]>
Commit | Line | Data |
---|---|---|
1da177e4 LT |
1 | #ifndef _LINUX_SCHED_H |
2 | #define _LINUX_SCHED_H | |
3 | ||
607ca46e | 4 | #include <uapi/linux/sched.h> |
b7b3c76a | 5 | |
5c228079 DY |
6 | #include <linux/sched/prio.h> |
7 | ||
b7b3c76a DW |
8 | |
9 | struct sched_param { | |
10 | int sched_priority; | |
11 | }; | |
12 | ||
1da177e4 LT |
13 | #include <asm/param.h> /* for HZ */ |
14 | ||
1da177e4 LT |
15 | #include <linux/capability.h> |
16 | #include <linux/threads.h> | |
17 | #include <linux/kernel.h> | |
18 | #include <linux/types.h> | |
19 | #include <linux/timex.h> | |
20 | #include <linux/jiffies.h> | |
b69339ba | 21 | #include <linux/mutex.h> |
fb00aca4 | 22 | #include <linux/plist.h> |
1da177e4 LT |
23 | #include <linux/rbtree.h> |
24 | #include <linux/thread_info.h> | |
25 | #include <linux/cpumask.h> | |
26 | #include <linux/errno.h> | |
27 | #include <linux/nodemask.h> | |
c92ff1bd | 28 | #include <linux/mm_types.h> |
92cf2118 | 29 | #include <linux/preempt.h> |
1da177e4 | 30 | |
1da177e4 LT |
31 | #include <asm/page.h> |
32 | #include <asm/ptrace.h> | |
1da177e4 LT |
33 | |
34 | #include <linux/smp.h> | |
35 | #include <linux/sem.h> | |
ab602f79 | 36 | #include <linux/shm.h> |
1da177e4 | 37 | #include <linux/signal.h> |
1da177e4 LT |
38 | #include <linux/compiler.h> |
39 | #include <linux/completion.h> | |
f361bf4a | 40 | #include <linux/signal_types.h> |
1da177e4 LT |
41 | #include <linux/pid.h> |
42 | #include <linux/percpu.h> | |
43 | #include <linux/topology.h> | |
44 | #include <linux/seccomp.h> | |
e56d0903 | 45 | #include <linux/rcupdate.h> |
05725f7e | 46 | #include <linux/rculist.h> |
23f78d4a | 47 | #include <linux/rtmutex.h> |
1da177e4 | 48 | |
a3b6714e DW |
49 | #include <linux/time.h> |
50 | #include <linux/param.h> | |
51 | #include <linux/resource.h> | |
52 | #include <linux/timer.h> | |
53 | #include <linux/hrtimer.h> | |
5c9a8750 | 54 | #include <linux/kcov.h> |
7c3ab738 | 55 | #include <linux/task_io_accounting.h> |
9745512c | 56 | #include <linux/latencytop.h> |
9e2b2dc4 | 57 | #include <linux/cred.h> |
fa14ff4a | 58 | #include <linux/llist.h> |
7b44ab97 | 59 | #include <linux/uidgid.h> |
21caf2fc | 60 | #include <linux/gfp.h> |
fd771233 | 61 | #include <linux/topology.h> |
d4311ff1 | 62 | #include <linux/magic.h> |
7d7efec3 | 63 | #include <linux/cgroup-defs.h> |
a3b6714e DW |
64 | |
65 | #include <asm/processor.h> | |
36d57ac4 | 66 | |
d50dde5a DF |
67 | #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */ |
68 | ||
69 | /* | |
70 | * Extended scheduling parameters data structure. | |
71 | * | |
72 | * This is needed because the original struct sched_param can not be | |
73 | * altered without introducing ABI issues with legacy applications | |
74 | * (e.g., in sched_getparam()). | |
75 | * | |
76 | * However, the possibility of specifying more than just a priority for | |
77 | * the tasks may be useful for a wide variety of application fields, e.g., | |
78 | * multimedia, streaming, automation and control, and many others. | |
79 | * | |
80 | * This variant (sched_attr) is meant at describing a so-called | |
81 | * sporadic time-constrained task. In such model a task is specified by: | |
82 | * - the activation period or minimum instance inter-arrival time; | |
83 | * - the maximum (or average, depending on the actual scheduling | |
84 | * discipline) computation time of all instances, a.k.a. runtime; | |
85 | * - the deadline (relative to the actual activation time) of each | |
86 | * instance. | |
87 | * Very briefly, a periodic (sporadic) task asks for the execution of | |
88 | * some specific computation --which is typically called an instance-- | |
89 | * (at most) every period. Moreover, each instance typically lasts no more | |
90 | * than the runtime and must be completed by time instant t equal to | |
91 | * the instance activation time + the deadline. | |
92 | * | |
93 | * This is reflected by the actual fields of the sched_attr structure: | |
94 | * | |
95 | * @size size of the structure, for fwd/bwd compat. | |
96 | * | |
97 | * @sched_policy task's scheduling policy | |
98 | * @sched_flags for customizing the scheduler behaviour | |
99 | * @sched_nice task's nice value (SCHED_NORMAL/BATCH) | |
100 | * @sched_priority task's static priority (SCHED_FIFO/RR) | |
101 | * @sched_deadline representative of the task's deadline | |
102 | * @sched_runtime representative of the task's runtime | |
103 | * @sched_period representative of the task's period | |
104 | * | |
105 | * Given this task model, there are a multiplicity of scheduling algorithms | |
106 | * and policies, that can be used to ensure all the tasks will make their | |
107 | * timing constraints. | |
aab03e05 DF |
108 | * |
109 | * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the | |
110 | * only user of this new interface. More information about the algorithm | |
111 | * available in the scheduling class file or in Documentation/. | |
d50dde5a DF |
112 | */ |
113 | struct sched_attr { | |
114 | u32 size; | |
115 | ||
116 | u32 sched_policy; | |
117 | u64 sched_flags; | |
118 | ||
119 | /* SCHED_NORMAL, SCHED_BATCH */ | |
120 | s32 sched_nice; | |
121 | ||
122 | /* SCHED_FIFO, SCHED_RR */ | |
123 | u32 sched_priority; | |
124 | ||
125 | /* SCHED_DEADLINE */ | |
126 | u64 sched_runtime; | |
127 | u64 sched_deadline; | |
128 | u64 sched_period; | |
129 | }; | |
130 | ||
c87e2837 | 131 | struct futex_pi_state; |
286100a6 | 132 | struct robust_list_head; |
bddd87c7 | 133 | struct bio_list; |
5ad4e53b | 134 | struct fs_struct; |
cdd6c482 | 135 | struct perf_event_context; |
73c10101 | 136 | struct blk_plug; |
c4ad8f98 | 137 | struct filename; |
89076bc3 | 138 | struct nameidata; |
1da177e4 | 139 | |
1da177e4 LT |
140 | /* |
141 | * These are the constant used to fake the fixed-point load-average | |
142 | * counting. Some notes: | |
143 | * - 11 bit fractions expand to 22 bits by the multiplies: this gives | |
144 | * a load-average precision of 10 bits integer + 11 bits fractional | |
145 | * - if you want to count load-averages more often, you need more | |
146 | * precision, or rounding will get you. With 2-second counting freq, | |
147 | * the EXP_n values would be 1981, 2034 and 2043 if still using only | |
148 | * 11 bit fractions. | |
149 | */ | |
150 | extern unsigned long avenrun[]; /* Load averages */ | |
2d02494f | 151 | extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); |
1da177e4 LT |
152 | |
153 | #define FSHIFT 11 /* nr of bits of precision */ | |
154 | #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ | |
0c2043ab | 155 | #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ |
1da177e4 LT |
156 | #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ |
157 | #define EXP_5 2014 /* 1/exp(5sec/5min) */ | |
158 | #define EXP_15 2037 /* 1/exp(5sec/15min) */ | |
159 | ||
160 | #define CALC_LOAD(load,exp,n) \ | |
161 | load *= exp; \ | |
162 | load += n*(FIXED_1-exp); \ | |
163 | load >>= FSHIFT; | |
164 | ||
165 | extern unsigned long total_forks; | |
166 | extern int nr_threads; | |
1da177e4 LT |
167 | DECLARE_PER_CPU(unsigned long, process_counts); |
168 | extern int nr_processes(void); | |
169 | extern unsigned long nr_running(void); | |
2ee507c4 | 170 | extern bool single_task_running(void); |
1da177e4 | 171 | extern unsigned long nr_iowait(void); |
8c215bd3 | 172 | extern unsigned long nr_iowait_cpu(int cpu); |
372ba8cb | 173 | extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load); |
69d25870 | 174 | |
0f004f5a | 175 | extern void calc_global_load(unsigned long ticks); |
3289bdb4 PZ |
176 | |
177 | #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) | |
1f41906a FW |
178 | extern void cpu_load_update_nohz_start(void); |
179 | extern void cpu_load_update_nohz_stop(void); | |
3289bdb4 | 180 | #else |
1f41906a FW |
181 | static inline void cpu_load_update_nohz_start(void) { } |
182 | static inline void cpu_load_update_nohz_stop(void) { } | |
3289bdb4 | 183 | #endif |
1da177e4 | 184 | |
b637a328 PM |
185 | extern void dump_cpu_task(int cpu); |
186 | ||
43ae34cb IM |
187 | struct seq_file; |
188 | struct cfs_rq; | |
4cf86d77 | 189 | struct task_group; |
43ae34cb IM |
190 | #ifdef CONFIG_SCHED_DEBUG |
191 | extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); | |
192 | extern void proc_sched_set_task(struct task_struct *p); | |
43ae34cb | 193 | #endif |
1da177e4 | 194 | |
4a8342d2 LT |
195 | /* |
196 | * Task state bitmask. NOTE! These bits are also | |
197 | * encoded in fs/proc/array.c: get_task_state(). | |
198 | * | |
199 | * We have two separate sets of flags: task->state | |
200 | * is about runnability, while task->exit_state are | |
201 | * about the task exiting. Confusing, but this way | |
202 | * modifying one set can't modify the other one by | |
203 | * mistake. | |
204 | */ | |
1da177e4 LT |
205 | #define TASK_RUNNING 0 |
206 | #define TASK_INTERRUPTIBLE 1 | |
207 | #define TASK_UNINTERRUPTIBLE 2 | |
f021a3c2 MW |
208 | #define __TASK_STOPPED 4 |
209 | #define __TASK_TRACED 8 | |
4a8342d2 | 210 | /* in tsk->exit_state */ |
ad86622b ON |
211 | #define EXIT_DEAD 16 |
212 | #define EXIT_ZOMBIE 32 | |
abd50b39 | 213 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) |
4a8342d2 | 214 | /* in tsk->state again */ |
af927232 | 215 | #define TASK_DEAD 64 |
f021a3c2 | 216 | #define TASK_WAKEKILL 128 |
e9c84311 | 217 | #define TASK_WAKING 256 |
f2530dc7 | 218 | #define TASK_PARKED 512 |
80ed87c8 | 219 | #define TASK_NOLOAD 1024 |
7dc603c9 PZ |
220 | #define TASK_NEW 2048 |
221 | #define TASK_STATE_MAX 4096 | |
f021a3c2 | 222 | |
7dc603c9 | 223 | #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn" |
73342151 | 224 | |
642fa448 | 225 | /* Convenience macros for the sake of set_current_state */ |
f021a3c2 MW |
226 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) |
227 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
228 | #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) | |
1da177e4 | 229 | |
80ed87c8 PZ |
230 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) |
231 | ||
92a1f4bc MW |
232 | /* Convenience macros for the sake of wake_up */ |
233 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
f021a3c2 | 234 | #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) |
92a1f4bc MW |
235 | |
236 | /* get_task_state() */ | |
237 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
f021a3c2 | 238 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ |
74e37200 | 239 | __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD) |
92a1f4bc | 240 | |
f021a3c2 MW |
241 | #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) |
242 | #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) | |
92a1f4bc | 243 | #define task_is_stopped_or_traced(task) \ |
f021a3c2 | 244 | ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) |
92a1f4bc | 245 | #define task_contributes_to_load(task) \ |
e3c8ca83 | 246 | ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ |
80ed87c8 PZ |
247 | (task->flags & PF_FROZEN) == 0 && \ |
248 | (task->state & TASK_NOLOAD) == 0) | |
1da177e4 | 249 | |
8eb23b9f PZ |
250 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
251 | ||
8eb23b9f PZ |
252 | #define __set_current_state(state_value) \ |
253 | do { \ | |
254 | current->task_state_change = _THIS_IP_; \ | |
255 | current->state = (state_value); \ | |
256 | } while (0) | |
257 | #define set_current_state(state_value) \ | |
258 | do { \ | |
259 | current->task_state_change = _THIS_IP_; \ | |
a2250238 | 260 | smp_store_mb(current->state, (state_value)); \ |
8eb23b9f PZ |
261 | } while (0) |
262 | ||
263 | #else | |
498d0c57 AM |
264 | /* |
265 | * set_current_state() includes a barrier so that the write of current->state | |
266 | * is correctly serialised wrt the caller's subsequent test of whether to | |
267 | * actually sleep: | |
268 | * | |
a2250238 | 269 | * for (;;) { |
498d0c57 | 270 | * set_current_state(TASK_UNINTERRUPTIBLE); |
a2250238 PZ |
271 | * if (!need_sleep) |
272 | * break; | |
273 | * | |
274 | * schedule(); | |
275 | * } | |
276 | * __set_current_state(TASK_RUNNING); | |
277 | * | |
278 | * If the caller does not need such serialisation (because, for instance, the | |
279 | * condition test and condition change and wakeup are under the same lock) then | |
280 | * use __set_current_state(). | |
281 | * | |
282 | * The above is typically ordered against the wakeup, which does: | |
283 | * | |
284 | * need_sleep = false; | |
285 | * wake_up_state(p, TASK_UNINTERRUPTIBLE); | |
286 | * | |
287 | * Where wake_up_state() (and all other wakeup primitives) imply enough | |
288 | * barriers to order the store of the variable against wakeup. | |
289 | * | |
290 | * Wakeup will do: if (@state & p->state) p->state = TASK_RUNNING, that is, | |
291 | * once it observes the TASK_UNINTERRUPTIBLE store the waking CPU can issue a | |
292 | * TASK_RUNNING store which can collide with __set_current_state(TASK_RUNNING). | |
498d0c57 | 293 | * |
a2250238 | 294 | * This is obviously fine, since they both store the exact same value. |
498d0c57 | 295 | * |
a2250238 | 296 | * Also see the comments of try_to_wake_up(). |
498d0c57 | 297 | */ |
8eb23b9f | 298 | #define __set_current_state(state_value) \ |
1da177e4 | 299 | do { current->state = (state_value); } while (0) |
8eb23b9f | 300 | #define set_current_state(state_value) \ |
b92b8b35 | 301 | smp_store_mb(current->state, (state_value)) |
1da177e4 | 302 | |
8eb23b9f PZ |
303 | #endif |
304 | ||
1da177e4 LT |
305 | /* Task command name length */ |
306 | #define TASK_COMM_LEN 16 | |
307 | ||
1da177e4 LT |
308 | #include <linux/spinlock.h> |
309 | ||
310 | /* | |
311 | * This serializes "schedule()" and also protects | |
312 | * the run-queue from deletions/modifications (but | |
313 | * _adding_ to the beginning of the run-queue has | |
314 | * a separate lock). | |
315 | */ | |
316 | extern rwlock_t tasklist_lock; | |
317 | extern spinlock_t mmlist_lock; | |
318 | ||
36c8b586 | 319 | struct task_struct; |
1da177e4 | 320 | |
db1466b3 PM |
321 | #ifdef CONFIG_PROVE_RCU |
322 | extern int lockdep_tasklist_lock_is_held(void); | |
323 | #endif /* #ifdef CONFIG_PROVE_RCU */ | |
324 | ||
1da177e4 LT |
325 | extern void sched_init(void); |
326 | extern void sched_init_smp(void); | |
2d07b255 | 327 | extern asmlinkage void schedule_tail(struct task_struct *prev); |
36c8b586 | 328 | extern void init_idle(struct task_struct *idle, int cpu); |
1df21055 | 329 | extern void init_idle_bootup_task(struct task_struct *idle); |
1da177e4 | 330 | |
3fa0818b RR |
331 | extern cpumask_var_t cpu_isolated_map; |
332 | ||
89f19f04 | 333 | extern int runqueue_is_locked(int cpu); |
017730c1 | 334 | |
3451d024 | 335 | #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) |
c1cc017c | 336 | extern void nohz_balance_enter_idle(int cpu); |
69e1e811 | 337 | extern void set_cpu_sd_state_idle(void); |
bc7a34b8 | 338 | extern int get_nohz_timer_target(void); |
46cb4b7c | 339 | #else |
c1cc017c | 340 | static inline void nohz_balance_enter_idle(int cpu) { } |
fdaabd80 | 341 | static inline void set_cpu_sd_state_idle(void) { } |
46cb4b7c | 342 | #endif |
1da177e4 | 343 | |
e59e2ae2 | 344 | /* |
39bc89fd | 345 | * Only dump TASK_* tasks. (0 for all tasks) |
e59e2ae2 IM |
346 | */ |
347 | extern void show_state_filter(unsigned long state_filter); | |
348 | ||
349 | static inline void show_state(void) | |
350 | { | |
39bc89fd | 351 | show_state_filter(0); |
e59e2ae2 IM |
352 | } |
353 | ||
1da177e4 LT |
354 | extern void show_regs(struct pt_regs *); |
355 | ||
356 | /* | |
357 | * TASK is a pointer to the task whose backtrace we want to see (or NULL for current | |
358 | * task), SP is the stack pointer of the first frame that should be shown in the back | |
359 | * trace (or NULL if the entire call-chain of the task should be shown). | |
360 | */ | |
361 | extern void show_stack(struct task_struct *task, unsigned long *sp); | |
362 | ||
1da177e4 LT |
363 | extern void cpu_init (void); |
364 | extern void trap_init(void); | |
365 | extern void update_process_times(int user); | |
366 | extern void scheduler_tick(void); | |
9cf7243d | 367 | extern int sched_cpu_starting(unsigned int cpu); |
40190a78 TG |
368 | extern int sched_cpu_activate(unsigned int cpu); |
369 | extern int sched_cpu_deactivate(unsigned int cpu); | |
1da177e4 | 370 | |
f2785ddb TG |
371 | #ifdef CONFIG_HOTPLUG_CPU |
372 | extern int sched_cpu_dying(unsigned int cpu); | |
373 | #else | |
374 | # define sched_cpu_dying NULL | |
375 | #endif | |
1da177e4 | 376 | |
82a1fcb9 IM |
377 | extern void sched_show_task(struct task_struct *p); |
378 | ||
19cc36c0 | 379 | #ifdef CONFIG_LOCKUP_DETECTOR |
03e0d461 | 380 | extern void touch_softlockup_watchdog_sched(void); |
8446f1d3 | 381 | extern void touch_softlockup_watchdog(void); |
d6ad3e28 | 382 | extern void touch_softlockup_watchdog_sync(void); |
04c9167f | 383 | extern void touch_all_softlockup_watchdogs(void); |
332fbdbc DZ |
384 | extern int proc_dowatchdog_thresh(struct ctl_table *table, int write, |
385 | void __user *buffer, | |
386 | size_t *lenp, loff_t *ppos); | |
9c44bc03 | 387 | extern unsigned int softlockup_panic; |
ac1f5912 | 388 | extern unsigned int hardlockup_panic; |
004417a6 | 389 | void lockup_detector_init(void); |
8446f1d3 | 390 | #else |
03e0d461 TH |
391 | static inline void touch_softlockup_watchdog_sched(void) |
392 | { | |
393 | } | |
8446f1d3 IM |
394 | static inline void touch_softlockup_watchdog(void) |
395 | { | |
396 | } | |
d6ad3e28 JW |
397 | static inline void touch_softlockup_watchdog_sync(void) |
398 | { | |
399 | } | |
04c9167f JF |
400 | static inline void touch_all_softlockup_watchdogs(void) |
401 | { | |
402 | } | |
004417a6 PZ |
403 | static inline void lockup_detector_init(void) |
404 | { | |
405 | } | |
8446f1d3 IM |
406 | #endif |
407 | ||
8b414521 MT |
408 | #ifdef CONFIG_DETECT_HUNG_TASK |
409 | void reset_hung_task_detector(void); | |
410 | #else | |
411 | static inline void reset_hung_task_detector(void) | |
412 | { | |
413 | } | |
414 | #endif | |
415 | ||
1da177e4 LT |
416 | /* Attach to any functions which should be ignored in wchan output. */ |
417 | #define __sched __attribute__((__section__(".sched.text"))) | |
deaf2227 IM |
418 | |
419 | /* Linker adds these: start and end of __sched functions */ | |
420 | extern char __sched_text_start[], __sched_text_end[]; | |
421 | ||
1da177e4 LT |
422 | /* Is this address in the __sched functions? */ |
423 | extern int in_sched_functions(unsigned long addr); | |
424 | ||
425 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX | |
b3c97528 | 426 | extern signed long schedule_timeout(signed long timeout); |
64ed93a2 | 427 | extern signed long schedule_timeout_interruptible(signed long timeout); |
294d5cc2 | 428 | extern signed long schedule_timeout_killable(signed long timeout); |
64ed93a2 | 429 | extern signed long schedule_timeout_uninterruptible(signed long timeout); |
69b27baf | 430 | extern signed long schedule_timeout_idle(signed long timeout); |
1da177e4 | 431 | asmlinkage void schedule(void); |
c5491ea7 | 432 | extern void schedule_preempt_disabled(void); |
1da177e4 | 433 | |
10ab5643 TH |
434 | extern int __must_check io_schedule_prepare(void); |
435 | extern void io_schedule_finish(int token); | |
9cff8ade | 436 | extern long io_schedule_timeout(long timeout); |
10ab5643 | 437 | extern void io_schedule(void); |
9cff8ade | 438 | |
9af6528e PZ |
439 | void __noreturn do_task_dead(void); |
440 | ||
ab516013 | 441 | struct nsproxy; |
acce292c | 442 | struct user_namespace; |
1da177e4 | 443 | |
efc1a3b1 DH |
444 | #ifdef CONFIG_MMU |
445 | extern void arch_pick_mmap_layout(struct mm_struct *mm); | |
1da177e4 LT |
446 | extern unsigned long |
447 | arch_get_unmapped_area(struct file *, unsigned long, unsigned long, | |
448 | unsigned long, unsigned long); | |
449 | extern unsigned long | |
450 | arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, | |
451 | unsigned long len, unsigned long pgoff, | |
452 | unsigned long flags); | |
efc1a3b1 DH |
453 | #else |
454 | static inline void arch_pick_mmap_layout(struct mm_struct *mm) {} | |
455 | #endif | |
1da177e4 | 456 | |
d049f74f KC |
457 | #define SUID_DUMP_DISABLE 0 /* No setuid dumping */ |
458 | #define SUID_DUMP_USER 1 /* Dump as user of process */ | |
459 | #define SUID_DUMP_ROOT 2 /* Dump as root */ | |
460 | ||
6c5d5238 | 461 | /* mm flags */ |
f8af4da3 | 462 | |
7288e118 | 463 | /* for SUID_DUMP_* above */ |
3cb4a0bb | 464 | #define MMF_DUMPABLE_BITS 2 |
f8af4da3 | 465 | #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1) |
3cb4a0bb | 466 | |
942be387 ON |
467 | extern void set_dumpable(struct mm_struct *mm, int value); |
468 | /* | |
469 | * This returns the actual value of the suid_dumpable flag. For things | |
470 | * that are using this for checking for privilege transitions, it must | |
471 | * test against SUID_DUMP_USER rather than treating it as a boolean | |
472 | * value. | |
473 | */ | |
474 | static inline int __get_dumpable(unsigned long mm_flags) | |
475 | { | |
476 | return mm_flags & MMF_DUMPABLE_MASK; | |
477 | } | |
478 | ||
479 | static inline int get_dumpable(struct mm_struct *mm) | |
480 | { | |
481 | return __get_dumpable(mm->flags); | |
482 | } | |
483 | ||
3cb4a0bb KH |
484 | /* coredump filter bits */ |
485 | #define MMF_DUMP_ANON_PRIVATE 2 | |
486 | #define MMF_DUMP_ANON_SHARED 3 | |
487 | #define MMF_DUMP_MAPPED_PRIVATE 4 | |
488 | #define MMF_DUMP_MAPPED_SHARED 5 | |
82df3973 | 489 | #define MMF_DUMP_ELF_HEADERS 6 |
e575f111 KM |
490 | #define MMF_DUMP_HUGETLB_PRIVATE 7 |
491 | #define MMF_DUMP_HUGETLB_SHARED 8 | |
5037835c RZ |
492 | #define MMF_DUMP_DAX_PRIVATE 9 |
493 | #define MMF_DUMP_DAX_SHARED 10 | |
f8af4da3 | 494 | |
3cb4a0bb | 495 | #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS |
5037835c | 496 | #define MMF_DUMP_FILTER_BITS 9 |
3cb4a0bb KH |
497 | #define MMF_DUMP_FILTER_MASK \ |
498 | (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) | |
499 | #define MMF_DUMP_FILTER_DEFAULT \ | |
e575f111 | 500 | ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ |
656eb2cd RM |
501 | (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) |
502 | ||
503 | #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS | |
504 | # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) | |
505 | #else | |
506 | # define MMF_DUMP_MASK_DEFAULT_ELF 0 | |
507 | #endif | |
f8af4da3 HD |
508 | /* leave room for more dump flags */ |
509 | #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */ | |
ba76149f | 510 | #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */ |
3fb4afd9 SK |
511 | /* |
512 | * This one-shot flag is dropped due to necessity of changing exe once again | |
513 | * on NFS restore | |
514 | */ | |
515 | //#define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */ | |
f8af4da3 | 516 | |
9f68f672 ON |
517 | #define MMF_HAS_UPROBES 19 /* has uprobes */ |
518 | #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */ | |
862e3073 | 519 | #define MMF_OOM_SKIP 21 /* mm is of no interest for the OOM killer */ |
3f70dc38 | 520 | #define MMF_UNSTABLE 22 /* mm is unstable for copy_from_user */ |
6fcb52a5 | 521 | #define MMF_HUGE_ZERO_PAGE 23 /* mm has ever used the global huge zero page */ |
f8ac4ec9 | 522 | |
f8af4da3 | 523 | #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) |
6c5d5238 | 524 | |
1da177e4 LT |
525 | struct sighand_struct { |
526 | atomic_t count; | |
527 | struct k_sigaction action[_NSIG]; | |
528 | spinlock_t siglock; | |
b8fceee1 | 529 | wait_queue_head_t signalfd_wqh; |
1da177e4 LT |
530 | }; |
531 | ||
0e464814 | 532 | struct pacct_struct { |
f6ec29a4 KK |
533 | int ac_flag; |
534 | long ac_exitcode; | |
0e464814 | 535 | unsigned long ac_mem; |
d4bc42af | 536 | u64 ac_utime, ac_stime; |
77787bfb | 537 | unsigned long ac_minflt, ac_majflt; |
0e464814 KK |
538 | }; |
539 | ||
42c4ab41 | 540 | struct cpu_itimer { |
858cf3a8 FW |
541 | u64 expires; |
542 | u64 incr; | |
42c4ab41 SG |
543 | }; |
544 | ||
d37f761d | 545 | /** |
9d7fb042 | 546 | * struct prev_cputime - snaphsot of system and user cputime |
d37f761d FW |
547 | * @utime: time spent in user mode |
548 | * @stime: time spent in system mode | |
9d7fb042 | 549 | * @lock: protects the above two fields |
d37f761d | 550 | * |
9d7fb042 PZ |
551 | * Stores previous user/system time values such that we can guarantee |
552 | * monotonicity. | |
d37f761d | 553 | */ |
9d7fb042 PZ |
554 | struct prev_cputime { |
555 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
5613fda9 FW |
556 | u64 utime; |
557 | u64 stime; | |
9d7fb042 PZ |
558 | raw_spinlock_t lock; |
559 | #endif | |
d37f761d FW |
560 | }; |
561 | ||
9d7fb042 PZ |
562 | static inline void prev_cputime_init(struct prev_cputime *prev) |
563 | { | |
564 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
565 | prev->utime = prev->stime = 0; | |
566 | raw_spin_lock_init(&prev->lock); | |
567 | #endif | |
568 | } | |
569 | ||
f06febc9 FM |
570 | /** |
571 | * struct task_cputime - collected CPU time counts | |
5613fda9 FW |
572 | * @utime: time spent in user mode, in nanoseconds |
573 | * @stime: time spent in kernel mode, in nanoseconds | |
f06febc9 | 574 | * @sum_exec_runtime: total time spent on the CPU, in nanoseconds |
5ce73a4a | 575 | * |
9d7fb042 PZ |
576 | * This structure groups together three kinds of CPU time that are tracked for |
577 | * threads and thread groups. Most things considering CPU time want to group | |
578 | * these counts together and treat all three of them in parallel. | |
f06febc9 FM |
579 | */ |
580 | struct task_cputime { | |
5613fda9 FW |
581 | u64 utime; |
582 | u64 stime; | |
f06febc9 FM |
583 | unsigned long long sum_exec_runtime; |
584 | }; | |
9d7fb042 | 585 | |
f06febc9 | 586 | /* Alternate field names when used to cache expirations. */ |
f06febc9 | 587 | #define virt_exp utime |
9d7fb042 | 588 | #define prof_exp stime |
f06febc9 FM |
589 | #define sched_exp sum_exec_runtime |
590 | ||
971e8a98 JL |
591 | /* |
592 | * This is the atomic variant of task_cputime, which can be used for | |
593 | * storing and updating task_cputime statistics without locking. | |
594 | */ | |
595 | struct task_cputime_atomic { | |
596 | atomic64_t utime; | |
597 | atomic64_t stime; | |
598 | atomic64_t sum_exec_runtime; | |
599 | }; | |
600 | ||
601 | #define INIT_CPUTIME_ATOMIC \ | |
602 | (struct task_cputime_atomic) { \ | |
603 | .utime = ATOMIC64_INIT(0), \ | |
604 | .stime = ATOMIC64_INIT(0), \ | |
605 | .sum_exec_runtime = ATOMIC64_INIT(0), \ | |
606 | } | |
607 | ||
609ca066 | 608 | #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED) |
a233f112 | 609 | |
c99e6efe | 610 | /* |
87dcbc06 PZ |
611 | * Disable preemption until the scheduler is running -- use an unconditional |
612 | * value so that it also works on !PREEMPT_COUNT kernels. | |
d86ee480 | 613 | * |
87dcbc06 | 614 | * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count(). |
c99e6efe | 615 | */ |
87dcbc06 | 616 | #define INIT_PREEMPT_COUNT PREEMPT_OFFSET |
a233f112 | 617 | |
c99e6efe | 618 | /* |
609ca066 PZ |
619 | * Initial preempt_count value; reflects the preempt_count schedule invariant |
620 | * which states that during context switches: | |
d86ee480 | 621 | * |
609ca066 PZ |
622 | * preempt_count() == 2*PREEMPT_DISABLE_OFFSET |
623 | * | |
624 | * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels. | |
625 | * Note: See finish_task_switch(). | |
c99e6efe | 626 | */ |
609ca066 | 627 | #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED) |
c99e6efe | 628 | |
f06febc9 | 629 | /** |
4cd4c1b4 | 630 | * struct thread_group_cputimer - thread group interval timer counts |
920ce39f | 631 | * @cputime_atomic: atomic thread group interval timers. |
d5c373eb JL |
632 | * @running: true when there are timers running and |
633 | * @cputime_atomic receives updates. | |
c8d75aa4 JL |
634 | * @checking_timer: true when a thread in the group is in the |
635 | * process of checking for thread group timers. | |
f06febc9 FM |
636 | * |
637 | * This structure contains the version of task_cputime, above, that is | |
4cd4c1b4 | 638 | * used for thread group CPU timer calculations. |
f06febc9 | 639 | */ |
4cd4c1b4 | 640 | struct thread_group_cputimer { |
71107445 | 641 | struct task_cputime_atomic cputime_atomic; |
d5c373eb | 642 | bool running; |
c8d75aa4 | 643 | bool checking_timer; |
f06febc9 | 644 | }; |
f06febc9 | 645 | |
4714d1d3 | 646 | #include <linux/rwsem.h> |
5091faa4 MG |
647 | struct autogroup; |
648 | ||
1da177e4 | 649 | /* |
e815f0a8 | 650 | * NOTE! "signal_struct" does not have its own |
1da177e4 LT |
651 | * locking, because a shared signal_struct always |
652 | * implies a shared sighand_struct, so locking | |
653 | * sighand_struct is always a proper superset of | |
654 | * the locking of signal_struct. | |
655 | */ | |
656 | struct signal_struct { | |
ea6d290c | 657 | atomic_t sigcnt; |
1da177e4 | 658 | atomic_t live; |
b3ac022c | 659 | int nr_threads; |
0c740d0a | 660 | struct list_head thread_head; |
1da177e4 LT |
661 | |
662 | wait_queue_head_t wait_chldexit; /* for wait4() */ | |
663 | ||
664 | /* current thread group signal load-balancing target: */ | |
36c8b586 | 665 | struct task_struct *curr_target; |
1da177e4 LT |
666 | |
667 | /* shared signal handling: */ | |
668 | struct sigpending shared_pending; | |
669 | ||
670 | /* thread group exit support */ | |
671 | int group_exit_code; | |
672 | /* overloaded: | |
673 | * - notify group_exit_task when ->count is equal to notify_count | |
674 | * - everyone except group_exit_task is stopped during signal delivery | |
675 | * of fatal signals, group_exit_task processes the signal. | |
676 | */ | |
1da177e4 | 677 | int notify_count; |
07dd20e0 | 678 | struct task_struct *group_exit_task; |
1da177e4 LT |
679 | |
680 | /* thread group stop support, overloads group_exit_code too */ | |
681 | int group_stop_count; | |
682 | unsigned int flags; /* see SIGNAL_* flags below */ | |
683 | ||
ebec18a6 LP |
684 | /* |
685 | * PR_SET_CHILD_SUBREAPER marks a process, like a service | |
686 | * manager, to re-parent orphan (double-forking) child processes | |
687 | * to this process instead of 'init'. The service manager is | |
688 | * able to receive SIGCHLD signals and is able to investigate | |
689 | * the process until it calls wait(). All children of this | |
690 | * process will inherit a flag if they should look for a | |
691 | * child_subreaper process at exit. | |
692 | */ | |
693 | unsigned int is_child_subreaper:1; | |
694 | unsigned int has_child_subreaper:1; | |
695 | ||
b18b6a9c NP |
696 | #ifdef CONFIG_POSIX_TIMERS |
697 | ||
1da177e4 | 698 | /* POSIX.1b Interval Timers */ |
5ed67f05 PE |
699 | int posix_timer_id; |
700 | struct list_head posix_timers; | |
1da177e4 LT |
701 | |
702 | /* ITIMER_REAL timer for the process */ | |
2ff678b8 TG |
703 | struct hrtimer real_timer; |
704 | ktime_t it_real_incr; | |
1da177e4 | 705 | |
42c4ab41 SG |
706 | /* |
707 | * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use | |
708 | * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these | |
709 | * values are defined to 0 and 1 respectively | |
710 | */ | |
711 | struct cpu_itimer it[2]; | |
1da177e4 | 712 | |
f06febc9 | 713 | /* |
4cd4c1b4 PZ |
714 | * Thread group totals for process CPU timers. |
715 | * See thread_group_cputimer(), et al, for details. | |
f06febc9 | 716 | */ |
4cd4c1b4 | 717 | struct thread_group_cputimer cputimer; |
f06febc9 FM |
718 | |
719 | /* Earliest-expiration cache. */ | |
720 | struct task_cputime cputime_expires; | |
721 | ||
b18b6a9c NP |
722 | struct list_head cpu_timers[3]; |
723 | ||
724 | #endif | |
725 | ||
726 | struct pid *leader_pid; | |
727 | ||
d027d45d | 728 | #ifdef CONFIG_NO_HZ_FULL |
f009a7a7 | 729 | atomic_t tick_dep_mask; |
d027d45d FW |
730 | #endif |
731 | ||
ab521dc0 | 732 | struct pid *tty_old_pgrp; |
1ec320af | 733 | |
1da177e4 LT |
734 | /* boolean value for session group leader */ |
735 | int leader; | |
736 | ||
737 | struct tty_struct *tty; /* NULL if no tty */ | |
738 | ||
5091faa4 MG |
739 | #ifdef CONFIG_SCHED_AUTOGROUP |
740 | struct autogroup *autogroup; | |
741 | #endif | |
1da177e4 LT |
742 | /* |
743 | * Cumulative resource counters for dead threads in the group, | |
744 | * and for reaped dead child processes forked by this group. | |
745 | * Live threads maintain their own counters and add to these | |
746 | * in __exit_signal, except for the group leader. | |
747 | */ | |
e78c3496 | 748 | seqlock_t stats_lock; |
5613fda9 | 749 | u64 utime, stime, cutime, cstime; |
16a6d9be FW |
750 | u64 gtime; |
751 | u64 cgtime; | |
9d7fb042 | 752 | struct prev_cputime prev_cputime; |
1da177e4 LT |
753 | unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; |
754 | unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; | |
6eaeeaba | 755 | unsigned long inblock, oublock, cinblock, coublock; |
1f10206c | 756 | unsigned long maxrss, cmaxrss; |
940389b8 | 757 | struct task_io_accounting ioac; |
1da177e4 | 758 | |
32bd671d PZ |
759 | /* |
760 | * Cumulative ns of schedule CPU time fo dead threads in the | |
761 | * group, not including a zombie group leader, (This only differs | |
762 | * from jiffies_to_ns(utime + stime) if sched_clock uses something | |
763 | * other than jiffies.) | |
764 | */ | |
765 | unsigned long long sum_sched_runtime; | |
766 | ||
1da177e4 LT |
767 | /* |
768 | * We don't bother to synchronize most readers of this at all, | |
769 | * because there is no reader checking a limit that actually needs | |
770 | * to get both rlim_cur and rlim_max atomically, and either one | |
771 | * alone is a single word that can safely be read normally. | |
772 | * getrlimit/setrlimit use task_lock(current->group_leader) to | |
773 | * protect this instead of the siglock, because they really | |
774 | * have no need to disable irqs. | |
775 | */ | |
776 | struct rlimit rlim[RLIM_NLIMITS]; | |
777 | ||
0e464814 KK |
778 | #ifdef CONFIG_BSD_PROCESS_ACCT |
779 | struct pacct_struct pacct; /* per-process accounting information */ | |
780 | #endif | |
ad4ecbcb | 781 | #ifdef CONFIG_TASKSTATS |
ad4ecbcb SN |
782 | struct taskstats *stats; |
783 | #endif | |
522ed776 MT |
784 | #ifdef CONFIG_AUDIT |
785 | unsigned audit_tty; | |
786 | struct tty_audit_buf *tty_audit_buf; | |
787 | #endif | |
28b83c51 | 788 | |
c96fc2d8 TH |
789 | /* |
790 | * Thread is the potential origin of an oom condition; kill first on | |
791 | * oom | |
792 | */ | |
793 | bool oom_flag_origin; | |
a9c58b90 DR |
794 | short oom_score_adj; /* OOM kill score adjustment */ |
795 | short oom_score_adj_min; /* OOM kill score adjustment min value. | |
796 | * Only settable by CAP_SYS_RESOURCE. */ | |
26db62f1 MH |
797 | struct mm_struct *oom_mm; /* recorded mm when the thread group got |
798 | * killed by the oom killer */ | |
9b1bf12d KM |
799 | |
800 | struct mutex cred_guard_mutex; /* guard against foreign influences on | |
801 | * credential calculations | |
802 | * (notably. ptrace) */ | |
1da177e4 LT |
803 | }; |
804 | ||
805 | /* | |
806 | * Bits in flags field of signal_struct. | |
807 | */ | |
808 | #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ | |
ee77f075 ON |
809 | #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */ |
810 | #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */ | |
403bad72 | 811 | #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */ |
e4420551 ON |
812 | /* |
813 | * Pending notifications to parent. | |
814 | */ | |
815 | #define SIGNAL_CLD_STOPPED 0x00000010 | |
816 | #define SIGNAL_CLD_CONTINUED 0x00000020 | |
817 | #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) | |
1da177e4 | 818 | |
fae5fa44 ON |
819 | #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ |
820 | ||
2d39b3cd JI |
821 | #define SIGNAL_STOP_MASK (SIGNAL_CLD_MASK | SIGNAL_STOP_STOPPED | \ |
822 | SIGNAL_STOP_CONTINUED) | |
823 | ||
824 | static inline void signal_set_stop_flags(struct signal_struct *sig, | |
825 | unsigned int flags) | |
826 | { | |
827 | WARN_ON(sig->flags & (SIGNAL_GROUP_EXIT|SIGNAL_GROUP_COREDUMP)); | |
828 | sig->flags = (sig->flags & ~SIGNAL_STOP_MASK) | flags; | |
829 | } | |
830 | ||
ed5d2cac ON |
831 | /* If true, all threads except ->group_exit_task have pending SIGKILL */ |
832 | static inline int signal_group_exit(const struct signal_struct *sig) | |
833 | { | |
834 | return (sig->flags & SIGNAL_GROUP_EXIT) || | |
835 | (sig->group_exit_task != NULL); | |
836 | } | |
837 | ||
1da177e4 LT |
838 | /* |
839 | * Some day this will be a full-fledged user tracking system.. | |
840 | */ | |
841 | struct user_struct { | |
842 | atomic_t __count; /* reference count */ | |
843 | atomic_t processes; /* How many processes does this user have? */ | |
1da177e4 | 844 | atomic_t sigpending; /* How many pending signals does this user have? */ |
4afeff85 EP |
845 | #ifdef CONFIG_FANOTIFY |
846 | atomic_t fanotify_listeners; | |
847 | #endif | |
7ef9964e | 848 | #ifdef CONFIG_EPOLL |
52bd19f7 | 849 | atomic_long_t epoll_watches; /* The number of file descriptors currently watched */ |
7ef9964e | 850 | #endif |
970a8645 | 851 | #ifdef CONFIG_POSIX_MQUEUE |
1da177e4 LT |
852 | /* protected by mq_lock */ |
853 | unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ | |
970a8645 | 854 | #endif |
1da177e4 | 855 | unsigned long locked_shm; /* How many pages of mlocked shm ? */ |
712f4aad | 856 | unsigned long unix_inflight; /* How many files in flight in unix sockets */ |
759c0114 | 857 | atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */ |
1da177e4 LT |
858 | |
859 | #ifdef CONFIG_KEYS | |
860 | struct key *uid_keyring; /* UID specific keyring */ | |
861 | struct key *session_keyring; /* UID's default session keyring */ | |
862 | #endif | |
863 | ||
864 | /* Hash table maintenance information */ | |
735de223 | 865 | struct hlist_node uidhash_node; |
7b44ab97 | 866 | kuid_t uid; |
24e377a8 | 867 | |
aaac3ba9 | 868 | #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL) |
789f90fc PZ |
869 | atomic_long_t locked_vm; |
870 | #endif | |
1da177e4 LT |
871 | }; |
872 | ||
eb41d946 | 873 | extern int uids_sysfs_init(void); |
5cb350ba | 874 | |
7b44ab97 | 875 | extern struct user_struct *find_user(kuid_t); |
1da177e4 LT |
876 | |
877 | extern struct user_struct root_user; | |
878 | #define INIT_USER (&root_user) | |
879 | ||
b6dff3ec | 880 | |
1da177e4 LT |
881 | struct backing_dev_info; |
882 | struct reclaim_state; | |
883 | ||
f6db8347 | 884 | #ifdef CONFIG_SCHED_INFO |
1da177e4 LT |
885 | struct sched_info { |
886 | /* cumulative counters */ | |
2d72376b | 887 | unsigned long pcount; /* # of times run on this cpu */ |
9c2c4802 | 888 | unsigned long long run_delay; /* time spent waiting on a runqueue */ |
1da177e4 LT |
889 | |
890 | /* timestamps */ | |
172ba844 BS |
891 | unsigned long long last_arrival,/* when we last ran on a cpu */ |
892 | last_queued; /* when we were last queued to run */ | |
1da177e4 | 893 | }; |
f6db8347 | 894 | #endif /* CONFIG_SCHED_INFO */ |
1da177e4 | 895 | |
ca74e92b SN |
896 | #ifdef CONFIG_TASK_DELAY_ACCT |
897 | struct task_delay_info { | |
898 | spinlock_t lock; | |
899 | unsigned int flags; /* Private per-task flags */ | |
900 | ||
901 | /* For each stat XXX, add following, aligned appropriately | |
902 | * | |
903 | * struct timespec XXX_start, XXX_end; | |
904 | * u64 XXX_delay; | |
905 | * u32 XXX_count; | |
906 | * | |
907 | * Atomicity of updates to XXX_delay, XXX_count protected by | |
908 | * single lock above (split into XXX_lock if contention is an issue). | |
909 | */ | |
0ff92245 SN |
910 | |
911 | /* | |
912 | * XXX_count is incremented on every XXX operation, the delay | |
913 | * associated with the operation is added to XXX_delay. | |
914 | * XXX_delay contains the accumulated delay time in nanoseconds. | |
915 | */ | |
9667a23d | 916 | u64 blkio_start; /* Shared by blkio, swapin */ |
0ff92245 SN |
917 | u64 blkio_delay; /* wait for sync block io completion */ |
918 | u64 swapin_delay; /* wait for swapin block io completion */ | |
919 | u32 blkio_count; /* total count of the number of sync block */ | |
920 | /* io operations performed */ | |
921 | u32 swapin_count; /* total count of the number of swapin block */ | |
922 | /* io operations performed */ | |
873b4771 | 923 | |
9667a23d | 924 | u64 freepages_start; |
873b4771 KK |
925 | u64 freepages_delay; /* wait for memory reclaim */ |
926 | u32 freepages_count; /* total count of memory reclaim */ | |
ca74e92b | 927 | }; |
52f17b6c CS |
928 | #endif /* CONFIG_TASK_DELAY_ACCT */ |
929 | ||
930 | static inline int sched_info_on(void) | |
931 | { | |
932 | #ifdef CONFIG_SCHEDSTATS | |
933 | return 1; | |
934 | #elif defined(CONFIG_TASK_DELAY_ACCT) | |
935 | extern int delayacct_on; | |
936 | return delayacct_on; | |
937 | #else | |
938 | return 0; | |
ca74e92b | 939 | #endif |
52f17b6c | 940 | } |
ca74e92b | 941 | |
cb251765 MG |
942 | #ifdef CONFIG_SCHEDSTATS |
943 | void force_schedstat_enabled(void); | |
944 | #endif | |
945 | ||
6ecdd749 YD |
946 | /* |
947 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
948 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
949 | * | |
950 | * We define a basic fixed point arithmetic range, and then formalize | |
951 | * all these metrics based on that basic range. | |
952 | */ | |
953 | # define SCHED_FIXEDPOINT_SHIFT 10 | |
954 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
955 | ||
76751049 PZ |
956 | /* |
957 | * Wake-queues are lists of tasks with a pending wakeup, whose | |
958 | * callers have already marked the task as woken internally, | |
959 | * and can thus carry on. A common use case is being able to | |
960 | * do the wakeups once the corresponding user lock as been | |
961 | * released. | |
962 | * | |
963 | * We hold reference to each task in the list across the wakeup, | |
964 | * thus guaranteeing that the memory is still valid by the time | |
965 | * the actual wakeups are performed in wake_up_q(). | |
966 | * | |
967 | * One per task suffices, because there's never a need for a task to be | |
968 | * in two wake queues simultaneously; it is forbidden to abandon a task | |
969 | * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is | |
970 | * already in a wake queue, the wakeup will happen soon and the second | |
971 | * waker can just skip it. | |
972 | * | |
194a6b5b | 973 | * The DEFINE_WAKE_Q macro declares and initializes the list head. |
76751049 | 974 | * wake_up_q() does NOT reinitialize the list; it's expected to be |
0754445d DB |
975 | * called near the end of a function. Otherwise, the list can be |
976 | * re-initialized for later re-use by wake_q_init(). | |
76751049 PZ |
977 | * |
978 | * Note that this can cause spurious wakeups. schedule() callers | |
979 | * must ensure the call is done inside a loop, confirming that the | |
980 | * wakeup condition has in fact occurred. | |
981 | */ | |
982 | struct wake_q_node { | |
983 | struct wake_q_node *next; | |
984 | }; | |
985 | ||
986 | struct wake_q_head { | |
987 | struct wake_q_node *first; | |
988 | struct wake_q_node **lastp; | |
989 | }; | |
990 | ||
991 | #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01) | |
992 | ||
194a6b5b | 993 | #define DEFINE_WAKE_Q(name) \ |
76751049 PZ |
994 | struct wake_q_head name = { WAKE_Q_TAIL, &name.first } |
995 | ||
bcc9a76d WL |
996 | static inline void wake_q_init(struct wake_q_head *head) |
997 | { | |
998 | head->first = WAKE_Q_TAIL; | |
999 | head->lastp = &head->first; | |
1000 | } | |
1001 | ||
76751049 PZ |
1002 | extern void wake_q_add(struct wake_q_head *head, |
1003 | struct task_struct *task); | |
1004 | extern void wake_up_q(struct wake_q_head *head); | |
1005 | ||
1da177e4 | 1006 | struct io_context; /* See blkdev.h */ |
1da177e4 | 1007 | |
1da177e4 | 1008 | |
383f2835 | 1009 | #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK |
36c8b586 | 1010 | extern void prefetch_stack(struct task_struct *t); |
383f2835 KC |
1011 | #else |
1012 | static inline void prefetch_stack(struct task_struct *t) { } | |
1013 | #endif | |
1da177e4 LT |
1014 | |
1015 | struct audit_context; /* See audit.c */ | |
1016 | struct mempolicy; | |
b92ce558 | 1017 | struct pipe_inode_info; |
4865ecf1 | 1018 | struct uts_namespace; |
1da177e4 | 1019 | |
20b8a59f | 1020 | struct load_weight { |
9dbdb155 PZ |
1021 | unsigned long weight; |
1022 | u32 inv_weight; | |
20b8a59f IM |
1023 | }; |
1024 | ||
9d89c257 | 1025 | /* |
7b595334 YD |
1026 | * The load_avg/util_avg accumulates an infinite geometric series |
1027 | * (see __update_load_avg() in kernel/sched/fair.c). | |
1028 | * | |
1029 | * [load_avg definition] | |
1030 | * | |
1031 | * load_avg = runnable% * scale_load_down(load) | |
1032 | * | |
1033 | * where runnable% is the time ratio that a sched_entity is runnable. | |
1034 | * For cfs_rq, it is the aggregated load_avg of all runnable and | |
9d89c257 | 1035 | * blocked sched_entities. |
7b595334 YD |
1036 | * |
1037 | * load_avg may also take frequency scaling into account: | |
1038 | * | |
1039 | * load_avg = runnable% * scale_load_down(load) * freq% | |
1040 | * | |
1041 | * where freq% is the CPU frequency normalized to the highest frequency. | |
1042 | * | |
1043 | * [util_avg definition] | |
1044 | * | |
1045 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
1046 | * | |
1047 | * where running% is the time ratio that a sched_entity is running on | |
1048 | * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable | |
1049 | * and blocked sched_entities. | |
1050 | * | |
1051 | * util_avg may also factor frequency scaling and CPU capacity scaling: | |
1052 | * | |
1053 | * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity% | |
1054 | * | |
1055 | * where freq% is the same as above, and capacity% is the CPU capacity | |
1056 | * normalized to the greatest capacity (due to uarch differences, etc). | |
1057 | * | |
1058 | * N.B., the above ratios (runnable%, running%, freq%, and capacity%) | |
1059 | * themselves are in the range of [0, 1]. To do fixed point arithmetics, | |
1060 | * we therefore scale them to as large a range as necessary. This is for | |
1061 | * example reflected by util_avg's SCHED_CAPACITY_SCALE. | |
1062 | * | |
1063 | * [Overflow issue] | |
1064 | * | |
1065 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
1066 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
1067 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
1068 | * | |
1069 | * For all other cases (including 32-bit kernels), struct load_weight's | |
1070 | * weight will overflow first before we do, because: | |
1071 | * | |
1072 | * Max(load_avg) <= Max(load.weight) | |
1073 | * | |
1074 | * Then it is the load_weight's responsibility to consider overflow | |
1075 | * issues. | |
9d89c257 | 1076 | */ |
9d85f21c | 1077 | struct sched_avg { |
9d89c257 YD |
1078 | u64 last_update_time, load_sum; |
1079 | u32 util_sum, period_contrib; | |
1080 | unsigned long load_avg, util_avg; | |
9d85f21c PT |
1081 | }; |
1082 | ||
94c18227 | 1083 | #ifdef CONFIG_SCHEDSTATS |
41acab88 | 1084 | struct sched_statistics { |
20b8a59f | 1085 | u64 wait_start; |
94c18227 | 1086 | u64 wait_max; |
6d082592 AV |
1087 | u64 wait_count; |
1088 | u64 wait_sum; | |
8f0dfc34 AV |
1089 | u64 iowait_count; |
1090 | u64 iowait_sum; | |
94c18227 | 1091 | |
20b8a59f | 1092 | u64 sleep_start; |
20b8a59f | 1093 | u64 sleep_max; |
94c18227 IM |
1094 | s64 sum_sleep_runtime; |
1095 | ||
1096 | u64 block_start; | |
20b8a59f IM |
1097 | u64 block_max; |
1098 | u64 exec_max; | |
eba1ed4b | 1099 | u64 slice_max; |
cc367732 | 1100 | |
cc367732 IM |
1101 | u64 nr_migrations_cold; |
1102 | u64 nr_failed_migrations_affine; | |
1103 | u64 nr_failed_migrations_running; | |
1104 | u64 nr_failed_migrations_hot; | |
1105 | u64 nr_forced_migrations; | |
cc367732 IM |
1106 | |
1107 | u64 nr_wakeups; | |
1108 | u64 nr_wakeups_sync; | |
1109 | u64 nr_wakeups_migrate; | |
1110 | u64 nr_wakeups_local; | |
1111 | u64 nr_wakeups_remote; | |
1112 | u64 nr_wakeups_affine; | |
1113 | u64 nr_wakeups_affine_attempts; | |
1114 | u64 nr_wakeups_passive; | |
1115 | u64 nr_wakeups_idle; | |
41acab88 LDM |
1116 | }; |
1117 | #endif | |
1118 | ||
1119 | struct sched_entity { | |
1120 | struct load_weight load; /* for load-balancing */ | |
1121 | struct rb_node run_node; | |
1122 | struct list_head group_node; | |
1123 | unsigned int on_rq; | |
1124 | ||
1125 | u64 exec_start; | |
1126 | u64 sum_exec_runtime; | |
1127 | u64 vruntime; | |
1128 | u64 prev_sum_exec_runtime; | |
1129 | ||
41acab88 LDM |
1130 | u64 nr_migrations; |
1131 | ||
41acab88 LDM |
1132 | #ifdef CONFIG_SCHEDSTATS |
1133 | struct sched_statistics statistics; | |
94c18227 IM |
1134 | #endif |
1135 | ||
20b8a59f | 1136 | #ifdef CONFIG_FAIR_GROUP_SCHED |
fed14d45 | 1137 | int depth; |
20b8a59f IM |
1138 | struct sched_entity *parent; |
1139 | /* rq on which this entity is (to be) queued: */ | |
1140 | struct cfs_rq *cfs_rq; | |
1141 | /* rq "owned" by this entity/group: */ | |
1142 | struct cfs_rq *my_q; | |
1143 | #endif | |
8bd75c77 | 1144 | |
141965c7 | 1145 | #ifdef CONFIG_SMP |
5a107804 JO |
1146 | /* |
1147 | * Per entity load average tracking. | |
1148 | * | |
1149 | * Put into separate cache line so it does not | |
1150 | * collide with read-mostly values above. | |
1151 | */ | |
1152 | struct sched_avg avg ____cacheline_aligned_in_smp; | |
9d85f21c | 1153 | #endif |
20b8a59f | 1154 | }; |
70b97a7f | 1155 | |
fa717060 PZ |
1156 | struct sched_rt_entity { |
1157 | struct list_head run_list; | |
78f2c7db | 1158 | unsigned long timeout; |
57d2aa00 | 1159 | unsigned long watchdog_stamp; |
bee367ed | 1160 | unsigned int time_slice; |
ff77e468 PZ |
1161 | unsigned short on_rq; |
1162 | unsigned short on_list; | |
6f505b16 | 1163 | |
58d6c2d7 | 1164 | struct sched_rt_entity *back; |
052f1dc7 | 1165 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 PZ |
1166 | struct sched_rt_entity *parent; |
1167 | /* rq on which this entity is (to be) queued: */ | |
1168 | struct rt_rq *rt_rq; | |
1169 | /* rq "owned" by this entity/group: */ | |
1170 | struct rt_rq *my_q; | |
1171 | #endif | |
fa717060 PZ |
1172 | }; |
1173 | ||
aab03e05 DF |
1174 | struct sched_dl_entity { |
1175 | struct rb_node rb_node; | |
1176 | ||
1177 | /* | |
1178 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 1179 | * during sched_setattr(), they will remain the same until |
1180 | * the next sched_setattr(). | |
aab03e05 DF |
1181 | */ |
1182 | u64 dl_runtime; /* maximum runtime for each instance */ | |
1183 | u64 dl_deadline; /* relative deadline of each instance */ | |
755378a4 | 1184 | u64 dl_period; /* separation of two instances (period) */ |
332ac17e | 1185 | u64 dl_bw; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
1186 | |
1187 | /* | |
1188 | * Actual scheduling parameters. Initialized with the values above, | |
1189 | * they are continously updated during task execution. Note that | |
1190 | * the remaining runtime could be < 0 in case we are in overrun. | |
1191 | */ | |
1192 | s64 runtime; /* remaining runtime for this instance */ | |
1193 | u64 deadline; /* absolute deadline for this instance */ | |
1194 | unsigned int flags; /* specifying the scheduler behaviour */ | |
1195 | ||
1196 | /* | |
1197 | * Some bool flags: | |
1198 | * | |
1199 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
1200 | * task has to wait for a replenishment to be performed at the | |
1201 | * next firing of dl_timer. | |
1202 | * | |
2d3d891d DF |
1203 | * @dl_boosted tells if we are boosted due to DI. If so we are |
1204 | * outside bandwidth enforcement mechanism (but only until we | |
5bfd126e JL |
1205 | * exit the critical section); |
1206 | * | |
1207 | * @dl_yielded tells if task gave up the cpu before consuming | |
1208 | * all its available runtime during the last job. | |
aab03e05 | 1209 | */ |
72f9f3fd | 1210 | int dl_throttled, dl_boosted, dl_yielded; |
aab03e05 DF |
1211 | |
1212 | /* | |
1213 | * Bandwidth enforcement timer. Each -deadline task has its | |
1214 | * own bandwidth to be enforced, thus we need one timer per task. | |
1215 | */ | |
1216 | struct hrtimer dl_timer; | |
1217 | }; | |
8bd75c77 | 1218 | |
1d082fd0 PM |
1219 | union rcu_special { |
1220 | struct { | |
8203d6d0 PM |
1221 | u8 blocked; |
1222 | u8 need_qs; | |
1223 | u8 exp_need_qs; | |
1224 | u8 pad; /* Otherwise the compiler can store garbage here. */ | |
1225 | } b; /* Bits. */ | |
1226 | u32 s; /* Set of bits. */ | |
1d082fd0 | 1227 | }; |
86848966 PM |
1228 | struct rcu_node; |
1229 | ||
8dc85d54 PZ |
1230 | enum perf_event_task_context { |
1231 | perf_invalid_context = -1, | |
1232 | perf_hw_context = 0, | |
89a1e187 | 1233 | perf_sw_context, |
8dc85d54 PZ |
1234 | perf_nr_task_contexts, |
1235 | }; | |
1236 | ||
72b252ae MG |
1237 | /* Track pages that require TLB flushes */ |
1238 | struct tlbflush_unmap_batch { | |
1239 | /* | |
1240 | * Each bit set is a CPU that potentially has a TLB entry for one of | |
1241 | * the PFNs being flushed. See set_tlb_ubc_flush_pending(). | |
1242 | */ | |
1243 | struct cpumask cpumask; | |
1244 | ||
1245 | /* True if any bit in cpumask is set */ | |
1246 | bool flush_required; | |
d950c947 MG |
1247 | |
1248 | /* | |
1249 | * If true then the PTE was dirty when unmapped. The entry must be | |
1250 | * flushed before IO is initiated or a stale TLB entry potentially | |
1251 | * allows an update without redirtying the page. | |
1252 | */ | |
1253 | bool writable; | |
72b252ae MG |
1254 | }; |
1255 | ||
1da177e4 | 1256 | struct task_struct { |
c65eacbe AL |
1257 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1258 | /* | |
1259 | * For reasons of header soup (see current_thread_info()), this | |
1260 | * must be the first element of task_struct. | |
1261 | */ | |
1262 | struct thread_info thread_info; | |
1263 | #endif | |
1da177e4 | 1264 | volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ |
f7e4217b | 1265 | void *stack; |
1da177e4 | 1266 | atomic_t usage; |
97dc32cd WC |
1267 | unsigned int flags; /* per process flags, defined below */ |
1268 | unsigned int ptrace; | |
1da177e4 | 1269 | |
2dd73a4f | 1270 | #ifdef CONFIG_SMP |
fa14ff4a | 1271 | struct llist_node wake_entry; |
3ca7a440 | 1272 | int on_cpu; |
c65eacbe AL |
1273 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1274 | unsigned int cpu; /* current CPU */ | |
1275 | #endif | |
63b0e9ed | 1276 | unsigned int wakee_flips; |
62470419 | 1277 | unsigned long wakee_flip_decay_ts; |
63b0e9ed | 1278 | struct task_struct *last_wakee; |
ac66f547 PZ |
1279 | |
1280 | int wake_cpu; | |
2dd73a4f | 1281 | #endif |
fd2f4419 | 1282 | int on_rq; |
50e645a8 | 1283 | |
b29739f9 | 1284 | int prio, static_prio, normal_prio; |
c7aceaba | 1285 | unsigned int rt_priority; |
5522d5d5 | 1286 | const struct sched_class *sched_class; |
20b8a59f | 1287 | struct sched_entity se; |
fa717060 | 1288 | struct sched_rt_entity rt; |
8323f26c PZ |
1289 | #ifdef CONFIG_CGROUP_SCHED |
1290 | struct task_group *sched_task_group; | |
1291 | #endif | |
aab03e05 | 1292 | struct sched_dl_entity dl; |
1da177e4 | 1293 | |
e107be36 AK |
1294 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
1295 | /* list of struct preempt_notifier: */ | |
1296 | struct hlist_head preempt_notifiers; | |
1297 | #endif | |
1298 | ||
6c5c9341 | 1299 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
2056a782 | 1300 | unsigned int btrace_seq; |
6c5c9341 | 1301 | #endif |
1da177e4 | 1302 | |
97dc32cd | 1303 | unsigned int policy; |
29baa747 | 1304 | int nr_cpus_allowed; |
1da177e4 | 1305 | cpumask_t cpus_allowed; |
1da177e4 | 1306 | |
a57eb940 | 1307 | #ifdef CONFIG_PREEMPT_RCU |
e260be67 | 1308 | int rcu_read_lock_nesting; |
1d082fd0 | 1309 | union rcu_special rcu_read_unlock_special; |
f41d911f | 1310 | struct list_head rcu_node_entry; |
a57eb940 | 1311 | struct rcu_node *rcu_blocked_node; |
28f6569a | 1312 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
8315f422 PM |
1313 | #ifdef CONFIG_TASKS_RCU |
1314 | unsigned long rcu_tasks_nvcsw; | |
1315 | bool rcu_tasks_holdout; | |
1316 | struct list_head rcu_tasks_holdout_list; | |
176f8f7a | 1317 | int rcu_tasks_idle_cpu; |
8315f422 | 1318 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 1319 | |
f6db8347 | 1320 | #ifdef CONFIG_SCHED_INFO |
1da177e4 LT |
1321 | struct sched_info sched_info; |
1322 | #endif | |
1323 | ||
1324 | struct list_head tasks; | |
806c09a7 | 1325 | #ifdef CONFIG_SMP |
917b627d | 1326 | struct plist_node pushable_tasks; |
1baca4ce | 1327 | struct rb_node pushable_dl_tasks; |
806c09a7 | 1328 | #endif |
1da177e4 LT |
1329 | |
1330 | struct mm_struct *mm, *active_mm; | |
314ff785 IM |
1331 | |
1332 | /* Per-thread vma caching: */ | |
1333 | struct vmacache vmacache; | |
1334 | ||
34e55232 KH |
1335 | #if defined(SPLIT_RSS_COUNTING) |
1336 | struct task_rss_stat rss_stat; | |
1337 | #endif | |
1da177e4 | 1338 | /* task state */ |
97dc32cd | 1339 | int exit_state; |
1da177e4 LT |
1340 | int exit_code, exit_signal; |
1341 | int pdeath_signal; /* The signal sent when the parent dies */ | |
e7cc4173 | 1342 | unsigned long jobctl; /* JOBCTL_*, siglock protected */ |
9b89f6ba AE |
1343 | |
1344 | /* Used for emulating ABI behavior of previous Linux versions */ | |
97dc32cd | 1345 | unsigned int personality; |
9b89f6ba | 1346 | |
be958bdc | 1347 | /* scheduler bits, serialized by scheduler locks */ |
ca94c442 | 1348 | unsigned sched_reset_on_fork:1; |
a8e4f2ea | 1349 | unsigned sched_contributes_to_load:1; |
ff303e66 | 1350 | unsigned sched_migrated:1; |
b7e7ade3 | 1351 | unsigned sched_remote_wakeup:1; |
be958bdc PZ |
1352 | unsigned :0; /* force alignment to the next boundary */ |
1353 | ||
1354 | /* unserialized, strictly 'current' */ | |
1355 | unsigned in_execve:1; /* bit to tell LSMs we're in execve */ | |
1356 | unsigned in_iowait:1; | |
7e781418 AL |
1357 | #if !defined(TIF_RESTORE_SIGMASK) |
1358 | unsigned restore_sigmask:1; | |
1359 | #endif | |
626ebc41 TH |
1360 | #ifdef CONFIG_MEMCG |
1361 | unsigned memcg_may_oom:1; | |
127424c8 | 1362 | #ifndef CONFIG_SLOB |
6f185c29 VD |
1363 | unsigned memcg_kmem_skip_account:1; |
1364 | #endif | |
127424c8 | 1365 | #endif |
ff303e66 PZ |
1366 | #ifdef CONFIG_COMPAT_BRK |
1367 | unsigned brk_randomized:1; | |
1368 | #endif | |
6f185c29 | 1369 | |
1d4457f9 KC |
1370 | unsigned long atomic_flags; /* Flags needing atomic access. */ |
1371 | ||
f56141e3 AL |
1372 | struct restart_block restart_block; |
1373 | ||
1da177e4 LT |
1374 | pid_t pid; |
1375 | pid_t tgid; | |
0a425405 | 1376 | |
1314562a | 1377 | #ifdef CONFIG_CC_STACKPROTECTOR |
0a425405 AV |
1378 | /* Canary value for the -fstack-protector gcc feature */ |
1379 | unsigned long stack_canary; | |
1314562a | 1380 | #endif |
4d1d61a6 | 1381 | /* |
1da177e4 | 1382 | * pointers to (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 1383 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 1384 | * p->real_parent->pid) |
1da177e4 | 1385 | */ |
abd63bc3 KC |
1386 | struct task_struct __rcu *real_parent; /* real parent process */ |
1387 | struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */ | |
1da177e4 | 1388 | /* |
f470021a | 1389 | * children/sibling forms the list of my natural children |
1da177e4 LT |
1390 | */ |
1391 | struct list_head children; /* list of my children */ | |
1392 | struct list_head sibling; /* linkage in my parent's children list */ | |
1393 | struct task_struct *group_leader; /* threadgroup leader */ | |
1394 | ||
f470021a RM |
1395 | /* |
1396 | * ptraced is the list of tasks this task is using ptrace on. | |
1397 | * This includes both natural children and PTRACE_ATTACH targets. | |
1398 | * p->ptrace_entry is p's link on the p->parent->ptraced list. | |
1399 | */ | |
1400 | struct list_head ptraced; | |
1401 | struct list_head ptrace_entry; | |
1402 | ||
1da177e4 | 1403 | /* PID/PID hash table linkage. */ |
92476d7f | 1404 | struct pid_link pids[PIDTYPE_MAX]; |
47e65328 | 1405 | struct list_head thread_group; |
0c740d0a | 1406 | struct list_head thread_node; |
1da177e4 LT |
1407 | |
1408 | struct completion *vfork_done; /* for vfork() */ | |
1409 | int __user *set_child_tid; /* CLONE_CHILD_SETTID */ | |
1410 | int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ | |
1411 | ||
5613fda9 | 1412 | u64 utime, stime; |
40565b5a | 1413 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME |
5613fda9 | 1414 | u64 utimescaled, stimescaled; |
40565b5a | 1415 | #endif |
16a6d9be | 1416 | u64 gtime; |
9d7fb042 | 1417 | struct prev_cputime prev_cputime; |
6a61671b | 1418 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
b7ce2277 | 1419 | seqcount_t vtime_seqcount; |
6a61671b FW |
1420 | unsigned long long vtime_snap; |
1421 | enum { | |
7098c1ea FW |
1422 | /* Task is sleeping or running in a CPU with VTIME inactive */ |
1423 | VTIME_INACTIVE = 0, | |
1424 | /* Task runs in userspace in a CPU with VTIME active */ | |
6a61671b | 1425 | VTIME_USER, |
7098c1ea | 1426 | /* Task runs in kernelspace in a CPU with VTIME active */ |
6a61671b FW |
1427 | VTIME_SYS, |
1428 | } vtime_snap_whence; | |
d99ca3b9 | 1429 | #endif |
d027d45d FW |
1430 | |
1431 | #ifdef CONFIG_NO_HZ_FULL | |
f009a7a7 | 1432 | atomic_t tick_dep_mask; |
d027d45d | 1433 | #endif |
1da177e4 | 1434 | unsigned long nvcsw, nivcsw; /* context switch counts */ |
ccbf62d8 | 1435 | u64 start_time; /* monotonic time in nsec */ |
57e0be04 | 1436 | u64 real_start_time; /* boot based time in nsec */ |
1da177e4 LT |
1437 | /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ |
1438 | unsigned long min_flt, maj_flt; | |
1439 | ||
b18b6a9c | 1440 | #ifdef CONFIG_POSIX_TIMERS |
f06febc9 | 1441 | struct task_cputime cputime_expires; |
1da177e4 | 1442 | struct list_head cpu_timers[3]; |
b18b6a9c | 1443 | #endif |
1da177e4 LT |
1444 | |
1445 | /* process credentials */ | |
64b875f7 | 1446 | const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */ |
1b0ba1c9 | 1447 | const struct cred __rcu *real_cred; /* objective and real subjective task |
3b11a1de | 1448 | * credentials (COW) */ |
1b0ba1c9 | 1449 | const struct cred __rcu *cred; /* effective (overridable) subjective task |
3b11a1de | 1450 | * credentials (COW) */ |
36772092 PBG |
1451 | char comm[TASK_COMM_LEN]; /* executable name excluding path |
1452 | - access with [gs]et_task_comm (which lock | |
1453 | it with task_lock()) | |
221af7f8 | 1454 | - initialized normally by setup_new_exec */ |
1da177e4 | 1455 | /* file system info */ |
756daf26 | 1456 | struct nameidata *nameidata; |
3d5b6fcc | 1457 | #ifdef CONFIG_SYSVIPC |
1da177e4 LT |
1458 | /* ipc stuff */ |
1459 | struct sysv_sem sysvsem; | |
ab602f79 | 1460 | struct sysv_shm sysvshm; |
3d5b6fcc | 1461 | #endif |
e162b39a | 1462 | #ifdef CONFIG_DETECT_HUNG_TASK |
82a1fcb9 | 1463 | /* hung task detection */ |
82a1fcb9 IM |
1464 | unsigned long last_switch_count; |
1465 | #endif | |
1da177e4 LT |
1466 | /* filesystem information */ |
1467 | struct fs_struct *fs; | |
1468 | /* open file information */ | |
1469 | struct files_struct *files; | |
1651e14e | 1470 | /* namespaces */ |
ab516013 | 1471 | struct nsproxy *nsproxy; |
1da177e4 LT |
1472 | /* signal handlers */ |
1473 | struct signal_struct *signal; | |
1474 | struct sighand_struct *sighand; | |
1475 | ||
1476 | sigset_t blocked, real_blocked; | |
f3de272b | 1477 | sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ |
1da177e4 LT |
1478 | struct sigpending pending; |
1479 | ||
1480 | unsigned long sas_ss_sp; | |
1481 | size_t sas_ss_size; | |
2a742138 | 1482 | unsigned sas_ss_flags; |
2e01fabe | 1483 | |
67d12145 | 1484 | struct callback_head *task_works; |
e73f8959 | 1485 | |
1da177e4 | 1486 | struct audit_context *audit_context; |
bfef93a5 | 1487 | #ifdef CONFIG_AUDITSYSCALL |
e1760bd5 | 1488 | kuid_t loginuid; |
4746ec5b | 1489 | unsigned int sessionid; |
bfef93a5 | 1490 | #endif |
932ecebb | 1491 | struct seccomp seccomp; |
1da177e4 LT |
1492 | |
1493 | /* Thread group tracking */ | |
1494 | u32 parent_exec_id; | |
1495 | u32 self_exec_id; | |
58568d2a MX |
1496 | /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, |
1497 | * mempolicy */ | |
1da177e4 | 1498 | spinlock_t alloc_lock; |
1da177e4 | 1499 | |
b29739f9 | 1500 | /* Protection of the PI data structures: */ |
1d615482 | 1501 | raw_spinlock_t pi_lock; |
b29739f9 | 1502 | |
76751049 PZ |
1503 | struct wake_q_node wake_q; |
1504 | ||
23f78d4a IM |
1505 | #ifdef CONFIG_RT_MUTEXES |
1506 | /* PI waiters blocked on a rt_mutex held by this task */ | |
fb00aca4 PZ |
1507 | struct rb_root pi_waiters; |
1508 | struct rb_node *pi_waiters_leftmost; | |
23f78d4a IM |
1509 | /* Deadlock detection and priority inheritance handling */ |
1510 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
1511 | #endif |
1512 | ||
408894ee IM |
1513 | #ifdef CONFIG_DEBUG_MUTEXES |
1514 | /* mutex deadlock detection */ | |
1515 | struct mutex_waiter *blocked_on; | |
1516 | #endif | |
de30a2b3 IM |
1517 | #ifdef CONFIG_TRACE_IRQFLAGS |
1518 | unsigned int irq_events; | |
de30a2b3 | 1519 | unsigned long hardirq_enable_ip; |
de30a2b3 | 1520 | unsigned long hardirq_disable_ip; |
fa1452e8 | 1521 | unsigned int hardirq_enable_event; |
de30a2b3 | 1522 | unsigned int hardirq_disable_event; |
fa1452e8 HS |
1523 | int hardirqs_enabled; |
1524 | int hardirq_context; | |
de30a2b3 | 1525 | unsigned long softirq_disable_ip; |
de30a2b3 | 1526 | unsigned long softirq_enable_ip; |
fa1452e8 | 1527 | unsigned int softirq_disable_event; |
de30a2b3 | 1528 | unsigned int softirq_enable_event; |
fa1452e8 | 1529 | int softirqs_enabled; |
de30a2b3 IM |
1530 | int softirq_context; |
1531 | #endif | |
fbb9ce95 | 1532 | #ifdef CONFIG_LOCKDEP |
bdb9441e | 1533 | # define MAX_LOCK_DEPTH 48UL |
fbb9ce95 IM |
1534 | u64 curr_chain_key; |
1535 | int lockdep_depth; | |
fbb9ce95 | 1536 | unsigned int lockdep_recursion; |
c7aceaba | 1537 | struct held_lock held_locks[MAX_LOCK_DEPTH]; |
cf40bd16 | 1538 | gfp_t lockdep_reclaim_gfp; |
fbb9ce95 | 1539 | #endif |
c6d30853 AR |
1540 | #ifdef CONFIG_UBSAN |
1541 | unsigned int in_ubsan; | |
1542 | #endif | |
408894ee | 1543 | |
1da177e4 LT |
1544 | /* journalling filesystem info */ |
1545 | void *journal_info; | |
1546 | ||
d89d8796 | 1547 | /* stacked block device info */ |
bddd87c7 | 1548 | struct bio_list *bio_list; |
d89d8796 | 1549 | |
73c10101 JA |
1550 | #ifdef CONFIG_BLOCK |
1551 | /* stack plugging */ | |
1552 | struct blk_plug *plug; | |
1553 | #endif | |
1554 | ||
1da177e4 LT |
1555 | /* VM state */ |
1556 | struct reclaim_state *reclaim_state; | |
1557 | ||
1da177e4 LT |
1558 | struct backing_dev_info *backing_dev_info; |
1559 | ||
1560 | struct io_context *io_context; | |
1561 | ||
1562 | unsigned long ptrace_message; | |
1563 | siginfo_t *last_siginfo; /* For ptrace use. */ | |
7c3ab738 | 1564 | struct task_io_accounting ioac; |
8f0ab514 | 1565 | #if defined(CONFIG_TASK_XACCT) |
1da177e4 LT |
1566 | u64 acct_rss_mem1; /* accumulated rss usage */ |
1567 | u64 acct_vm_mem1; /* accumulated virtual memory usage */ | |
605dc2b3 | 1568 | u64 acct_timexpd; /* stime + utime since last update */ |
1da177e4 LT |
1569 | #endif |
1570 | #ifdef CONFIG_CPUSETS | |
58568d2a | 1571 | nodemask_t mems_allowed; /* Protected by alloc_lock */ |
cc9a6c87 | 1572 | seqcount_t mems_allowed_seq; /* Seqence no to catch updates */ |
825a46af | 1573 | int cpuset_mem_spread_rotor; |
6adef3eb | 1574 | int cpuset_slab_spread_rotor; |
1da177e4 | 1575 | #endif |
ddbcc7e8 | 1576 | #ifdef CONFIG_CGROUPS |
817929ec | 1577 | /* Control Group info protected by css_set_lock */ |
2c392b8c | 1578 | struct css_set __rcu *cgroups; |
817929ec PM |
1579 | /* cg_list protected by css_set_lock and tsk->alloc_lock */ |
1580 | struct list_head cg_list; | |
ddbcc7e8 | 1581 | #endif |
e02737d5 FY |
1582 | #ifdef CONFIG_INTEL_RDT_A |
1583 | int closid; | |
1584 | #endif | |
42b2dd0a | 1585 | #ifdef CONFIG_FUTEX |
0771dfef | 1586 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
1587 | #ifdef CONFIG_COMPAT |
1588 | struct compat_robust_list_head __user *compat_robust_list; | |
1589 | #endif | |
c87e2837 IM |
1590 | struct list_head pi_state_list; |
1591 | struct futex_pi_state *pi_state_cache; | |
c7aceaba | 1592 | #endif |
cdd6c482 | 1593 | #ifdef CONFIG_PERF_EVENTS |
8dc85d54 | 1594 | struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; |
cdd6c482 IM |
1595 | struct mutex perf_event_mutex; |
1596 | struct list_head perf_event_list; | |
a63eaf34 | 1597 | #endif |
8f47b187 TG |
1598 | #ifdef CONFIG_DEBUG_PREEMPT |
1599 | unsigned long preempt_disable_ip; | |
1600 | #endif | |
c7aceaba | 1601 | #ifdef CONFIG_NUMA |
58568d2a | 1602 | struct mempolicy *mempolicy; /* Protected by alloc_lock */ |
c7aceaba | 1603 | short il_next; |
207205a2 | 1604 | short pref_node_fork; |
42b2dd0a | 1605 | #endif |
cbee9f88 PZ |
1606 | #ifdef CONFIG_NUMA_BALANCING |
1607 | int numa_scan_seq; | |
cbee9f88 | 1608 | unsigned int numa_scan_period; |
598f0ec0 | 1609 | unsigned int numa_scan_period_max; |
de1c9ce6 | 1610 | int numa_preferred_nid; |
6b9a7460 | 1611 | unsigned long numa_migrate_retry; |
cbee9f88 | 1612 | u64 node_stamp; /* migration stamp */ |
7e2703e6 RR |
1613 | u64 last_task_numa_placement; |
1614 | u64 last_sum_exec_runtime; | |
cbee9f88 | 1615 | struct callback_head numa_work; |
f809ca9a | 1616 | |
8c8a743c PZ |
1617 | struct list_head numa_entry; |
1618 | struct numa_group *numa_group; | |
1619 | ||
745d6147 | 1620 | /* |
44dba3d5 IM |
1621 | * numa_faults is an array split into four regions: |
1622 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
1623 | * in this precise order. | |
1624 | * | |
1625 | * faults_memory: Exponential decaying average of faults on a per-node | |
1626 | * basis. Scheduling placement decisions are made based on these | |
1627 | * counts. The values remain static for the duration of a PTE scan. | |
1628 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
1629 | * hinting fault was incurred. | |
1630 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
1631 | * during the current scan window. When the scan completes, the counts | |
1632 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 1633 | */ |
44dba3d5 | 1634 | unsigned long *numa_faults; |
83e1d2cd | 1635 | unsigned long total_numa_faults; |
745d6147 | 1636 | |
04bb2f94 RR |
1637 | /* |
1638 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
1639 | * scan window were remote/local or failed to migrate. The task scan |
1640 | * period is adapted based on the locality of the faults with different | |
1641 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 1642 | */ |
074c2381 | 1643 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 1644 | |
b32e86b4 | 1645 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
1646 | #endif /* CONFIG_NUMA_BALANCING */ |
1647 | ||
72b252ae MG |
1648 | #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH |
1649 | struct tlbflush_unmap_batch tlb_ubc; | |
1650 | #endif | |
1651 | ||
e56d0903 | 1652 | struct rcu_head rcu; |
b92ce558 JA |
1653 | |
1654 | /* | |
1655 | * cache last used pipe for splice | |
1656 | */ | |
1657 | struct pipe_inode_info *splice_pipe; | |
5640f768 ED |
1658 | |
1659 | struct page_frag task_frag; | |
1660 | ||
ca74e92b SN |
1661 | #ifdef CONFIG_TASK_DELAY_ACCT |
1662 | struct task_delay_info *delays; | |
f4f154fd AM |
1663 | #endif |
1664 | #ifdef CONFIG_FAULT_INJECTION | |
1665 | int make_it_fail; | |
ca74e92b | 1666 | #endif |
9d823e8f WF |
1667 | /* |
1668 | * when (nr_dirtied >= nr_dirtied_pause), it's time to call | |
1669 | * balance_dirty_pages() for some dirty throttling pause | |
1670 | */ | |
1671 | int nr_dirtied; | |
1672 | int nr_dirtied_pause; | |
83712358 | 1673 | unsigned long dirty_paused_when; /* start of a write-and-pause period */ |
9d823e8f | 1674 | |
9745512c AV |
1675 | #ifdef CONFIG_LATENCYTOP |
1676 | int latency_record_count; | |
1677 | struct latency_record latency_record[LT_SAVECOUNT]; | |
1678 | #endif | |
6976675d AV |
1679 | /* |
1680 | * time slack values; these are used to round up poll() and | |
1681 | * select() etc timeout values. These are in nanoseconds. | |
1682 | */ | |
da8b44d5 JS |
1683 | u64 timer_slack_ns; |
1684 | u64 default_timer_slack_ns; | |
f8d570a4 | 1685 | |
0b24becc AR |
1686 | #ifdef CONFIG_KASAN |
1687 | unsigned int kasan_depth; | |
1688 | #endif | |
fb52607a | 1689 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
3ad2f3fb | 1690 | /* Index of current stored address in ret_stack */ |
f201ae23 FW |
1691 | int curr_ret_stack; |
1692 | /* Stack of return addresses for return function tracing */ | |
1693 | struct ftrace_ret_stack *ret_stack; | |
8aef2d28 SR |
1694 | /* time stamp for last schedule */ |
1695 | unsigned long long ftrace_timestamp; | |
f201ae23 FW |
1696 | /* |
1697 | * Number of functions that haven't been traced | |
1698 | * because of depth overrun. | |
1699 | */ | |
1700 | atomic_t trace_overrun; | |
380c4b14 FW |
1701 | /* Pause for the tracing */ |
1702 | atomic_t tracing_graph_pause; | |
f201ae23 | 1703 | #endif |
ea4e2bc4 SR |
1704 | #ifdef CONFIG_TRACING |
1705 | /* state flags for use by tracers */ | |
1706 | unsigned long trace; | |
b1cff0ad | 1707 | /* bitmask and counter of trace recursion */ |
261842b7 SR |
1708 | unsigned long trace_recursion; |
1709 | #endif /* CONFIG_TRACING */ | |
5c9a8750 DV |
1710 | #ifdef CONFIG_KCOV |
1711 | /* Coverage collection mode enabled for this task (0 if disabled). */ | |
1712 | enum kcov_mode kcov_mode; | |
1713 | /* Size of the kcov_area. */ | |
1714 | unsigned kcov_size; | |
1715 | /* Buffer for coverage collection. */ | |
1716 | void *kcov_area; | |
1717 | /* kcov desciptor wired with this task or NULL. */ | |
1718 | struct kcov *kcov; | |
1719 | #endif | |
6f185c29 | 1720 | #ifdef CONFIG_MEMCG |
626ebc41 TH |
1721 | struct mem_cgroup *memcg_in_oom; |
1722 | gfp_t memcg_oom_gfp_mask; | |
1723 | int memcg_oom_order; | |
b23afb93 TH |
1724 | |
1725 | /* number of pages to reclaim on returning to userland */ | |
1726 | unsigned int memcg_nr_pages_over_high; | |
569b846d | 1727 | #endif |
0326f5a9 SD |
1728 | #ifdef CONFIG_UPROBES |
1729 | struct uprobe_task *utask; | |
0326f5a9 | 1730 | #endif |
cafe5635 KO |
1731 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
1732 | unsigned int sequential_io; | |
1733 | unsigned int sequential_io_avg; | |
1734 | #endif | |
8eb23b9f PZ |
1735 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
1736 | unsigned long task_state_change; | |
1737 | #endif | |
8bcbde54 | 1738 | int pagefault_disabled; |
03049269 | 1739 | #ifdef CONFIG_MMU |
29c696e1 | 1740 | struct task_struct *oom_reaper_list; |
03049269 | 1741 | #endif |
ba14a194 AL |
1742 | #ifdef CONFIG_VMAP_STACK |
1743 | struct vm_struct *stack_vm_area; | |
1744 | #endif | |
68f24b08 AL |
1745 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
1746 | /* A live task holds one reference. */ | |
1747 | atomic_t stack_refcount; | |
1748 | #endif | |
0c8c0f03 DH |
1749 | /* CPU-specific state of this task */ |
1750 | struct thread_struct thread; | |
1751 | /* | |
1752 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1753 | * structure. It *MUST* be at the end of 'task_struct'. | |
1754 | * | |
1755 | * Do not put anything below here! | |
1756 | */ | |
1da177e4 LT |
1757 | }; |
1758 | ||
5aaeb5c0 IM |
1759 | #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT |
1760 | extern int arch_task_struct_size __read_mostly; | |
1761 | #else | |
1762 | # define arch_task_struct_size (sizeof(struct task_struct)) | |
1763 | #endif | |
0c8c0f03 | 1764 | |
ba14a194 AL |
1765 | #ifdef CONFIG_VMAP_STACK |
1766 | static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t) | |
1767 | { | |
1768 | return t->stack_vm_area; | |
1769 | } | |
1770 | #else | |
1771 | static inline struct vm_struct *task_stack_vm_area(const struct task_struct *t) | |
1772 | { | |
1773 | return NULL; | |
1774 | } | |
1775 | #endif | |
1776 | ||
6688cc05 PZ |
1777 | #define TNF_MIGRATED 0x01 |
1778 | #define TNF_NO_GROUP 0x02 | |
dabe1d99 | 1779 | #define TNF_SHARED 0x04 |
04bb2f94 | 1780 | #define TNF_FAULT_LOCAL 0x08 |
074c2381 | 1781 | #define TNF_MIGRATE_FAIL 0x10 |
6688cc05 | 1782 | |
b18dc5f2 MH |
1783 | static inline bool in_vfork(struct task_struct *tsk) |
1784 | { | |
1785 | bool ret; | |
1786 | ||
1787 | /* | |
1788 | * need RCU to access ->real_parent if CLONE_VM was used along with | |
1789 | * CLONE_PARENT. | |
1790 | * | |
1791 | * We check real_parent->mm == tsk->mm because CLONE_VFORK does not | |
1792 | * imply CLONE_VM | |
1793 | * | |
1794 | * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus | |
1795 | * ->real_parent is not necessarily the task doing vfork(), so in | |
1796 | * theory we can't rely on task_lock() if we want to dereference it. | |
1797 | * | |
1798 | * And in this case we can't trust the real_parent->mm == tsk->mm | |
1799 | * check, it can be false negative. But we do not care, if init or | |
1800 | * another oom-unkillable task does this it should blame itself. | |
1801 | */ | |
1802 | rcu_read_lock(); | |
1803 | ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm; | |
1804 | rcu_read_unlock(); | |
1805 | ||
1806 | return ret; | |
1807 | } | |
1808 | ||
cbee9f88 | 1809 | #ifdef CONFIG_NUMA_BALANCING |
6688cc05 | 1810 | extern void task_numa_fault(int last_node, int node, int pages, int flags); |
e29cf08b | 1811 | extern pid_t task_numa_group_id(struct task_struct *p); |
1a687c2e | 1812 | extern void set_numabalancing_state(bool enabled); |
82727018 | 1813 | extern void task_numa_free(struct task_struct *p); |
10f39042 RR |
1814 | extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page, |
1815 | int src_nid, int dst_cpu); | |
cbee9f88 | 1816 | #else |
ac8e895b | 1817 | static inline void task_numa_fault(int last_node, int node, int pages, |
6688cc05 | 1818 | int flags) |
cbee9f88 PZ |
1819 | { |
1820 | } | |
e29cf08b MG |
1821 | static inline pid_t task_numa_group_id(struct task_struct *p) |
1822 | { | |
1823 | return 0; | |
1824 | } | |
1a687c2e MG |
1825 | static inline void set_numabalancing_state(bool enabled) |
1826 | { | |
1827 | } | |
82727018 RR |
1828 | static inline void task_numa_free(struct task_struct *p) |
1829 | { | |
1830 | } | |
10f39042 RR |
1831 | static inline bool should_numa_migrate_memory(struct task_struct *p, |
1832 | struct page *page, int src_nid, int dst_cpu) | |
1833 | { | |
1834 | return true; | |
1835 | } | |
cbee9f88 PZ |
1836 | #endif |
1837 | ||
e868171a | 1838 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 EB |
1839 | { |
1840 | return task->pids[PIDTYPE_PID].pid; | |
1841 | } | |
1842 | ||
e868171a | 1843 | static inline struct pid *task_tgid(struct task_struct *task) |
22c935f4 EB |
1844 | { |
1845 | return task->group_leader->pids[PIDTYPE_PID].pid; | |
1846 | } | |
1847 | ||
6dda81f4 ON |
1848 | /* |
1849 | * Without tasklist or rcu lock it is not safe to dereference | |
1850 | * the result of task_pgrp/task_session even if task == current, | |
1851 | * we can race with another thread doing sys_setsid/sys_setpgid. | |
1852 | */ | |
e868171a | 1853 | static inline struct pid *task_pgrp(struct task_struct *task) |
22c935f4 EB |
1854 | { |
1855 | return task->group_leader->pids[PIDTYPE_PGID].pid; | |
1856 | } | |
1857 | ||
e868171a | 1858 | static inline struct pid *task_session(struct task_struct *task) |
22c935f4 EB |
1859 | { |
1860 | return task->group_leader->pids[PIDTYPE_SID].pid; | |
1861 | } | |
1862 | ||
7af57294 PE |
1863 | struct pid_namespace; |
1864 | ||
1865 | /* | |
1866 | * the helpers to get the task's different pids as they are seen | |
1867 | * from various namespaces | |
1868 | * | |
1869 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
1870 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
1871 | * current. | |
7af57294 PE |
1872 | * task_xid_nr_ns() : id seen from the ns specified; |
1873 | * | |
1874 | * set_task_vxid() : assigns a virtual id to a task; | |
1875 | * | |
7af57294 PE |
1876 | * see also pid_nr() etc in include/linux/pid.h |
1877 | */ | |
52ee2dfd ON |
1878 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, |
1879 | struct pid_namespace *ns); | |
7af57294 | 1880 | |
e868171a | 1881 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
1882 | { |
1883 | return tsk->pid; | |
1884 | } | |
1885 | ||
52ee2dfd ON |
1886 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, |
1887 | struct pid_namespace *ns) | |
1888 | { | |
1889 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
1890 | } | |
7af57294 PE |
1891 | |
1892 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
1893 | { | |
52ee2dfd | 1894 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
1895 | } |
1896 | ||
1897 | ||
e868171a | 1898 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
1899 | { |
1900 | return tsk->tgid; | |
1901 | } | |
1902 | ||
2f2a3a46 | 1903 | pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); |
7af57294 PE |
1904 | |
1905 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
1906 | { | |
1907 | return pid_vnr(task_tgid(tsk)); | |
1908 | } | |
1909 | ||
1910 | ||
80e0b6e8 | 1911 | static inline int pid_alive(const struct task_struct *p); |
ad36d282 RGB |
1912 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) |
1913 | { | |
1914 | pid_t pid = 0; | |
1915 | ||
1916 | rcu_read_lock(); | |
1917 | if (pid_alive(tsk)) | |
1918 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
1919 | rcu_read_unlock(); | |
1920 | ||
1921 | return pid; | |
1922 | } | |
1923 | ||
1924 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
1925 | { | |
1926 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
1927 | } | |
1928 | ||
52ee2dfd ON |
1929 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, |
1930 | struct pid_namespace *ns) | |
7af57294 | 1931 | { |
52ee2dfd | 1932 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
1933 | } |
1934 | ||
7af57294 PE |
1935 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
1936 | { | |
52ee2dfd | 1937 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
1938 | } |
1939 | ||
1940 | ||
52ee2dfd ON |
1941 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, |
1942 | struct pid_namespace *ns) | |
7af57294 | 1943 | { |
52ee2dfd | 1944 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
1945 | } |
1946 | ||
7af57294 PE |
1947 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
1948 | { | |
52ee2dfd | 1949 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
1950 | } |
1951 | ||
1b0f7ffd ON |
1952 | /* obsolete, do not use */ |
1953 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) | |
1954 | { | |
1955 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
1956 | } | |
7af57294 | 1957 | |
1da177e4 LT |
1958 | /** |
1959 | * pid_alive - check that a task structure is not stale | |
1960 | * @p: Task structure to be checked. | |
1961 | * | |
1962 | * Test if a process is not yet dead (at most zombie state) | |
1963 | * If pid_alive fails, then pointers within the task structure | |
1964 | * can be stale and must not be dereferenced. | |
e69f6186 YB |
1965 | * |
1966 | * Return: 1 if the process is alive. 0 otherwise. | |
1da177e4 | 1967 | */ |
ad36d282 | 1968 | static inline int pid_alive(const struct task_struct *p) |
1da177e4 | 1969 | { |
92476d7f | 1970 | return p->pids[PIDTYPE_PID].pid != NULL; |
1da177e4 LT |
1971 | } |
1972 | ||
f400e198 | 1973 | /** |
570f5241 SS |
1974 | * is_global_init - check if a task structure is init. Since init |
1975 | * is free to have sub-threads we need to check tgid. | |
3260259f HK |
1976 | * @tsk: Task structure to be checked. |
1977 | * | |
1978 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
1979 | * |
1980 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 1981 | */ |
e868171a | 1982 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 1983 | { |
570f5241 | 1984 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 1985 | } |
b460cbc5 | 1986 | |
9ec52099 CLG |
1987 | extern struct pid *cad_pid; |
1988 | ||
1da177e4 | 1989 | extern void free_task(struct task_struct *tsk); |
1da177e4 | 1990 | #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) |
e56d0903 | 1991 | |
158d9ebd | 1992 | extern void __put_task_struct(struct task_struct *t); |
e56d0903 IM |
1993 | |
1994 | static inline void put_task_struct(struct task_struct *t) | |
1995 | { | |
1996 | if (atomic_dec_and_test(&t->usage)) | |
8c7904a0 | 1997 | __put_task_struct(t); |
e56d0903 | 1998 | } |
1da177e4 | 1999 | |
150593bf ON |
2000 | struct task_struct *task_rcu_dereference(struct task_struct **ptask); |
2001 | struct task_struct *try_get_task_struct(struct task_struct **ptask); | |
2002 | ||
6a61671b FW |
2003 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
2004 | extern void task_cputime(struct task_struct *t, | |
5613fda9 | 2005 | u64 *utime, u64 *stime); |
16a6d9be | 2006 | extern u64 task_gtime(struct task_struct *t); |
6a61671b | 2007 | #else |
6fac4829 | 2008 | static inline void task_cputime(struct task_struct *t, |
5613fda9 | 2009 | u64 *utime, u64 *stime) |
6fac4829 | 2010 | { |
353c50eb SG |
2011 | *utime = t->utime; |
2012 | *stime = t->stime; | |
6fac4829 FW |
2013 | } |
2014 | ||
16a6d9be | 2015 | static inline u64 task_gtime(struct task_struct *t) |
40565b5a SG |
2016 | { |
2017 | return t->gtime; | |
2018 | } | |
2019 | #endif | |
2020 | ||
2021 | #ifdef CONFIG_ARCH_HAS_SCALED_CPUTIME | |
6fac4829 | 2022 | static inline void task_cputime_scaled(struct task_struct *t, |
5613fda9 FW |
2023 | u64 *utimescaled, |
2024 | u64 *stimescaled) | |
6fac4829 | 2025 | { |
353c50eb SG |
2026 | *utimescaled = t->utimescaled; |
2027 | *stimescaled = t->stimescaled; | |
6fac4829 | 2028 | } |
40565b5a SG |
2029 | #else |
2030 | static inline void task_cputime_scaled(struct task_struct *t, | |
5613fda9 FW |
2031 | u64 *utimescaled, |
2032 | u64 *stimescaled) | |
6a61671b | 2033 | { |
40565b5a | 2034 | task_cputime(t, utimescaled, stimescaled); |
6a61671b FW |
2035 | } |
2036 | #endif | |
40565b5a | 2037 | |
5613fda9 FW |
2038 | extern void task_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st); |
2039 | extern void thread_group_cputime_adjusted(struct task_struct *p, u64 *ut, u64 *st); | |
49048622 | 2040 | |
1da177e4 LT |
2041 | /* |
2042 | * Per process flags | |
2043 | */ | |
c1de45ca | 2044 | #define PF_IDLE 0x00000002 /* I am an IDLE thread */ |
1da177e4 | 2045 | #define PF_EXITING 0x00000004 /* getting shut down */ |
778e9a9c | 2046 | #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ |
94886b84 | 2047 | #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ |
21aa9af0 | 2048 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ |
1da177e4 | 2049 | #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ |
4db96cf0 | 2050 | #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ |
1da177e4 LT |
2051 | #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ |
2052 | #define PF_DUMPCORE 0x00000200 /* dumped core */ | |
2053 | #define PF_SIGNALED 0x00000400 /* killed by a signal */ | |
2054 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
72fa5997 | 2055 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */ |
1da177e4 | 2056 | #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ |
774a1221 | 2057 | #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */ |
1da177e4 LT |
2058 | #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ |
2059 | #define PF_FROZEN 0x00010000 /* frozen for system suspend */ | |
2060 | #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ | |
2061 | #define PF_KSWAPD 0x00040000 /* I am kswapd */ | |
21caf2fc | 2062 | #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */ |
1da177e4 | 2063 | #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ |
246bb0b1 | 2064 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ |
b31dc66a JA |
2065 | #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ |
2066 | #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ | |
14a40ffc | 2067 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */ |
4db96cf0 | 2068 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ |
61a87122 | 2069 | #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ |
58a69cb4 | 2070 | #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ |
2b44c4db | 2071 | #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */ |
1da177e4 LT |
2072 | |
2073 | /* | |
2074 | * Only the _current_ task can read/write to tsk->flags, but other | |
2075 | * tasks can access tsk->flags in readonly mode for example | |
2076 | * with tsk_used_math (like during threaded core dumping). | |
2077 | * There is however an exception to this rule during ptrace | |
2078 | * or during fork: the ptracer task is allowed to write to the | |
2079 | * child->flags of its traced child (same goes for fork, the parent | |
2080 | * can write to the child->flags), because we're guaranteed the | |
2081 | * child is not running and in turn not changing child->flags | |
2082 | * at the same time the parent does it. | |
2083 | */ | |
2084 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) | |
2085 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
2086 | #define clear_used_math() clear_stopped_child_used_math(current) | |
2087 | #define set_used_math() set_stopped_child_used_math(current) | |
2088 | #define conditional_stopped_child_used_math(condition, child) \ | |
2089 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
2090 | #define conditional_used_math(condition) \ | |
2091 | conditional_stopped_child_used_math(condition, current) | |
2092 | #define copy_to_stopped_child_used_math(child) \ | |
2093 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
2094 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ | |
2095 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) | |
2096 | #define used_math() tsk_used_math(current) | |
2097 | ||
934f3072 JB |
2098 | /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags |
2099 | * __GFP_FS is also cleared as it implies __GFP_IO. | |
2100 | */ | |
21caf2fc ML |
2101 | static inline gfp_t memalloc_noio_flags(gfp_t flags) |
2102 | { | |
2103 | if (unlikely(current->flags & PF_MEMALLOC_NOIO)) | |
934f3072 | 2104 | flags &= ~(__GFP_IO | __GFP_FS); |
21caf2fc ML |
2105 | return flags; |
2106 | } | |
2107 | ||
2108 | static inline unsigned int memalloc_noio_save(void) | |
2109 | { | |
2110 | unsigned int flags = current->flags & PF_MEMALLOC_NOIO; | |
2111 | current->flags |= PF_MEMALLOC_NOIO; | |
2112 | return flags; | |
2113 | } | |
2114 | ||
2115 | static inline void memalloc_noio_restore(unsigned int flags) | |
2116 | { | |
2117 | current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags; | |
2118 | } | |
2119 | ||
1d4457f9 | 2120 | /* Per-process atomic flags. */ |
a2b86f77 | 2121 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
2ad654bc ZL |
2122 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ |
2123 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
77ed2c57 | 2124 | #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */ |
2ad654bc | 2125 | |
1d4457f9 | 2126 | |
e0e5070b ZL |
2127 | #define TASK_PFA_TEST(name, func) \ |
2128 | static inline bool task_##func(struct task_struct *p) \ | |
2129 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
2130 | #define TASK_PFA_SET(name, func) \ | |
2131 | static inline void task_set_##func(struct task_struct *p) \ | |
2132 | { set_bit(PFA_##name, &p->atomic_flags); } | |
2133 | #define TASK_PFA_CLEAR(name, func) \ | |
2134 | static inline void task_clear_##func(struct task_struct *p) \ | |
2135 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
2136 | ||
2137 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
2138 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 2139 | |
2ad654bc ZL |
2140 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
2141 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
2142 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
2143 | ||
2144 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
2145 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
2146 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 2147 | |
77ed2c57 TH |
2148 | TASK_PFA_TEST(LMK_WAITING, lmk_waiting) |
2149 | TASK_PFA_SET(LMK_WAITING, lmk_waiting) | |
2150 | ||
e5c1902e | 2151 | /* |
a8f072c1 | 2152 | * task->jobctl flags |
e5c1902e | 2153 | */ |
a8f072c1 | 2154 | #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */ |
e5c1902e | 2155 | |
a8f072c1 TH |
2156 | #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */ |
2157 | #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */ | |
2158 | #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */ | |
73ddff2b | 2159 | #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */ |
fb1d910c | 2160 | #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */ |
a8f072c1 | 2161 | #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */ |
544b2c91 | 2162 | #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */ |
a8f072c1 | 2163 | |
b76808e6 PD |
2164 | #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT) |
2165 | #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT) | |
2166 | #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT) | |
2167 | #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT) | |
2168 | #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT) | |
2169 | #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT) | |
2170 | #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT) | |
a8f072c1 | 2171 | |
fb1d910c | 2172 | #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY) |
73ddff2b | 2173 | #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK) |
3759a0d9 | 2174 | |
7dd3db54 | 2175 | extern bool task_set_jobctl_pending(struct task_struct *task, |
b76808e6 | 2176 | unsigned long mask); |
73ddff2b | 2177 | extern void task_clear_jobctl_trapping(struct task_struct *task); |
3759a0d9 | 2178 | extern void task_clear_jobctl_pending(struct task_struct *task, |
b76808e6 | 2179 | unsigned long mask); |
39efa3ef | 2180 | |
f41d911f PM |
2181 | static inline void rcu_copy_process(struct task_struct *p) |
2182 | { | |
8315f422 | 2183 | #ifdef CONFIG_PREEMPT_RCU |
f41d911f | 2184 | p->rcu_read_lock_nesting = 0; |
1d082fd0 | 2185 | p->rcu_read_unlock_special.s = 0; |
dd5d19ba | 2186 | p->rcu_blocked_node = NULL; |
f41d911f | 2187 | INIT_LIST_HEAD(&p->rcu_node_entry); |
8315f422 PM |
2188 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
2189 | #ifdef CONFIG_TASKS_RCU | |
2190 | p->rcu_tasks_holdout = false; | |
2191 | INIT_LIST_HEAD(&p->rcu_tasks_holdout_list); | |
176f8f7a | 2192 | p->rcu_tasks_idle_cpu = -1; |
8315f422 | 2193 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
f41d911f PM |
2194 | } |
2195 | ||
907aed48 MG |
2196 | static inline void tsk_restore_flags(struct task_struct *task, |
2197 | unsigned long orig_flags, unsigned long flags) | |
2198 | { | |
2199 | task->flags &= ~flags; | |
2200 | task->flags |= orig_flags & flags; | |
2201 | } | |
2202 | ||
f82f8042 JL |
2203 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, |
2204 | const struct cpumask *trial); | |
7f51412a JL |
2205 | extern int task_can_attach(struct task_struct *p, |
2206 | const struct cpumask *cs_cpus_allowed); | |
1da177e4 | 2207 | #ifdef CONFIG_SMP |
1e1b6c51 KM |
2208 | extern void do_set_cpus_allowed(struct task_struct *p, |
2209 | const struct cpumask *new_mask); | |
2210 | ||
cd8ba7cd | 2211 | extern int set_cpus_allowed_ptr(struct task_struct *p, |
96f874e2 | 2212 | const struct cpumask *new_mask); |
1da177e4 | 2213 | #else |
1e1b6c51 KM |
2214 | static inline void do_set_cpus_allowed(struct task_struct *p, |
2215 | const struct cpumask *new_mask) | |
2216 | { | |
2217 | } | |
cd8ba7cd | 2218 | static inline int set_cpus_allowed_ptr(struct task_struct *p, |
96f874e2 | 2219 | const struct cpumask *new_mask) |
1da177e4 | 2220 | { |
96f874e2 | 2221 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
2222 | return -EINVAL; |
2223 | return 0; | |
2224 | } | |
2225 | #endif | |
e0ad9556 | 2226 | |
3451d024 | 2227 | #ifdef CONFIG_NO_HZ_COMMON |
5167e8d5 PZ |
2228 | void calc_load_enter_idle(void); |
2229 | void calc_load_exit_idle(void); | |
2230 | #else | |
2231 | static inline void calc_load_enter_idle(void) { } | |
2232 | static inline void calc_load_exit_idle(void) { } | |
3451d024 | 2233 | #endif /* CONFIG_NO_HZ_COMMON */ |
5167e8d5 | 2234 | |
6d0d2878 CB |
2235 | #ifndef cpu_relax_yield |
2236 | #define cpu_relax_yield() cpu_relax() | |
2237 | #endif | |
2238 | ||
b342501c | 2239 | /* |
c676329a PZ |
2240 | * Do not use outside of architecture code which knows its limitations. |
2241 | * | |
2242 | * sched_clock() has no promise of monotonicity or bounded drift between | |
2243 | * CPUs, use (which you should not) requires disabling IRQs. | |
2244 | * | |
2245 | * Please use one of the three interfaces below. | |
b342501c | 2246 | */ |
1bbfa6f2 | 2247 | extern unsigned long long notrace sched_clock(void); |
c676329a | 2248 | /* |
489a71b0 | 2249 | * See the comment in kernel/sched/clock.c |
c676329a | 2250 | */ |
545a2bf7 | 2251 | extern u64 running_clock(void); |
c676329a PZ |
2252 | extern u64 sched_clock_cpu(int cpu); |
2253 | ||
e436d800 | 2254 | |
c1955a3d | 2255 | extern void sched_clock_init(void); |
3e51f33f | 2256 | |
c1955a3d | 2257 | #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK |
9881b024 PZ |
2258 | static inline void sched_clock_init_late(void) |
2259 | { | |
2260 | } | |
2261 | ||
3e51f33f PZ |
2262 | static inline void sched_clock_tick(void) |
2263 | { | |
2264 | } | |
2265 | ||
733ce725 PG |
2266 | static inline void clear_sched_clock_stable(void) |
2267 | { | |
2268 | } | |
2269 | ||
3e51f33f PZ |
2270 | static inline void sched_clock_idle_sleep_event(void) |
2271 | { | |
2272 | } | |
2273 | ||
2274 | static inline void sched_clock_idle_wakeup_event(u64 delta_ns) | |
2275 | { | |
2276 | } | |
2c923e94 DL |
2277 | |
2278 | static inline u64 cpu_clock(int cpu) | |
2279 | { | |
2280 | return sched_clock(); | |
2281 | } | |
2282 | ||
2283 | static inline u64 local_clock(void) | |
2284 | { | |
2285 | return sched_clock(); | |
2286 | } | |
3e51f33f | 2287 | #else |
9881b024 | 2288 | extern void sched_clock_init_late(void); |
c676329a PZ |
2289 | /* |
2290 | * Architectures can set this to 1 if they have specified | |
2291 | * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, | |
2292 | * but then during bootup it turns out that sched_clock() | |
2293 | * is reliable after all: | |
2294 | */ | |
35af99e6 | 2295 | extern int sched_clock_stable(void); |
35af99e6 | 2296 | extern void clear_sched_clock_stable(void); |
c676329a | 2297 | |
3e51f33f PZ |
2298 | extern void sched_clock_tick(void); |
2299 | extern void sched_clock_idle_sleep_event(void); | |
2300 | extern void sched_clock_idle_wakeup_event(u64 delta_ns); | |
2c923e94 DL |
2301 | |
2302 | /* | |
2303 | * As outlined in clock.c, provides a fast, high resolution, nanosecond | |
2304 | * time source that is monotonic per cpu argument and has bounded drift | |
2305 | * between cpus. | |
2306 | * | |
2307 | * ######################### BIG FAT WARNING ########################## | |
2308 | * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can # | |
2309 | * # go backwards !! # | |
2310 | * #################################################################### | |
2311 | */ | |
2312 | static inline u64 cpu_clock(int cpu) | |
2313 | { | |
2314 | return sched_clock_cpu(cpu); | |
2315 | } | |
2316 | ||
2317 | static inline u64 local_clock(void) | |
2318 | { | |
2319 | return sched_clock_cpu(raw_smp_processor_id()); | |
2320 | } | |
3e51f33f PZ |
2321 | #endif |
2322 | ||
b52bfee4 VP |
2323 | #ifdef CONFIG_IRQ_TIME_ACCOUNTING |
2324 | /* | |
2325 | * An i/f to runtime opt-in for irq time accounting based off of sched_clock. | |
2326 | * The reason for this explicit opt-in is not to have perf penalty with | |
2327 | * slow sched_clocks. | |
2328 | */ | |
2329 | extern void enable_sched_clock_irqtime(void); | |
2330 | extern void disable_sched_clock_irqtime(void); | |
2331 | #else | |
2332 | static inline void enable_sched_clock_irqtime(void) {} | |
2333 | static inline void disable_sched_clock_irqtime(void) {} | |
2334 | #endif | |
2335 | ||
36c8b586 | 2336 | extern unsigned long long |
41b86e9c | 2337 | task_sched_runtime(struct task_struct *task); |
1da177e4 LT |
2338 | |
2339 | /* sched_exec is called by processes performing an exec */ | |
2340 | #ifdef CONFIG_SMP | |
2341 | extern void sched_exec(void); | |
2342 | #else | |
2343 | #define sched_exec() {} | |
2344 | #endif | |
2345 | ||
2aa44d05 IM |
2346 | extern void sched_clock_idle_sleep_event(void); |
2347 | extern void sched_clock_idle_wakeup_event(u64 delta_ns); | |
bb29ab26 | 2348 | |
1da177e4 LT |
2349 | #ifdef CONFIG_HOTPLUG_CPU |
2350 | extern void idle_task_exit(void); | |
2351 | #else | |
2352 | static inline void idle_task_exit(void) {} | |
2353 | #endif | |
2354 | ||
3451d024 | 2355 | #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP) |
1c20091e | 2356 | extern void wake_up_nohz_cpu(int cpu); |
06d8308c | 2357 | #else |
1c20091e | 2358 | static inline void wake_up_nohz_cpu(int cpu) { } |
06d8308c TG |
2359 | #endif |
2360 | ||
ce831b38 | 2361 | #ifdef CONFIG_NO_HZ_FULL |
265f22a9 | 2362 | extern u64 scheduler_tick_max_deferment(void); |
06d8308c TG |
2363 | #endif |
2364 | ||
5091faa4 | 2365 | #ifdef CONFIG_SCHED_AUTOGROUP |
5091faa4 MG |
2366 | extern void sched_autogroup_create_attach(struct task_struct *p); |
2367 | extern void sched_autogroup_detach(struct task_struct *p); | |
2368 | extern void sched_autogroup_fork(struct signal_struct *sig); | |
2369 | extern void sched_autogroup_exit(struct signal_struct *sig); | |
8e5bfa8c | 2370 | extern void sched_autogroup_exit_task(struct task_struct *p); |
5091faa4 MG |
2371 | #ifdef CONFIG_PROC_FS |
2372 | extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m); | |
2e5b5b3a | 2373 | extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice); |
5091faa4 MG |
2374 | #endif |
2375 | #else | |
2376 | static inline void sched_autogroup_create_attach(struct task_struct *p) { } | |
2377 | static inline void sched_autogroup_detach(struct task_struct *p) { } | |
2378 | static inline void sched_autogroup_fork(struct signal_struct *sig) { } | |
2379 | static inline void sched_autogroup_exit(struct signal_struct *sig) { } | |
8e5bfa8c | 2380 | static inline void sched_autogroup_exit_task(struct task_struct *p) { } |
5091faa4 MG |
2381 | #endif |
2382 | ||
fa93384f | 2383 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
2384 | extern void set_user_nice(struct task_struct *p, long nice); |
2385 | extern int task_prio(const struct task_struct *p); | |
d0ea0268 DY |
2386 | /** |
2387 | * task_nice - return the nice value of a given task. | |
2388 | * @p: the task in question. | |
2389 | * | |
2390 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
2391 | */ | |
2392 | static inline int task_nice(const struct task_struct *p) | |
2393 | { | |
2394 | return PRIO_TO_NICE((p)->static_prio); | |
2395 | } | |
36c8b586 IM |
2396 | extern int can_nice(const struct task_struct *p, const int nice); |
2397 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 2398 | extern int idle_cpu(int cpu); |
fe7de49f KM |
2399 | extern int sched_setscheduler(struct task_struct *, int, |
2400 | const struct sched_param *); | |
961ccddd | 2401 | extern int sched_setscheduler_nocheck(struct task_struct *, int, |
fe7de49f | 2402 | const struct sched_param *); |
d50dde5a DF |
2403 | extern int sched_setattr(struct task_struct *, |
2404 | const struct sched_attr *); | |
36c8b586 | 2405 | extern struct task_struct *idle_task(int cpu); |
c4f30608 PM |
2406 | /** |
2407 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 2408 | * @p: the task in question. |
e69f6186 YB |
2409 | * |
2410 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 2411 | */ |
7061ca3b | 2412 | static inline bool is_idle_task(const struct task_struct *p) |
c4f30608 | 2413 | { |
c1de45ca | 2414 | return !!(p->flags & PF_IDLE); |
c4f30608 | 2415 | } |
36c8b586 | 2416 | extern struct task_struct *curr_task(int cpu); |
a458ae2e | 2417 | extern void ia64_set_curr_task(int cpu, struct task_struct *p); |
1da177e4 LT |
2418 | |
2419 | void yield(void); | |
2420 | ||
1da177e4 | 2421 | union thread_union { |
c65eacbe | 2422 | #ifndef CONFIG_THREAD_INFO_IN_TASK |
1da177e4 | 2423 | struct thread_info thread_info; |
c65eacbe | 2424 | #endif |
1da177e4 LT |
2425 | unsigned long stack[THREAD_SIZE/sizeof(long)]; |
2426 | }; | |
2427 | ||
2428 | #ifndef __HAVE_ARCH_KSTACK_END | |
2429 | static inline int kstack_end(void *addr) | |
2430 | { | |
2431 | /* Reliable end of stack detection: | |
2432 | * Some APM bios versions misalign the stack | |
2433 | */ | |
2434 | return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); | |
2435 | } | |
2436 | #endif | |
2437 | ||
2438 | extern union thread_union init_thread_union; | |
2439 | extern struct task_struct init_task; | |
2440 | ||
2441 | extern struct mm_struct init_mm; | |
2442 | ||
198fe21b PE |
2443 | extern struct pid_namespace init_pid_ns; |
2444 | ||
2445 | /* | |
2446 | * find a task by one of its numerical ids | |
2447 | * | |
198fe21b PE |
2448 | * find_task_by_pid_ns(): |
2449 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
2450 | * find_task_by_vpid(): |
2451 | * finds a task by its virtual pid | |
198fe21b | 2452 | * |
e49859e7 | 2453 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
2454 | */ |
2455 | ||
228ebcbe PE |
2456 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
2457 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, | |
2458 | struct pid_namespace *ns); | |
198fe21b | 2459 | |
1da177e4 | 2460 | /* per-UID process charging. */ |
7b44ab97 | 2461 | extern struct user_struct * alloc_uid(kuid_t); |
1da177e4 LT |
2462 | static inline struct user_struct *get_uid(struct user_struct *u) |
2463 | { | |
2464 | atomic_inc(&u->__count); | |
2465 | return u; | |
2466 | } | |
2467 | extern void free_uid(struct user_struct *); | |
1da177e4 LT |
2468 | |
2469 | #include <asm/current.h> | |
2470 | ||
f0af911a | 2471 | extern void xtime_update(unsigned long ticks); |
1da177e4 | 2472 | |
b3c97528 HH |
2473 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
2474 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 2475 | extern void wake_up_new_task(struct task_struct *tsk); |
1da177e4 LT |
2476 | #ifdef CONFIG_SMP |
2477 | extern void kick_process(struct task_struct *tsk); | |
2478 | #else | |
2479 | static inline void kick_process(struct task_struct *tsk) { } | |
2480 | #endif | |
aab03e05 | 2481 | extern int sched_fork(unsigned long clone_flags, struct task_struct *p); |
ad46c2c4 | 2482 | extern void sched_dead(struct task_struct *p); |
1da177e4 | 2483 | |
1da177e4 LT |
2484 | extern void proc_caches_init(void); |
2485 | extern void flush_signals(struct task_struct *); | |
10ab825b | 2486 | extern void ignore_signals(struct task_struct *); |
1da177e4 LT |
2487 | extern void flush_signal_handlers(struct task_struct *, int force_default); |
2488 | extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); | |
2489 | ||
be0e6f29 | 2490 | static inline int kernel_dequeue_signal(siginfo_t *info) |
1da177e4 | 2491 | { |
be0e6f29 ON |
2492 | struct task_struct *tsk = current; |
2493 | siginfo_t __info; | |
1da177e4 LT |
2494 | int ret; |
2495 | ||
be0e6f29 ON |
2496 | spin_lock_irq(&tsk->sighand->siglock); |
2497 | ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info); | |
2498 | spin_unlock_irq(&tsk->sighand->siglock); | |
1da177e4 LT |
2499 | |
2500 | return ret; | |
53c8f9f1 | 2501 | } |
1da177e4 | 2502 | |
9a13049e ON |
2503 | static inline void kernel_signal_stop(void) |
2504 | { | |
2505 | spin_lock_irq(¤t->sighand->siglock); | |
2506 | if (current->jobctl & JOBCTL_STOP_DEQUEUED) | |
2507 | __set_current_state(TASK_STOPPED); | |
2508 | spin_unlock_irq(¤t->sighand->siglock); | |
2509 | ||
2510 | schedule(); | |
2511 | } | |
2512 | ||
1da177e4 LT |
2513 | extern void release_task(struct task_struct * p); |
2514 | extern int send_sig_info(int, struct siginfo *, struct task_struct *); | |
1da177e4 LT |
2515 | extern int force_sigsegv(int, struct task_struct *); |
2516 | extern int force_sig_info(int, struct siginfo *, struct task_struct *); | |
c4b92fc1 | 2517 | extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); |
c4b92fc1 | 2518 | extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); |
d178bc3a SH |
2519 | extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *, |
2520 | const struct cred *, u32); | |
c4b92fc1 EB |
2521 | extern int kill_pgrp(struct pid *pid, int sig, int priv); |
2522 | extern int kill_pid(struct pid *pid, int sig, int priv); | |
c3de4b38 | 2523 | extern int kill_proc_info(int, struct siginfo *, pid_t); |
86773473 | 2524 | extern __must_check bool do_notify_parent(struct task_struct *, int); |
a7f0765e | 2525 | extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent); |
1da177e4 | 2526 | extern void force_sig(int, struct task_struct *); |
1da177e4 | 2527 | extern int send_sig(int, struct task_struct *, int); |
09faef11 | 2528 | extern int zap_other_threads(struct task_struct *p); |
1da177e4 LT |
2529 | extern struct sigqueue *sigqueue_alloc(void); |
2530 | extern void sigqueue_free(struct sigqueue *); | |
ac5c2153 | 2531 | extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); |
9ac95f2f | 2532 | extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); |
1da177e4 | 2533 | |
7e781418 AL |
2534 | #ifdef TIF_RESTORE_SIGMASK |
2535 | /* | |
2536 | * Legacy restore_sigmask accessors. These are inefficient on | |
2537 | * SMP architectures because they require atomic operations. | |
2538 | */ | |
2539 | ||
2540 | /** | |
2541 | * set_restore_sigmask() - make sure saved_sigmask processing gets done | |
2542 | * | |
2543 | * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code | |
2544 | * will run before returning to user mode, to process the flag. For | |
2545 | * all callers, TIF_SIGPENDING is already set or it's no harm to set | |
2546 | * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the | |
2547 | * arch code will notice on return to user mode, in case those bits | |
2548 | * are scarce. We set TIF_SIGPENDING here to ensure that the arch | |
2549 | * signal code always gets run when TIF_RESTORE_SIGMASK is set. | |
2550 | */ | |
2551 | static inline void set_restore_sigmask(void) | |
2552 | { | |
2553 | set_thread_flag(TIF_RESTORE_SIGMASK); | |
2554 | WARN_ON(!test_thread_flag(TIF_SIGPENDING)); | |
2555 | } | |
2556 | static inline void clear_restore_sigmask(void) | |
2557 | { | |
2558 | clear_thread_flag(TIF_RESTORE_SIGMASK); | |
2559 | } | |
2560 | static inline bool test_restore_sigmask(void) | |
2561 | { | |
2562 | return test_thread_flag(TIF_RESTORE_SIGMASK); | |
2563 | } | |
2564 | static inline bool test_and_clear_restore_sigmask(void) | |
2565 | { | |
2566 | return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK); | |
2567 | } | |
2568 | ||
2569 | #else /* TIF_RESTORE_SIGMASK */ | |
2570 | ||
2571 | /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */ | |
2572 | static inline void set_restore_sigmask(void) | |
2573 | { | |
2574 | current->restore_sigmask = true; | |
2575 | WARN_ON(!test_thread_flag(TIF_SIGPENDING)); | |
2576 | } | |
2577 | static inline void clear_restore_sigmask(void) | |
2578 | { | |
2579 | current->restore_sigmask = false; | |
2580 | } | |
2581 | static inline bool test_restore_sigmask(void) | |
2582 | { | |
2583 | return current->restore_sigmask; | |
2584 | } | |
2585 | static inline bool test_and_clear_restore_sigmask(void) | |
2586 | { | |
2587 | if (!current->restore_sigmask) | |
2588 | return false; | |
2589 | current->restore_sigmask = false; | |
2590 | return true; | |
2591 | } | |
2592 | #endif | |
2593 | ||
51a7b448 AV |
2594 | static inline void restore_saved_sigmask(void) |
2595 | { | |
2596 | if (test_and_clear_restore_sigmask()) | |
77097ae5 | 2597 | __set_current_blocked(¤t->saved_sigmask); |
51a7b448 AV |
2598 | } |
2599 | ||
b7f9a11a AV |
2600 | static inline sigset_t *sigmask_to_save(void) |
2601 | { | |
2602 | sigset_t *res = ¤t->blocked; | |
2603 | if (unlikely(test_restore_sigmask())) | |
2604 | res = ¤t->saved_sigmask; | |
2605 | return res; | |
2606 | } | |
2607 | ||
9ec52099 CLG |
2608 | static inline int kill_cad_pid(int sig, int priv) |
2609 | { | |
2610 | return kill_pid(cad_pid, sig, priv); | |
2611 | } | |
2612 | ||
1da177e4 LT |
2613 | /* These can be the second arg to send_sig_info/send_group_sig_info. */ |
2614 | #define SEND_SIG_NOINFO ((struct siginfo *) 0) | |
2615 | #define SEND_SIG_PRIV ((struct siginfo *) 1) | |
2616 | #define SEND_SIG_FORCED ((struct siginfo *) 2) | |
2617 | ||
2a855dd0 SAS |
2618 | /* |
2619 | * True if we are on the alternate signal stack. | |
2620 | */ | |
1da177e4 LT |
2621 | static inline int on_sig_stack(unsigned long sp) |
2622 | { | |
c876eeab AL |
2623 | /* |
2624 | * If the signal stack is SS_AUTODISARM then, by construction, we | |
2625 | * can't be on the signal stack unless user code deliberately set | |
2626 | * SS_AUTODISARM when we were already on it. | |
2627 | * | |
2628 | * This improves reliability: if user state gets corrupted such that | |
2629 | * the stack pointer points very close to the end of the signal stack, | |
2630 | * then this check will enable the signal to be handled anyway. | |
2631 | */ | |
2632 | if (current->sas_ss_flags & SS_AUTODISARM) | |
2633 | return 0; | |
2634 | ||
2a855dd0 SAS |
2635 | #ifdef CONFIG_STACK_GROWSUP |
2636 | return sp >= current->sas_ss_sp && | |
2637 | sp - current->sas_ss_sp < current->sas_ss_size; | |
2638 | #else | |
2639 | return sp > current->sas_ss_sp && | |
2640 | sp - current->sas_ss_sp <= current->sas_ss_size; | |
2641 | #endif | |
1da177e4 LT |
2642 | } |
2643 | ||
2644 | static inline int sas_ss_flags(unsigned long sp) | |
2645 | { | |
72f15c03 RW |
2646 | if (!current->sas_ss_size) |
2647 | return SS_DISABLE; | |
2648 | ||
2649 | return on_sig_stack(sp) ? SS_ONSTACK : 0; | |
1da177e4 LT |
2650 | } |
2651 | ||
2a742138 SS |
2652 | static inline void sas_ss_reset(struct task_struct *p) |
2653 | { | |
2654 | p->sas_ss_sp = 0; | |
2655 | p->sas_ss_size = 0; | |
2656 | p->sas_ss_flags = SS_DISABLE; | |
2657 | } | |
2658 | ||
5a1b98d3 AV |
2659 | static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig) |
2660 | { | |
2661 | if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp)) | |
2662 | #ifdef CONFIG_STACK_GROWSUP | |
2663 | return current->sas_ss_sp; | |
2664 | #else | |
2665 | return current->sas_ss_sp + current->sas_ss_size; | |
2666 | #endif | |
2667 | return sp; | |
2668 | } | |
2669 | ||
1da177e4 LT |
2670 | /* |
2671 | * Routines for handling mm_structs | |
2672 | */ | |
2673 | extern struct mm_struct * mm_alloc(void); | |
2674 | ||
f1f10076 VN |
2675 | /** |
2676 | * mmgrab() - Pin a &struct mm_struct. | |
2677 | * @mm: The &struct mm_struct to pin. | |
2678 | * | |
2679 | * Make sure that @mm will not get freed even after the owning task | |
2680 | * exits. This doesn't guarantee that the associated address space | |
2681 | * will still exist later on and mmget_not_zero() has to be used before | |
2682 | * accessing it. | |
2683 | * | |
2684 | * This is a preferred way to to pin @mm for a longer/unbounded amount | |
2685 | * of time. | |
2686 | * | |
2687 | * Use mmdrop() to release the reference acquired by mmgrab(). | |
2688 | * | |
2689 | * See also <Documentation/vm/active_mm.txt> for an in-depth explanation | |
2690 | * of &mm_struct.mm_count vs &mm_struct.mm_users. | |
2691 | */ | |
2692 | static inline void mmgrab(struct mm_struct *mm) | |
2693 | { | |
2694 | atomic_inc(&mm->mm_count); | |
2695 | } | |
2696 | ||
1da177e4 | 2697 | /* mmdrop drops the mm and the page tables */ |
b3c97528 | 2698 | extern void __mmdrop(struct mm_struct *); |
d2005e3f | 2699 | static inline void mmdrop(struct mm_struct *mm) |
1da177e4 | 2700 | { |
6fb43d7b | 2701 | if (unlikely(atomic_dec_and_test(&mm->mm_count))) |
1da177e4 LT |
2702 | __mmdrop(mm); |
2703 | } | |
2704 | ||
7283094e MH |
2705 | static inline void mmdrop_async_fn(struct work_struct *work) |
2706 | { | |
2707 | struct mm_struct *mm = container_of(work, struct mm_struct, async_put_work); | |
2708 | __mmdrop(mm); | |
2709 | } | |
2710 | ||
2711 | static inline void mmdrop_async(struct mm_struct *mm) | |
2712 | { | |
2713 | if (unlikely(atomic_dec_and_test(&mm->mm_count))) { | |
2714 | INIT_WORK(&mm->async_put_work, mmdrop_async_fn); | |
2715 | schedule_work(&mm->async_put_work); | |
2716 | } | |
2717 | } | |
2718 | ||
3fce371b VN |
2719 | /** |
2720 | * mmget() - Pin the address space associated with a &struct mm_struct. | |
2721 | * @mm: The address space to pin. | |
2722 | * | |
2723 | * Make sure that the address space of the given &struct mm_struct doesn't | |
2724 | * go away. This does not protect against parts of the address space being | |
2725 | * modified or freed, however. | |
2726 | * | |
2727 | * Never use this function to pin this address space for an | |
2728 | * unbounded/indefinite amount of time. | |
2729 | * | |
2730 | * Use mmput() to release the reference acquired by mmget(). | |
2731 | * | |
2732 | * See also <Documentation/vm/active_mm.txt> for an in-depth explanation | |
2733 | * of &mm_struct.mm_count vs &mm_struct.mm_users. | |
2734 | */ | |
2735 | static inline void mmget(struct mm_struct *mm) | |
2736 | { | |
2737 | atomic_inc(&mm->mm_users); | |
2738 | } | |
2739 | ||
d2005e3f ON |
2740 | static inline bool mmget_not_zero(struct mm_struct *mm) |
2741 | { | |
2742 | return atomic_inc_not_zero(&mm->mm_users); | |
2743 | } | |
2744 | ||
1da177e4 LT |
2745 | /* mmput gets rid of the mappings and all user-space */ |
2746 | extern void mmput(struct mm_struct *); | |
7ef949d7 MH |
2747 | #ifdef CONFIG_MMU |
2748 | /* same as above but performs the slow path from the async context. Can | |
ec8d7c14 MH |
2749 | * be called from the atomic context as well |
2750 | */ | |
2751 | extern void mmput_async(struct mm_struct *); | |
7ef949d7 | 2752 | #endif |
ec8d7c14 | 2753 | |
1da177e4 LT |
2754 | /* Grab a reference to a task's mm, if it is not already going away */ |
2755 | extern struct mm_struct *get_task_mm(struct task_struct *task); | |
8cdb878d CY |
2756 | /* |
2757 | * Grab a reference to a task's mm, if it is not already going away | |
2758 | * and ptrace_may_access with the mode parameter passed to it | |
2759 | * succeeds. | |
2760 | */ | |
2761 | extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode); | |
1da177e4 LT |
2762 | /* Remove the current tasks stale references to the old mm_struct */ |
2763 | extern void mm_release(struct task_struct *, struct mm_struct *); | |
2764 | ||
3033f14a JT |
2765 | #ifdef CONFIG_HAVE_COPY_THREAD_TLS |
2766 | extern int copy_thread_tls(unsigned long, unsigned long, unsigned long, | |
2767 | struct task_struct *, unsigned long); | |
2768 | #else | |
6f2c55b8 | 2769 | extern int copy_thread(unsigned long, unsigned long, unsigned long, |
afa86fc4 | 2770 | struct task_struct *); |
3033f14a JT |
2771 | |
2772 | /* Architectures that haven't opted into copy_thread_tls get the tls argument | |
2773 | * via pt_regs, so ignore the tls argument passed via C. */ | |
2774 | static inline int copy_thread_tls( | |
2775 | unsigned long clone_flags, unsigned long sp, unsigned long arg, | |
2776 | struct task_struct *p, unsigned long tls) | |
2777 | { | |
2778 | return copy_thread(clone_flags, sp, arg, p); | |
2779 | } | |
2780 | #endif | |
1da177e4 | 2781 | extern void flush_thread(void); |
5f56a5df JS |
2782 | |
2783 | #ifdef CONFIG_HAVE_EXIT_THREAD | |
e6464694 | 2784 | extern void exit_thread(struct task_struct *tsk); |
5f56a5df | 2785 | #else |
e6464694 | 2786 | static inline void exit_thread(struct task_struct *tsk) |
5f56a5df JS |
2787 | { |
2788 | } | |
2789 | #endif | |
1da177e4 | 2790 | |
1da177e4 | 2791 | extern void exit_files(struct task_struct *); |
a7e5328a | 2792 | extern void __cleanup_sighand(struct sighand_struct *); |
cbaffba1 | 2793 | |
1da177e4 | 2794 | extern void exit_itimers(struct signal_struct *); |
cbaffba1 | 2795 | extern void flush_itimer_signals(void); |
1da177e4 | 2796 | |
9402c95f | 2797 | extern void do_group_exit(int); |
1da177e4 | 2798 | |
c4ad8f98 | 2799 | extern int do_execve(struct filename *, |
d7627467 | 2800 | const char __user * const __user *, |
da3d4c5f | 2801 | const char __user * const __user *); |
51f39a1f DD |
2802 | extern int do_execveat(int, struct filename *, |
2803 | const char __user * const __user *, | |
2804 | const char __user * const __user *, | |
2805 | int); | |
3033f14a | 2806 | extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long); |
e80d6661 | 2807 | extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *); |
36c8b586 | 2808 | struct task_struct *fork_idle(int); |
2aa3a7f8 | 2809 | extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags); |
1da177e4 | 2810 | |
82b89778 AH |
2811 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
2812 | static inline void set_task_comm(struct task_struct *tsk, const char *from) | |
2813 | { | |
2814 | __set_task_comm(tsk, from, false); | |
2815 | } | |
59714d65 | 2816 | extern char *get_task_comm(char *to, struct task_struct *tsk); |
1da177e4 LT |
2817 | |
2818 | #ifdef CONFIG_SMP | |
317f3941 | 2819 | void scheduler_ipi(void); |
85ba2d86 | 2820 | extern unsigned long wait_task_inactive(struct task_struct *, long match_state); |
1da177e4 | 2821 | #else |
184748cc | 2822 | static inline void scheduler_ipi(void) { } |
85ba2d86 RM |
2823 | static inline unsigned long wait_task_inactive(struct task_struct *p, |
2824 | long match_state) | |
2825 | { | |
2826 | return 1; | |
2827 | } | |
1da177e4 LT |
2828 | #endif |
2829 | ||
fafe870f FW |
2830 | #define tasklist_empty() \ |
2831 | list_empty(&init_task.tasks) | |
2832 | ||
05725f7e JP |
2833 | #define next_task(p) \ |
2834 | list_entry_rcu((p)->tasks.next, struct task_struct, tasks) | |
1da177e4 LT |
2835 | |
2836 | #define for_each_process(p) \ | |
2837 | for (p = &init_task ; (p = next_task(p)) != &init_task ; ) | |
2838 | ||
5bb459bb | 2839 | extern bool current_is_single_threaded(void); |
d84f4f99 | 2840 | |
1da177e4 LT |
2841 | /* |
2842 | * Careful: do_each_thread/while_each_thread is a double loop so | |
2843 | * 'break' will not work as expected - use goto instead. | |
2844 | */ | |
2845 | #define do_each_thread(g, t) \ | |
2846 | for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do | |
2847 | ||
2848 | #define while_each_thread(g, t) \ | |
2849 | while ((t = next_thread(t)) != g) | |
2850 | ||
0c740d0a ON |
2851 | #define __for_each_thread(signal, t) \ |
2852 | list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node) | |
2853 | ||
2854 | #define for_each_thread(p, t) \ | |
2855 | __for_each_thread((p)->signal, t) | |
2856 | ||
2857 | /* Careful: this is a double loop, 'break' won't work as expected. */ | |
2858 | #define for_each_process_thread(p, t) \ | |
2859 | for_each_process(p) for_each_thread(p, t) | |
2860 | ||
0f1b92cb ON |
2861 | typedef int (*proc_visitor)(struct task_struct *p, void *data); |
2862 | void walk_process_tree(struct task_struct *top, proc_visitor, void *); | |
2863 | ||
7e49827c ON |
2864 | static inline int get_nr_threads(struct task_struct *tsk) |
2865 | { | |
b3ac022c | 2866 | return tsk->signal->nr_threads; |
7e49827c ON |
2867 | } |
2868 | ||
087806b1 ON |
2869 | static inline bool thread_group_leader(struct task_struct *p) |
2870 | { | |
2871 | return p->exit_signal >= 0; | |
2872 | } | |
1da177e4 | 2873 | |
0804ef4b EB |
2874 | /* Do to the insanities of de_thread it is possible for a process |
2875 | * to have the pid of the thread group leader without actually being | |
2876 | * the thread group leader. For iteration through the pids in proc | |
2877 | * all we care about is that we have a task with the appropriate | |
2878 | * pid, we don't actually care if we have the right task. | |
2879 | */ | |
e1403b8e | 2880 | static inline bool has_group_leader_pid(struct task_struct *p) |
0804ef4b | 2881 | { |
e1403b8e | 2882 | return task_pid(p) == p->signal->leader_pid; |
0804ef4b EB |
2883 | } |
2884 | ||
bac0abd6 | 2885 | static inline |
e1403b8e | 2886 | bool same_thread_group(struct task_struct *p1, struct task_struct *p2) |
bac0abd6 | 2887 | { |
e1403b8e | 2888 | return p1->signal == p2->signal; |
bac0abd6 PE |
2889 | } |
2890 | ||
36c8b586 | 2891 | static inline struct task_struct *next_thread(const struct task_struct *p) |
47e65328 | 2892 | { |
05725f7e JP |
2893 | return list_entry_rcu(p->thread_group.next, |
2894 | struct task_struct, thread_group); | |
47e65328 ON |
2895 | } |
2896 | ||
e868171a | 2897 | static inline int thread_group_empty(struct task_struct *p) |
1da177e4 | 2898 | { |
47e65328 | 2899 | return list_empty(&p->thread_group); |
1da177e4 LT |
2900 | } |
2901 | ||
2902 | #define delay_group_leader(p) \ | |
2903 | (thread_group_leader(p) && !thread_group_empty(p)) | |
2904 | ||
1da177e4 | 2905 | /* |
260ea101 | 2906 | * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring |
22e2c507 | 2907 | * subscriptions and synchronises with wait4(). Also used in procfs. Also |
ddbcc7e8 | 2908 | * pins the final release of task.io_context. Also protects ->cpuset and |
d68b46fe | 2909 | * ->cgroup.subsys[]. And ->vfork_done. |
1da177e4 LT |
2910 | * |
2911 | * Nests both inside and outside of read_lock(&tasklist_lock). | |
2912 | * It must not be nested with write_lock_irq(&tasklist_lock), | |
2913 | * neither inside nor outside. | |
2914 | */ | |
2915 | static inline void task_lock(struct task_struct *p) | |
2916 | { | |
2917 | spin_lock(&p->alloc_lock); | |
2918 | } | |
2919 | ||
2920 | static inline void task_unlock(struct task_struct *p) | |
2921 | { | |
2922 | spin_unlock(&p->alloc_lock); | |
2923 | } | |
2924 | ||
b8ed374e | 2925 | extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk, |
f63ee72e ON |
2926 | unsigned long *flags); |
2927 | ||
9388dc30 AV |
2928 | static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk, |
2929 | unsigned long *flags) | |
2930 | { | |
2931 | struct sighand_struct *ret; | |
2932 | ||
2933 | ret = __lock_task_sighand(tsk, flags); | |
2934 | (void)__cond_lock(&tsk->sighand->siglock, ret); | |
2935 | return ret; | |
2936 | } | |
b8ed374e | 2937 | |
f63ee72e ON |
2938 | static inline void unlock_task_sighand(struct task_struct *tsk, |
2939 | unsigned long *flags) | |
2940 | { | |
2941 | spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); | |
2942 | } | |
2943 | ||
c65eacbe AL |
2944 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
2945 | ||
2946 | static inline struct thread_info *task_thread_info(struct task_struct *task) | |
2947 | { | |
2948 | return &task->thread_info; | |
2949 | } | |
c6c314a6 AL |
2950 | |
2951 | /* | |
2952 | * When accessing the stack of a non-current task that might exit, use | |
2953 | * try_get_task_stack() instead. task_stack_page will return a pointer | |
2954 | * that could get freed out from under you. | |
2955 | */ | |
c65eacbe AL |
2956 | static inline void *task_stack_page(const struct task_struct *task) |
2957 | { | |
2958 | return task->stack; | |
2959 | } | |
c6c314a6 | 2960 | |
c65eacbe | 2961 | #define setup_thread_stack(new,old) do { } while(0) |
c6c314a6 | 2962 | |
c65eacbe AL |
2963 | static inline unsigned long *end_of_stack(const struct task_struct *task) |
2964 | { | |
2965 | return task->stack; | |
2966 | } | |
2967 | ||
2968 | #elif !defined(__HAVE_THREAD_FUNCTIONS) | |
f037360f | 2969 | |
f7e4217b | 2970 | #define task_thread_info(task) ((struct thread_info *)(task)->stack) |
c65eacbe | 2971 | #define task_stack_page(task) ((void *)(task)->stack) |
a1261f54 | 2972 | |
10ebffde AV |
2973 | static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) |
2974 | { | |
2975 | *task_thread_info(p) = *task_thread_info(org); | |
2976 | task_thread_info(p)->task = p; | |
2977 | } | |
2978 | ||
6a40281a CE |
2979 | /* |
2980 | * Return the address of the last usable long on the stack. | |
2981 | * | |
2982 | * When the stack grows down, this is just above the thread | |
2983 | * info struct. Going any lower will corrupt the threadinfo. | |
2984 | * | |
2985 | * When the stack grows up, this is the highest address. | |
2986 | * Beyond that position, we corrupt data on the next page. | |
2987 | */ | |
10ebffde AV |
2988 | static inline unsigned long *end_of_stack(struct task_struct *p) |
2989 | { | |
6a40281a CE |
2990 | #ifdef CONFIG_STACK_GROWSUP |
2991 | return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1; | |
2992 | #else | |
f7e4217b | 2993 | return (unsigned long *)(task_thread_info(p) + 1); |
6a40281a | 2994 | #endif |
10ebffde AV |
2995 | } |
2996 | ||
f037360f | 2997 | #endif |
c6c314a6 | 2998 | |
68f24b08 AL |
2999 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
3000 | static inline void *try_get_task_stack(struct task_struct *tsk) | |
3001 | { | |
3002 | return atomic_inc_not_zero(&tsk->stack_refcount) ? | |
3003 | task_stack_page(tsk) : NULL; | |
3004 | } | |
3005 | ||
3006 | extern void put_task_stack(struct task_struct *tsk); | |
3007 | #else | |
c6c314a6 AL |
3008 | static inline void *try_get_task_stack(struct task_struct *tsk) |
3009 | { | |
3010 | return task_stack_page(tsk); | |
3011 | } | |
3012 | ||
3013 | static inline void put_task_stack(struct task_struct *tsk) {} | |
68f24b08 | 3014 | #endif |
c6c314a6 | 3015 | |
a70857e4 AT |
3016 | #define task_stack_end_corrupted(task) \ |
3017 | (*(end_of_stack(task)) != STACK_END_MAGIC) | |
f037360f | 3018 | |
8b05c7e6 FT |
3019 | static inline int object_is_on_stack(void *obj) |
3020 | { | |
3021 | void *stack = task_stack_page(current); | |
3022 | ||
3023 | return (obj >= stack) && (obj < (stack + THREAD_SIZE)); | |
3024 | } | |
3025 | ||
b235beea | 3026 | extern void thread_stack_cache_init(void); |
8c9843e5 | 3027 | |
7c9f8861 ES |
3028 | #ifdef CONFIG_DEBUG_STACK_USAGE |
3029 | static inline unsigned long stack_not_used(struct task_struct *p) | |
3030 | { | |
3031 | unsigned long *n = end_of_stack(p); | |
3032 | ||
3033 | do { /* Skip over canary */ | |
6c31da34 HD |
3034 | # ifdef CONFIG_STACK_GROWSUP |
3035 | n--; | |
3036 | # else | |
7c9f8861 | 3037 | n++; |
6c31da34 | 3038 | # endif |
7c9f8861 ES |
3039 | } while (!*n); |
3040 | ||
6c31da34 HD |
3041 | # ifdef CONFIG_STACK_GROWSUP |
3042 | return (unsigned long)end_of_stack(p) - (unsigned long)n; | |
3043 | # else | |
7c9f8861 | 3044 | return (unsigned long)n - (unsigned long)end_of_stack(p); |
6c31da34 | 3045 | # endif |
7c9f8861 ES |
3046 | } |
3047 | #endif | |
d4311ff1 | 3048 | extern void set_task_stack_end_magic(struct task_struct *tsk); |
7c9f8861 | 3049 | |
1da177e4 LT |
3050 | /* set thread flags in other task's structures |
3051 | * - see asm/thread_info.h for TIF_xxxx flags available | |
3052 | */ | |
3053 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3054 | { | |
a1261f54 | 3055 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3056 | } |
3057 | ||
3058 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3059 | { | |
a1261f54 | 3060 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3061 | } |
3062 | ||
3063 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3064 | { | |
a1261f54 | 3065 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3066 | } |
3067 | ||
3068 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3069 | { | |
a1261f54 | 3070 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3071 | } |
3072 | ||
3073 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3074 | { | |
a1261f54 | 3075 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3076 | } |
3077 | ||
3078 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
3079 | { | |
3080 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
3081 | } | |
3082 | ||
3083 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
3084 | { | |
3085 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
3086 | } | |
3087 | ||
8ae121ac GH |
3088 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
3089 | { | |
3090 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
3091 | } | |
3092 | ||
690cc3ff EB |
3093 | static inline int restart_syscall(void) |
3094 | { | |
3095 | set_tsk_thread_flag(current, TIF_SIGPENDING); | |
3096 | return -ERESTARTNOINTR; | |
3097 | } | |
3098 | ||
1da177e4 LT |
3099 | static inline int signal_pending(struct task_struct *p) |
3100 | { | |
3101 | return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); | |
3102 | } | |
f776d12d | 3103 | |
d9588725 RM |
3104 | static inline int __fatal_signal_pending(struct task_struct *p) |
3105 | { | |
3106 | return unlikely(sigismember(&p->pending.signal, SIGKILL)); | |
3107 | } | |
f776d12d MW |
3108 | |
3109 | static inline int fatal_signal_pending(struct task_struct *p) | |
3110 | { | |
3111 | return signal_pending(p) && __fatal_signal_pending(p); | |
3112 | } | |
3113 | ||
16882c1e ON |
3114 | static inline int signal_pending_state(long state, struct task_struct *p) |
3115 | { | |
3116 | if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) | |
3117 | return 0; | |
3118 | if (!signal_pending(p)) | |
3119 | return 0; | |
3120 | ||
16882c1e ON |
3121 | return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); |
3122 | } | |
3123 | ||
1da177e4 LT |
3124 | /* |
3125 | * cond_resched() and cond_resched_lock(): latency reduction via | |
3126 | * explicit rescheduling in places that are safe. The return | |
3127 | * value indicates whether a reschedule was done in fact. | |
3128 | * cond_resched_lock() will drop the spinlock before scheduling, | |
3129 | * cond_resched_softirq() will enable bhs before scheduling. | |
3130 | */ | |
35a773a0 | 3131 | #ifndef CONFIG_PREEMPT |
c3921ab7 | 3132 | extern int _cond_resched(void); |
35a773a0 PZ |
3133 | #else |
3134 | static inline int _cond_resched(void) { return 0; } | |
3135 | #endif | |
6f80bd98 | 3136 | |
613afbf8 | 3137 | #define cond_resched() ({ \ |
3427445a | 3138 | ___might_sleep(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
3139 | _cond_resched(); \ |
3140 | }) | |
6f80bd98 | 3141 | |
613afbf8 FW |
3142 | extern int __cond_resched_lock(spinlock_t *lock); |
3143 | ||
3144 | #define cond_resched_lock(lock) ({ \ | |
3427445a | 3145 | ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ |
613afbf8 FW |
3146 | __cond_resched_lock(lock); \ |
3147 | }) | |
3148 | ||
3149 | extern int __cond_resched_softirq(void); | |
3150 | ||
75e1056f | 3151 | #define cond_resched_softirq() ({ \ |
3427445a | 3152 | ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \ |
75e1056f | 3153 | __cond_resched_softirq(); \ |
613afbf8 | 3154 | }) |
1da177e4 | 3155 | |
f6f3c437 SH |
3156 | static inline void cond_resched_rcu(void) |
3157 | { | |
3158 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
3159 | rcu_read_unlock(); | |
3160 | cond_resched(); | |
3161 | rcu_read_lock(); | |
3162 | #endif | |
3163 | } | |
3164 | ||
1da177e4 LT |
3165 | /* |
3166 | * Does a critical section need to be broken due to another | |
95c354fe NP |
3167 | * task waiting?: (technically does not depend on CONFIG_PREEMPT, |
3168 | * but a general need for low latency) | |
1da177e4 | 3169 | */ |
95c354fe | 3170 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 3171 | { |
95c354fe NP |
3172 | #ifdef CONFIG_PREEMPT |
3173 | return spin_is_contended(lock); | |
3174 | #else | |
1da177e4 | 3175 | return 0; |
95c354fe | 3176 | #endif |
1da177e4 LT |
3177 | } |
3178 | ||
75f93fed PZ |
3179 | static __always_inline bool need_resched(void) |
3180 | { | |
3181 | return unlikely(tif_need_resched()); | |
3182 | } | |
3183 | ||
f06febc9 FM |
3184 | /* |
3185 | * Thread group CPU time accounting. | |
3186 | */ | |
4cd4c1b4 | 3187 | void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); |
4da94d49 | 3188 | void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); |
f06febc9 | 3189 | |
7bb44ade RM |
3190 | /* |
3191 | * Reevaluate whether the task has signals pending delivery. | |
3192 | * Wake the task if so. | |
3193 | * This is required every time the blocked sigset_t changes. | |
3194 | * callers must hold sighand->siglock. | |
3195 | */ | |
3196 | extern void recalc_sigpending_and_wake(struct task_struct *t); | |
1da177e4 LT |
3197 | extern void recalc_sigpending(void); |
3198 | ||
910ffdb1 ON |
3199 | extern void signal_wake_up_state(struct task_struct *t, unsigned int state); |
3200 | ||
3201 | static inline void signal_wake_up(struct task_struct *t, bool resume) | |
3202 | { | |
3203 | signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0); | |
3204 | } | |
3205 | static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume) | |
3206 | { | |
3207 | signal_wake_up_state(t, resume ? __TASK_TRACED : 0); | |
3208 | } | |
1da177e4 LT |
3209 | |
3210 | /* | |
3211 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
3212 | */ | |
3213 | #ifdef CONFIG_SMP | |
3214 | ||
3215 | static inline unsigned int task_cpu(const struct task_struct *p) | |
3216 | { | |
c65eacbe AL |
3217 | #ifdef CONFIG_THREAD_INFO_IN_TASK |
3218 | return p->cpu; | |
3219 | #else | |
a1261f54 | 3220 | return task_thread_info(p)->cpu; |
c65eacbe | 3221 | #endif |
1da177e4 LT |
3222 | } |
3223 | ||
b32e86b4 IM |
3224 | static inline int task_node(const struct task_struct *p) |
3225 | { | |
3226 | return cpu_to_node(task_cpu(p)); | |
3227 | } | |
3228 | ||
c65cc870 | 3229 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
3230 | |
3231 | #else | |
3232 | ||
3233 | static inline unsigned int task_cpu(const struct task_struct *p) | |
3234 | { | |
3235 | return 0; | |
3236 | } | |
3237 | ||
3238 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
3239 | { | |
3240 | } | |
3241 | ||
3242 | #endif /* CONFIG_SMP */ | |
3243 | ||
d9345c65 PX |
3244 | /* |
3245 | * In order to reduce various lock holder preemption latencies provide an | |
3246 | * interface to see if a vCPU is currently running or not. | |
3247 | * | |
3248 | * This allows us to terminate optimistic spin loops and block, analogous to | |
3249 | * the native optimistic spin heuristic of testing if the lock owner task is | |
3250 | * running or not. | |
3251 | */ | |
3252 | #ifndef vcpu_is_preempted | |
3253 | # define vcpu_is_preempted(cpu) false | |
3254 | #endif | |
3255 | ||
96f874e2 RR |
3256 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
3257 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 3258 | |
7c941438 | 3259 | #ifdef CONFIG_CGROUP_SCHED |
07e06b01 | 3260 | extern struct task_group root_task_group; |
8323f26c | 3261 | #endif /* CONFIG_CGROUP_SCHED */ |
9b5b7751 | 3262 | |
54e99124 DG |
3263 | extern int task_can_switch_user(struct user_struct *up, |
3264 | struct task_struct *tsk); | |
3265 | ||
4b98d11b AD |
3266 | #ifdef CONFIG_TASK_XACCT |
3267 | static inline void add_rchar(struct task_struct *tsk, ssize_t amt) | |
3268 | { | |
940389b8 | 3269 | tsk->ioac.rchar += amt; |
4b98d11b AD |
3270 | } |
3271 | ||
3272 | static inline void add_wchar(struct task_struct *tsk, ssize_t amt) | |
3273 | { | |
940389b8 | 3274 | tsk->ioac.wchar += amt; |
4b98d11b AD |
3275 | } |
3276 | ||
3277 | static inline void inc_syscr(struct task_struct *tsk) | |
3278 | { | |
940389b8 | 3279 | tsk->ioac.syscr++; |
4b98d11b AD |
3280 | } |
3281 | ||
3282 | static inline void inc_syscw(struct task_struct *tsk) | |
3283 | { | |
940389b8 | 3284 | tsk->ioac.syscw++; |
4b98d11b AD |
3285 | } |
3286 | #else | |
3287 | static inline void add_rchar(struct task_struct *tsk, ssize_t amt) | |
3288 | { | |
3289 | } | |
3290 | ||
3291 | static inline void add_wchar(struct task_struct *tsk, ssize_t amt) | |
3292 | { | |
3293 | } | |
3294 | ||
3295 | static inline void inc_syscr(struct task_struct *tsk) | |
3296 | { | |
3297 | } | |
3298 | ||
3299 | static inline void inc_syscw(struct task_struct *tsk) | |
3300 | { | |
3301 | } | |
3302 | #endif | |
3303 | ||
82455257 DH |
3304 | #ifndef TASK_SIZE_OF |
3305 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
3306 | #endif | |
3307 | ||
f98bafa0 | 3308 | #ifdef CONFIG_MEMCG |
cf475ad2 | 3309 | extern void mm_update_next_owner(struct mm_struct *mm); |
cf475ad2 BS |
3310 | #else |
3311 | static inline void mm_update_next_owner(struct mm_struct *mm) | |
3312 | { | |
3313 | } | |
f98bafa0 | 3314 | #endif /* CONFIG_MEMCG */ |
cf475ad2 | 3315 | |
3e10e716 JS |
3316 | static inline unsigned long task_rlimit(const struct task_struct *tsk, |
3317 | unsigned int limit) | |
3318 | { | |
316c1608 | 3319 | return READ_ONCE(tsk->signal->rlim[limit].rlim_cur); |
3e10e716 JS |
3320 | } |
3321 | ||
3322 | static inline unsigned long task_rlimit_max(const struct task_struct *tsk, | |
3323 | unsigned int limit) | |
3324 | { | |
316c1608 | 3325 | return READ_ONCE(tsk->signal->rlim[limit].rlim_max); |
3e10e716 JS |
3326 | } |
3327 | ||
3328 | static inline unsigned long rlimit(unsigned int limit) | |
3329 | { | |
3330 | return task_rlimit(current, limit); | |
3331 | } | |
3332 | ||
3333 | static inline unsigned long rlimit_max(unsigned int limit) | |
3334 | { | |
3335 | return task_rlimit_max(current, limit); | |
3336 | } | |
3337 | ||
58919e83 RW |
3338 | #define SCHED_CPUFREQ_RT (1U << 0) |
3339 | #define SCHED_CPUFREQ_DL (1U << 1) | |
8c34ab19 | 3340 | #define SCHED_CPUFREQ_IOWAIT (1U << 2) |
58919e83 RW |
3341 | |
3342 | #define SCHED_CPUFREQ_RT_DL (SCHED_CPUFREQ_RT | SCHED_CPUFREQ_DL) | |
3343 | ||
adaf9fcd RW |
3344 | #ifdef CONFIG_CPU_FREQ |
3345 | struct update_util_data { | |
58919e83 | 3346 | void (*func)(struct update_util_data *data, u64 time, unsigned int flags); |
adaf9fcd RW |
3347 | }; |
3348 | ||
0bed612b | 3349 | void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data, |
58919e83 RW |
3350 | void (*func)(struct update_util_data *data, u64 time, |
3351 | unsigned int flags)); | |
0bed612b | 3352 | void cpufreq_remove_update_util_hook(int cpu); |
adaf9fcd RW |
3353 | #endif /* CONFIG_CPU_FREQ */ |
3354 | ||
1da177e4 | 3355 | #endif |