flag(t, TIF_NOTIFY_RESUME); | |
2334 | } | |
2335 | ||
784e0300 | 2336 | void __rseq_handle_notify_resume(struct ksignal *sig, struct pt_regs *regs); |
d7822b1e | 2337 | |
784e0300 WD |
2338 | static inline void rseq_handle_notify_resume(struct ksignal *ksig, |
2339 | struct pt_regs *regs) | |
d7822b1e MD |
2340 | { |
2341 | if (current->rseq) | |
784e0300 | 2342 | __rseq_handle_notify_resume(ksig, regs); |
d7822b1e MD |
2343 | } |
2344 | ||
784e0300 WD |
2345 | static inline void rseq_signal_deliver(struct ksignal *ksig, |
2346 | struct pt_regs *regs) | |
d7822b1e MD |
2347 | { |
2348 | preempt_disable(); | |
2349 | __set_bit(RSEQ_EVENT_SIGNAL_BIT, ¤t->rseq_event_mask); | |
2350 | preempt_enable(); | |
784e0300 | 2351 | rseq_handle_notify_resume(ksig, regs); |
d7822b1e MD |
2352 | } |
2353 | ||
2354 | /* rseq_preempt() requires preemption to be disabled. */ | |
2355 | static inline void rseq_preempt(struct task_struct *t) | |
2356 | { | |
2357 | __set_bit(RSEQ_EVENT_PREEMPT_BIT, &t->rseq_event_mask); | |
2358 | rseq_set_notify_resume(t); | |
2359 | } | |
2360 | ||
2361 | /* rseq_migrate() requires preemption to be disabled. */ | |
2362 | static inline void rseq_migrate(struct task_struct *t) | |
2363 | { | |
2364 | __set_bit(RSEQ_EVENT_MIGRATE_BIT, &t->rseq_event_mask); | |
2365 | rseq_set_notify_resume(t); | |
2366 | } | |
2367 | ||
2368 | /* | |
2369 | * If parent process has a registered restartable sequences area, the | |
463f550f | 2370 | * child inherits. Unregister rseq for a clone with CLONE_VM set. |
d7822b1e MD |
2371 | */ |
2372 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
2373 | { | |
463f550f | 2374 | if (clone_flags & CLONE_VM) { |
d7822b1e | 2375 | t->rseq = NULL; |
ee3e3ac0 | 2376 | t->rseq_len = 0; |
d7822b1e MD |
2377 | t->rseq_sig = 0; |
2378 | t->rseq_event_mask = 0; | |
2379 | } else { | |
2380 | t->rseq = current->rseq; | |
ee3e3ac0 | 2381 | t->rseq_len = current->rseq_len; |
d7822b1e MD |
2382 | t->rseq_sig = current->rseq_sig; |
2383 | t->rseq_event_mask = current->rseq_event_mask; | |
d7822b1e MD |
2384 | } |
2385 | } | |
2386 | ||
2387 | static inline void rseq_execve(struct task_struct *t) | |
2388 | { | |
2389 | t->rseq = NULL; | |
ee3e3ac0 | 2390 | t->rseq_len = 0; |
d7822b1e MD |
2391 | t->rseq_sig = 0; |
2392 | t->rseq_event_mask = 0; | |
2393 | } | |
2394 | ||
2395 | #else | |
2396 | ||
2397 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
2398 | { | |
2399 | } | |
784e0300 WD |
2400 | static inline void rseq_handle_notify_resume(struct ksignal *ksig, |
2401 | struct pt_regs *regs) | |
d7822b1e MD |
2402 | { |
2403 | } | |
784e0300 WD |
2404 | static inline void rseq_signal_deliver(struct ksignal *ksig, |
2405 | struct pt_regs *regs) | |
d7822b1e MD |
2406 | { |
2407 | } | |
2408 | static inline void rseq_preempt(struct task_struct *t) | |
2409 | { | |
2410 | } | |
2411 | static inline void rseq_migrate(struct task_struct *t) | |
2412 | { | |
2413 | } | |
2414 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
2415 | { | |
2416 | } | |
2417 | static inline void rseq_execve(struct task_struct *t) | |
2418 | { | |
2419 | } | |
2420 | ||
2421 | #endif | |
2422 | ||
2423 | #ifdef CONFIG_DEBUG_RSEQ | |
2424 | ||
2425 | void rseq_syscall(struct pt_regs *regs); | |
2426 | ||
2427 | #else | |
2428 | ||
2429 | static inline void rseq_syscall(struct pt_regs *regs) | |
2430 | { | |
2431 | } | |
2432 | ||
2433 | #endif | |
2434 | ||
6e33cad0 PZ |
2435 | #ifdef CONFIG_SCHED_CORE |
2436 | extern void sched_core_free(struct task_struct *tsk); | |
85dd3f61 | 2437 | extern void sched_core_fork(struct task_struct *p); |
7ac592aa CH |
2438 | extern int sched_core_share_pid(unsigned int cmd, pid_t pid, enum pid_type type, |
2439 | unsigned long uaddr); | |
6e33cad0 PZ |
2440 | #else |
2441 | static inline void sched_core_free(struct task_struct *tsk) { } | |
85dd3f61 | 2442 | static inline void sched_core_fork(struct task_struct *p) { } |
6e33cad0 PZ |
2443 | #endif |
2444 | ||
d664e399 TG |
2445 | extern void sched_set_stop_task(int cpu, struct task_struct *stop); |
2446 | ||
1da177e4 | 2447 | #endif |