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> | |
fb00aca4 | 21 | #include <linux/plist.h> |
1da177e4 LT |
22 | #include <linux/rbtree.h> |
23 | #include <linux/thread_info.h> | |
24 | #include <linux/cpumask.h> | |
25 | #include <linux/errno.h> | |
26 | #include <linux/nodemask.h> | |
c92ff1bd | 27 | #include <linux/mm_types.h> |
92cf2118 | 28 | #include <linux/preempt.h> |
1da177e4 | 29 | |
1da177e4 LT |
30 | #include <asm/page.h> |
31 | #include <asm/ptrace.h> | |
bfc3f028 | 32 | #include <linux/cputime.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> | |
40 | #include <linux/pid.h> | |
41 | #include <linux/percpu.h> | |
42 | #include <linux/topology.h> | |
43 | #include <linux/seccomp.h> | |
e56d0903 | 44 | #include <linux/rcupdate.h> |
05725f7e | 45 | #include <linux/rculist.h> |
23f78d4a | 46 | #include <linux/rtmutex.h> |
1da177e4 | 47 | |
a3b6714e DW |
48 | #include <linux/time.h> |
49 | #include <linux/param.h> | |
50 | #include <linux/resource.h> | |
51 | #include <linux/timer.h> | |
52 | #include <linux/hrtimer.h> | |
5c9a8750 | 53 | #include <linux/kcov.h> |
7c3ab738 | 54 | #include <linux/task_io_accounting.h> |
9745512c | 55 | #include <linux/latencytop.h> |
9e2b2dc4 | 56 | #include <linux/cred.h> |
fa14ff4a | 57 | #include <linux/llist.h> |
7b44ab97 | 58 | #include <linux/uidgid.h> |
21caf2fc | 59 | #include <linux/gfp.h> |
d4311ff1 | 60 | #include <linux/magic.h> |
7d7efec3 | 61 | #include <linux/cgroup-defs.h> |
91a034ed | 62 | #include <linux/rseq.h> |
a3b6714e DW |
63 | |
64 | #include <asm/processor.h> | |
36d57ac4 | 65 | |
d50dde5a DF |
66 | #define SCHED_ATTR_SIZE_VER0 48 /* sizeof first published struct */ |
67 | ||
68 | /* | |
69 | * Extended scheduling parameters data structure. | |
70 | * | |
71 | * This is needed because the original struct sched_param can not be | |
72 | * altered without introducing ABI issues with legacy applications | |
73 | * (e.g., in sched_getparam()). | |
74 | * | |
75 | * However, the possibility of specifying more than just a priority for | |
76 | * the tasks may be useful for a wide variety of application fields, e.g., | |
77 | * multimedia, streaming, automation and control, and many others. | |
78 | * | |
79 | * This variant (sched_attr) is meant at describing a so-called | |
80 | * sporadic time-constrained task. In such model a task is specified by: | |
81 | * - the activation period or minimum instance inter-arrival time; | |
82 | * - the maximum (or average, depending on the actual scheduling | |
83 | * discipline) computation time of all instances, a.k.a. runtime; | |
84 | * - the deadline (relative to the actual activation time) of each | |
85 | * instance. | |
86 | * Very briefly, a periodic (sporadic) task asks for the execution of | |
87 | * some specific computation --which is typically called an instance-- | |
88 | * (at most) every period. Moreover, each instance typically lasts no more | |
89 | * than the runtime and must be completed by time instant t equal to | |
90 | * the instance activation time + the deadline. | |
91 | * | |
92 | * This is reflected by the actual fields of the sched_attr structure: | |
93 | * | |
94 | * @size size of the structure, for fwd/bwd compat. | |
95 | * | |
96 | * @sched_policy task's scheduling policy | |
97 | * @sched_flags for customizing the scheduler behaviour | |
98 | * @sched_nice task's nice value (SCHED_NORMAL/BATCH) | |
99 | * @sched_priority task's static priority (SCHED_FIFO/RR) | |
100 | * @sched_deadline representative of the task's deadline | |
101 | * @sched_runtime representative of the task's runtime | |
102 | * @sched_period representative of the task's period | |
103 | * | |
104 | * Given this task model, there are a multiplicity of scheduling algorithms | |
105 | * and policies, that can be used to ensure all the tasks will make their | |
106 | * timing constraints. | |
aab03e05 DF |
107 | * |
108 | * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the | |
109 | * only user of this new interface. More information about the algorithm | |
110 | * available in the scheduling class file or in Documentation/. | |
d50dde5a DF |
111 | */ |
112 | struct sched_attr { | |
113 | u32 size; | |
114 | ||
115 | u32 sched_policy; | |
116 | u64 sched_flags; | |
117 | ||
118 | /* SCHED_NORMAL, SCHED_BATCH */ | |
119 | s32 sched_nice; | |
120 | ||
121 | /* SCHED_FIFO, SCHED_RR */ | |
122 | u32 sched_priority; | |
123 | ||
124 | /* SCHED_DEADLINE */ | |
125 | u64 sched_runtime; | |
126 | u64 sched_deadline; | |
127 | u64 sched_period; | |
128 | }; | |
129 | ||
c87e2837 | 130 | struct futex_pi_state; |
286100a6 | 131 | struct robust_list_head; |
bddd87c7 | 132 | struct bio_list; |
5ad4e53b | 133 | struct fs_struct; |
cdd6c482 | 134 | struct perf_event_context; |
73c10101 | 135 | struct blk_plug; |
c4ad8f98 | 136 | struct filename; |
89076bc3 | 137 | struct nameidata; |
1da177e4 | 138 | |
615d6e87 DB |
139 | #define VMACACHE_BITS 2 |
140 | #define VMACACHE_SIZE (1U << VMACACHE_BITS) | |
141 | #define VMACACHE_MASK (VMACACHE_SIZE - 1) | |
142 | ||
1da177e4 LT |
143 | /* |
144 | * These are the constant used to fake the fixed-point load-average | |
145 | * counting. Some notes: | |
146 | * - 11 bit fractions expand to 22 bits by the multiplies: this gives | |
147 | * a load-average precision of 10 bits integer + 11 bits fractional | |
148 | * - if you want to count load-averages more often, you need more | |
149 | * precision, or rounding will get you. With 2-second counting freq, | |
150 | * the EXP_n values would be 1981, 2034 and 2043 if still using only | |
151 | * 11 bit fractions. | |
152 | */ | |
153 | extern unsigned long avenrun[]; /* Load averages */ | |
2d02494f | 154 | extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); |
1da177e4 LT |
155 | |
156 | #define FSHIFT 11 /* nr of bits of precision */ | |
157 | #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ | |
0c2043ab | 158 | #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ |
1da177e4 LT |
159 | #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ |
160 | #define EXP_5 2014 /* 1/exp(5sec/5min) */ | |
161 | #define EXP_15 2037 /* 1/exp(5sec/15min) */ | |
162 | ||
163 | #define CALC_LOAD(load,exp,n) \ | |
164 | load *= exp; \ | |
165 | load += n*(FIXED_1-exp); \ | |
166 | load >>= FSHIFT; | |
167 | ||
168 | extern unsigned long total_forks; | |
169 | extern int nr_threads; | |
1da177e4 LT |
170 | DECLARE_PER_CPU(unsigned long, process_counts); |
171 | extern int nr_processes(void); | |
172 | extern unsigned long nr_running(void); | |
2ee507c4 | 173 | extern bool single_task_running(void); |
1da177e4 | 174 | extern unsigned long nr_iowait(void); |
8c215bd3 | 175 | extern unsigned long nr_iowait_cpu(int cpu); |
372ba8cb | 176 | extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load); |
69d25870 | 177 | |
0f004f5a | 178 | extern void calc_global_load(unsigned long ticks); |
3289bdb4 PZ |
179 | |
180 | #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) | |
1f41906a FW |
181 | extern void cpu_load_update_nohz_start(void); |
182 | extern void cpu_load_update_nohz_stop(void); | |
3289bdb4 | 183 | #else |
1f41906a FW |
184 | static inline void cpu_load_update_nohz_start(void) { } |
185 | static inline void cpu_load_update_nohz_stop(void) { } | |
3289bdb4 | 186 | #endif |
1da177e4 | 187 | |
b637a328 PM |
188 | extern void dump_cpu_task(int cpu); |
189 | ||
43ae34cb IM |
190 | struct seq_file; |
191 | struct cfs_rq; | |
4cf86d77 | 192 | struct task_group; |
43ae34cb IM |
193 | #ifdef CONFIG_SCHED_DEBUG |
194 | extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); | |
195 | extern void proc_sched_set_task(struct task_struct *p); | |
43ae34cb | 196 | #endif |
1da177e4 | 197 | |
4a8342d2 LT |
198 | /* |
199 | * Task state bitmask. NOTE! These bits are also | |
200 | * encoded in fs/proc/array.c: get_task_state(). | |
201 | * | |
202 | * We have two separate sets of flags: task->state | |
203 | * is about runnability, while task->exit_state are | |
204 | * about the task exiting. Confusing, but this way | |
205 | * modifying one set can't modify the other one by | |
206 | * mistake. | |
207 | */ | |
1da177e4 LT |
208 | #define TASK_RUNNING 0 |
209 | #define TASK_INTERRUPTIBLE 1 | |
210 | #define TASK_UNINTERRUPTIBLE 2 | |
f021a3c2 MW |
211 | #define __TASK_STOPPED 4 |
212 | #define __TASK_TRACED 8 | |
4a8342d2 | 213 | /* in tsk->exit_state */ |
ad86622b ON |
214 | #define EXIT_DEAD 16 |
215 | #define EXIT_ZOMBIE 32 | |
abd50b39 | 216 | #define EXIT_TRACE (EXIT_ZOMBIE | EXIT_DEAD) |
4a8342d2 | 217 | /* in tsk->state again */ |
af927232 | 218 | #define TASK_DEAD 64 |
f021a3c2 | 219 | #define TASK_WAKEKILL 128 |
e9c84311 | 220 | #define TASK_WAKING 256 |
f2530dc7 | 221 | #define TASK_PARKED 512 |
80ed87c8 | 222 | #define TASK_NOLOAD 1024 |
7dc603c9 PZ |
223 | #define TASK_NEW 2048 |
224 | #define TASK_STATE_MAX 4096 | |
f021a3c2 | 225 | |
7dc603c9 | 226 | #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn" |
73342151 | 227 | |
e1781538 PZ |
228 | extern char ___assert_task_state[1 - 2*!!( |
229 | sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)]; | |
f021a3c2 MW |
230 | |
231 | /* Convenience macros for the sake of set_task_state */ | |
232 | #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) | |
233 | #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) | |
234 | #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) | |
1da177e4 | 235 | |
80ed87c8 PZ |
236 | #define TASK_IDLE (TASK_UNINTERRUPTIBLE | TASK_NOLOAD) |
237 | ||
92a1f4bc MW |
238 | /* Convenience macros for the sake of wake_up */ |
239 | #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) | |
f021a3c2 | 240 | #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) |
92a1f4bc MW |
241 | |
242 | /* get_task_state() */ | |
243 | #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ | |
f021a3c2 | 244 | TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ |
74e37200 | 245 | __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD) |
92a1f4bc | 246 | |
f021a3c2 MW |
247 | #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) |
248 | #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) | |
92a1f4bc | 249 | #define task_is_stopped_or_traced(task) \ |
f021a3c2 | 250 | ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) |
92a1f4bc | 251 | #define task_contributes_to_load(task) \ |
e3c8ca83 | 252 | ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ |
80ed87c8 PZ |
253 | (task->flags & PF_FROZEN) == 0 && \ |
254 | (task->state & TASK_NOLOAD) == 0) | |
1da177e4 | 255 | |
8eb23b9f PZ |
256 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
257 | ||
258 | #define __set_task_state(tsk, state_value) \ | |
259 | do { \ | |
260 | (tsk)->task_state_change = _THIS_IP_; \ | |
261 | (tsk)->state = (state_value); \ | |
262 | } while (0) | |
263 | #define set_task_state(tsk, state_value) \ | |
264 | do { \ | |
265 | (tsk)->task_state_change = _THIS_IP_; \ | |
b92b8b35 | 266 | smp_store_mb((tsk)->state, (state_value)); \ |
8eb23b9f PZ |
267 | } while (0) |
268 | ||
269 | /* | |
270 | * set_current_state() includes a barrier so that the write of current->state | |
271 | * is correctly serialised wrt the caller's subsequent test of whether to | |
272 | * actually sleep: | |
273 | * | |
274 | * set_current_state(TASK_UNINTERRUPTIBLE); | |
275 | * if (do_i_need_to_sleep()) | |
276 | * schedule(); | |
277 | * | |
278 | * If the caller does not need such serialisation then use __set_current_state() | |
279 | */ | |
280 | #define __set_current_state(state_value) \ | |
281 | do { \ | |
282 | current->task_state_change = _THIS_IP_; \ | |
283 | current->state = (state_value); \ | |
284 | } while (0) | |
285 | #define set_current_state(state_value) \ | |
286 | do { \ | |
287 | current->task_state_change = _THIS_IP_; \ | |
b92b8b35 | 288 | smp_store_mb(current->state, (state_value)); \ |
8eb23b9f PZ |
289 | } while (0) |
290 | ||
291 | #else | |
292 | ||
1da177e4 LT |
293 | #define __set_task_state(tsk, state_value) \ |
294 | do { (tsk)->state = (state_value); } while (0) | |
295 | #define set_task_state(tsk, state_value) \ | |
b92b8b35 | 296 | smp_store_mb((tsk)->state, (state_value)) |
1da177e4 | 297 | |
498d0c57 AM |
298 | /* |
299 | * set_current_state() includes a barrier so that the write of current->state | |
300 | * is correctly serialised wrt the caller's subsequent test of whether to | |
301 | * actually sleep: | |
302 | * | |
303 | * set_current_state(TASK_UNINTERRUPTIBLE); | |
304 | * if (do_i_need_to_sleep()) | |
305 | * schedule(); | |
306 | * | |
307 | * If the caller does not need such serialisation then use __set_current_state() | |
308 | */ | |
8eb23b9f | 309 | #define __set_current_state(state_value) \ |
1da177e4 | 310 | do { current->state = (state_value); } while (0) |
8eb23b9f | 311 | #define set_current_state(state_value) \ |
b92b8b35 | 312 | smp_store_mb(current->state, (state_value)) |
1da177e4 | 313 | |
8eb23b9f PZ |
314 | #endif |
315 | ||
1da177e4 LT |
316 | /* Task command name length */ |
317 | #define TASK_COMM_LEN 16 | |
318 | ||
1da177e4 LT |
319 | #include <linux/spinlock.h> |
320 | ||
321 | /* | |
322 | * This serializes "schedule()" and also protects | |
323 | * the run-queue from deletions/modifications (but | |
324 | * _adding_ to the beginning of the run-queue has | |
325 | * a separate lock). | |
326 | */ | |
327 | extern rwlock_t tasklist_lock; | |
328 | extern spinlock_t mmlist_lock; | |
329 | ||
36c8b586 | 330 | struct task_struct; |
1da177e4 | 331 | |
db1466b3 PM |
332 | #ifdef CONFIG_PROVE_RCU |
333 | extern int lockdep_tasklist_lock_is_held(void); | |
334 | #endif /* #ifdef CONFIG_PROVE_RCU */ | |
335 | ||
1da177e4 LT |
336 | extern void sched_init(void); |
337 | extern void sched_init_smp(void); | |
2d07b255 | 338 | extern asmlinkage void schedule_tail(struct task_struct *prev); |
36c8b586 | 339 | extern void init_idle(struct task_struct *idle, int cpu); |
1df21055 | 340 | extern void init_idle_bootup_task(struct task_struct *idle); |
1da177e4 | 341 | |
3fa0818b RR |
342 | extern cpumask_var_t cpu_isolated_map; |
343 | ||
89f19f04 | 344 | extern int runqueue_is_locked(int cpu); |
017730c1 | 345 | |
3451d024 | 346 | #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON) |
c1cc017c | 347 | extern void nohz_balance_enter_idle(int cpu); |
69e1e811 | 348 | extern void set_cpu_sd_state_idle(void); |
bc7a34b8 | 349 | extern int get_nohz_timer_target(void); |
46cb4b7c | 350 | #else |
c1cc017c | 351 | static inline void nohz_balance_enter_idle(int cpu) { } |
fdaabd80 | 352 | static inline void set_cpu_sd_state_idle(void) { } |
46cb4b7c | 353 | #endif |
1da177e4 | 354 | |
e59e2ae2 | 355 | /* |
39bc89fd | 356 | * Only dump TASK_* tasks. (0 for all tasks) |
e59e2ae2 IM |
357 | */ |
358 | extern void show_state_filter(unsigned long state_filter); | |
359 | ||
360 | static inline void show_state(void) | |
361 | { | |
39bc89fd | 362 | show_state_filter(0); |
e59e2ae2 IM |
363 | } |
364 | ||
1da177e4 LT |
365 | extern void show_regs(struct pt_regs *); |
366 | ||
367 | /* | |
368 | * TASK is a pointer to the task whose backtrace we want to see (or NULL for current | |
369 | * task), SP is the stack pointer of the first frame that should be shown in the back | |
370 | * trace (or NULL if the entire call-chain of the task should be shown). | |
371 | */ | |
372 | extern void show_stack(struct task_struct *task, unsigned long *sp); | |
373 | ||
1da177e4 LT |
374 | extern void cpu_init (void); |
375 | extern void trap_init(void); | |
376 | extern void update_process_times(int user); | |
377 | extern void scheduler_tick(void); | |
9cf7243d | 378 | extern int sched_cpu_starting(unsigned int cpu); |
40190a78 TG |
379 | extern int sched_cpu_activate(unsigned int cpu); |
380 | extern int sched_cpu_deactivate(unsigned int cpu); | |
1da177e4 | 381 | |
f2785ddb TG |
382 | #ifdef CONFIG_HOTPLUG_CPU |
383 | extern int sched_cpu_dying(unsigned int cpu); | |
384 | #else | |
385 | # define sched_cpu_dying NULL | |
386 | #endif | |
1da177e4 | 387 | |
82a1fcb9 IM |
388 | extern void sched_show_task(struct task_struct *p); |
389 | ||
19cc36c0 | 390 | #ifdef CONFIG_LOCKUP_DETECTOR |
03e0d461 | 391 | extern void touch_softlockup_watchdog_sched(void); |
8446f1d3 | 392 | extern void touch_softlockup_watchdog(void); |
d6ad3e28 | 393 | extern void touch_softlockup_watchdog_sync(void); |
04c9167f | 394 | extern void touch_all_softlockup_watchdogs(void); |
332fbdbc DZ |
395 | extern int proc_dowatchdog_thresh(struct ctl_table *table, int write, |
396 | void __user *buffer, | |
397 | size_t *lenp, loff_t *ppos); | |
9c44bc03 | 398 | extern unsigned int softlockup_panic; |
ac1f5912 | 399 | extern unsigned int hardlockup_panic; |
004417a6 | 400 | void lockup_detector_init(void); |
8446f1d3 | 401 | #else |
03e0d461 TH |
402 | static inline void touch_softlockup_watchdog_sched(void) |
403 | { | |
404 | } | |
8446f1d3 IM |
405 | static inline void touch_softlockup_watchdog(void) |
406 | { | |
407 | } | |
d6ad3e28 JW |
408 | static inline void touch_softlockup_watchdog_sync(void) |
409 | { | |
410 | } | |
04c9167f JF |
411 | static inline void touch_all_softlockup_watchdogs(void) |
412 | { | |
413 | } | |
004417a6 PZ |
414 | static inline void lockup_detector_init(void) |
415 | { | |
416 | } | |
8446f1d3 IM |
417 | #endif |
418 | ||
8b414521 MT |
419 | #ifdef CONFIG_DETECT_HUNG_TASK |
420 | void reset_hung_task_detector(void); | |
421 | #else | |
422 | static inline void reset_hung_task_detector(void) | |
423 | { | |
424 | } | |
425 | #endif | |
426 | ||
1da177e4 LT |
427 | /* Attach to any functions which should be ignored in wchan output. */ |
428 | #define __sched __attribute__((__section__(".sched.text"))) | |
deaf2227 IM |
429 | |
430 | /* Linker adds these: start and end of __sched functions */ | |
431 | extern char __sched_text_start[], __sched_text_end[]; | |
432 | ||
1da177e4 LT |
433 | /* Is this address in the __sched functions? */ |
434 | extern int in_sched_functions(unsigned long addr); | |
435 | ||
436 | #define MAX_SCHEDULE_TIMEOUT LONG_MAX | |
b3c97528 | 437 | extern signed long schedule_timeout(signed long timeout); |
64ed93a2 | 438 | extern signed long schedule_timeout_interruptible(signed long timeout); |
294d5cc2 | 439 | extern signed long schedule_timeout_killable(signed long timeout); |
64ed93a2 | 440 | extern signed long schedule_timeout_uninterruptible(signed long timeout); |
69b27baf | 441 | extern signed long schedule_timeout_idle(signed long timeout); |
1da177e4 | 442 | asmlinkage void schedule(void); |
c5491ea7 | 443 | extern void schedule_preempt_disabled(void); |
1da177e4 | 444 | |
9cff8ade N |
445 | extern long io_schedule_timeout(long timeout); |
446 | ||
447 | static inline void io_schedule(void) | |
448 | { | |
449 | io_schedule_timeout(MAX_SCHEDULE_TIMEOUT); | |
450 | } | |
451 | ||
ab516013 | 452 | struct nsproxy; |
acce292c | 453 | struct user_namespace; |
1da177e4 | 454 | |
efc1a3b1 DH |
455 | #ifdef CONFIG_MMU |
456 | extern void arch_pick_mmap_layout(struct mm_struct *mm); | |
1da177e4 LT |
457 | extern unsigned long |
458 | arch_get_unmapped_area(struct file *, unsigned long, unsigned long, | |
459 | unsigned long, unsigned long); | |
460 | extern unsigned long | |
461 | arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, | |
462 | unsigned long len, unsigned long pgoff, | |
463 | unsigned long flags); | |
efc1a3b1 DH |
464 | #else |
465 | static inline void arch_pick_mmap_layout(struct mm_struct *mm) {} | |
466 | #endif | |
1da177e4 | 467 | |
d049f74f KC |
468 | #define SUID_DUMP_DISABLE 0 /* No setuid dumping */ |
469 | #define SUID_DUMP_USER 1 /* Dump as user of process */ | |
470 | #define SUID_DUMP_ROOT 2 /* Dump as root */ | |
471 | ||
6c5d5238 | 472 | /* mm flags */ |
f8af4da3 | 473 | |
7288e118 | 474 | /* for SUID_DUMP_* above */ |
3cb4a0bb | 475 | #define MMF_DUMPABLE_BITS 2 |
f8af4da3 | 476 | #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1) |
3cb4a0bb | 477 | |
942be387 ON |
478 | extern void set_dumpable(struct mm_struct *mm, int value); |
479 | /* | |
480 | * This returns the actual value of the suid_dumpable flag. For things | |
481 | * that are using this for checking for privilege transitions, it must | |
482 | * test against SUID_DUMP_USER rather than treating it as a boolean | |
483 | * value. | |
484 | */ | |
485 | static inline int __get_dumpable(unsigned long mm_flags) | |
486 | { | |
487 | return mm_flags & MMF_DUMPABLE_MASK; | |
488 | } | |
489 | ||
490 | static inline int get_dumpable(struct mm_struct *mm) | |
491 | { | |
492 | return __get_dumpable(mm->flags); | |
493 | } | |
494 | ||
3cb4a0bb KH |
495 | /* coredump filter bits */ |
496 | #define MMF_DUMP_ANON_PRIVATE 2 | |
497 | #define MMF_DUMP_ANON_SHARED 3 | |
498 | #define MMF_DUMP_MAPPED_PRIVATE 4 | |
499 | #define MMF_DUMP_MAPPED_SHARED 5 | |
82df3973 | 500 | #define MMF_DUMP_ELF_HEADERS 6 |
e575f111 KM |
501 | #define MMF_DUMP_HUGETLB_PRIVATE 7 |
502 | #define MMF_DUMP_HUGETLB_SHARED 8 | |
5037835c RZ |
503 | #define MMF_DUMP_DAX_PRIVATE 9 |
504 | #define MMF_DUMP_DAX_SHARED 10 | |
f8af4da3 | 505 | |
3cb4a0bb | 506 | #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS |
5037835c | 507 | #define MMF_DUMP_FILTER_BITS 9 |
3cb4a0bb KH |
508 | #define MMF_DUMP_FILTER_MASK \ |
509 | (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) | |
510 | #define MMF_DUMP_FILTER_DEFAULT \ | |
e575f111 | 511 | ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ |
656eb2cd RM |
512 | (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) |
513 | ||
514 | #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS | |
515 | # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) | |
516 | #else | |
517 | # define MMF_DUMP_MASK_DEFAULT_ELF 0 | |
518 | #endif | |
f8af4da3 HD |
519 | /* leave room for more dump flags */ |
520 | #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */ | |
ba76149f | 521 | #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */ |
bafb282d | 522 | #define MMF_EXE_FILE_CHANGED 18 /* see prctl_set_mm_exe_file() */ |
f8af4da3 | 523 | |
9f68f672 ON |
524 | #define MMF_HAS_UPROBES 19 /* has uprobes */ |
525 | #define MMF_RECALC_UPROBES 20 /* MMF_HAS_UPROBES can be wrong */ | |
bb8a4b7f | 526 | #define MMF_OOM_REAPED 21 /* mm has been already reaped */ |
11a410d5 | 527 | #define MMF_OOM_NOT_REAPABLE 22 /* mm couldn't be reaped */ |
f8ac4ec9 | 528 | |
f8af4da3 | 529 | #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) |
6c5d5238 | 530 | |
1da177e4 LT |
531 | struct sighand_struct { |
532 | atomic_t count; | |
533 | struct k_sigaction action[_NSIG]; | |
534 | spinlock_t siglock; | |
b8fceee1 | 535 | wait_queue_head_t signalfd_wqh; |
1da177e4 LT |
536 | }; |
537 | ||
0e464814 | 538 | struct pacct_struct { |
f6ec29a4 KK |
539 | int ac_flag; |
540 | long ac_exitcode; | |
0e464814 | 541 | unsigned long ac_mem; |
77787bfb KK |
542 | cputime_t ac_utime, ac_stime; |
543 | unsigned long ac_minflt, ac_majflt; | |
0e464814 KK |
544 | }; |
545 | ||
42c4ab41 SG |
546 | struct cpu_itimer { |
547 | cputime_t expires; | |
548 | cputime_t incr; | |
8356b5f9 SG |
549 | u32 error; |
550 | u32 incr_error; | |
42c4ab41 SG |
551 | }; |
552 | ||
d37f761d | 553 | /** |
9d7fb042 | 554 | * struct prev_cputime - snaphsot of system and user cputime |
d37f761d FW |
555 | * @utime: time spent in user mode |
556 | * @stime: time spent in system mode | |
9d7fb042 | 557 | * @lock: protects the above two fields |
d37f761d | 558 | * |
9d7fb042 PZ |
559 | * Stores previous user/system time values such that we can guarantee |
560 | * monotonicity. | |
d37f761d | 561 | */ |
9d7fb042 PZ |
562 | struct prev_cputime { |
563 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
d37f761d FW |
564 | cputime_t utime; |
565 | cputime_t stime; | |
9d7fb042 PZ |
566 | raw_spinlock_t lock; |
567 | #endif | |
d37f761d FW |
568 | }; |
569 | ||
9d7fb042 PZ |
570 | static inline void prev_cputime_init(struct prev_cputime *prev) |
571 | { | |
572 | #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE | |
573 | prev->utime = prev->stime = 0; | |
574 | raw_spin_lock_init(&prev->lock); | |
575 | #endif | |
576 | } | |
577 | ||
f06febc9 FM |
578 | /** |
579 | * struct task_cputime - collected CPU time counts | |
580 | * @utime: time spent in user mode, in &cputime_t units | |
581 | * @stime: time spent in kernel mode, in &cputime_t units | |
582 | * @sum_exec_runtime: total time spent on the CPU, in nanoseconds | |
5ce73a4a | 583 | * |
9d7fb042 PZ |
584 | * This structure groups together three kinds of CPU time that are tracked for |
585 | * threads and thread groups. Most things considering CPU time want to group | |
586 | * these counts together and treat all three of them in parallel. | |
f06febc9 FM |
587 | */ |
588 | struct task_cputime { | |
589 | cputime_t utime; | |
590 | cputime_t stime; | |
591 | unsigned long long sum_exec_runtime; | |
592 | }; | |
9d7fb042 | 593 | |
f06febc9 | 594 | /* Alternate field names when used to cache expirations. */ |
f06febc9 | 595 | #define virt_exp utime |
9d7fb042 | 596 | #define prof_exp stime |
f06febc9 FM |
597 | #define sched_exp sum_exec_runtime |
598 | ||
4cd4c1b4 PZ |
599 | #define INIT_CPUTIME \ |
600 | (struct task_cputime) { \ | |
64861634 MS |
601 | .utime = 0, \ |
602 | .stime = 0, \ | |
4cd4c1b4 PZ |
603 | .sum_exec_runtime = 0, \ |
604 | } | |
605 | ||
971e8a98 JL |
606 | /* |
607 | * This is the atomic variant of task_cputime, which can be used for | |
608 | * storing and updating task_cputime statistics without locking. | |
609 | */ | |
610 | struct task_cputime_atomic { | |
611 | atomic64_t utime; | |
612 | atomic64_t stime; | |
613 | atomic64_t sum_exec_runtime; | |
614 | }; | |
615 | ||
616 | #define INIT_CPUTIME_ATOMIC \ | |
617 | (struct task_cputime_atomic) { \ | |
618 | .utime = ATOMIC64_INIT(0), \ | |
619 | .stime = ATOMIC64_INIT(0), \ | |
620 | .sum_exec_runtime = ATOMIC64_INIT(0), \ | |
621 | } | |
622 | ||
609ca066 | 623 | #define PREEMPT_DISABLED (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED) |
a233f112 | 624 | |
c99e6efe | 625 | /* |
87dcbc06 PZ |
626 | * Disable preemption until the scheduler is running -- use an unconditional |
627 | * value so that it also works on !PREEMPT_COUNT kernels. | |
d86ee480 | 628 | * |
87dcbc06 | 629 | * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count(). |
c99e6efe | 630 | */ |
87dcbc06 | 631 | #define INIT_PREEMPT_COUNT PREEMPT_OFFSET |
a233f112 | 632 | |
c99e6efe | 633 | /* |
609ca066 PZ |
634 | * Initial preempt_count value; reflects the preempt_count schedule invariant |
635 | * which states that during context switches: | |
d86ee480 | 636 | * |
609ca066 PZ |
637 | * preempt_count() == 2*PREEMPT_DISABLE_OFFSET |
638 | * | |
639 | * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels. | |
640 | * Note: See finish_task_switch(). | |
c99e6efe | 641 | */ |
609ca066 | 642 | #define FORK_PREEMPT_COUNT (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED) |
c99e6efe | 643 | |
f06febc9 | 644 | /** |
4cd4c1b4 | 645 | * struct thread_group_cputimer - thread group interval timer counts |
920ce39f | 646 | * @cputime_atomic: atomic thread group interval timers. |
d5c373eb JL |
647 | * @running: true when there are timers running and |
648 | * @cputime_atomic receives updates. | |
c8d75aa4 JL |
649 | * @checking_timer: true when a thread in the group is in the |
650 | * process of checking for thread group timers. | |
f06febc9 FM |
651 | * |
652 | * This structure contains the version of task_cputime, above, that is | |
4cd4c1b4 | 653 | * used for thread group CPU timer calculations. |
f06febc9 | 654 | */ |
4cd4c1b4 | 655 | struct thread_group_cputimer { |
71107445 | 656 | struct task_cputime_atomic cputime_atomic; |
d5c373eb | 657 | bool running; |
c8d75aa4 | 658 | bool checking_timer; |
f06febc9 | 659 | }; |
f06febc9 | 660 | |
4714d1d3 | 661 | #include <linux/rwsem.h> |
5091faa4 MG |
662 | struct autogroup; |
663 | ||
1da177e4 | 664 | /* |
e815f0a8 | 665 | * NOTE! "signal_struct" does not have its own |
1da177e4 LT |
666 | * locking, because a shared signal_struct always |
667 | * implies a shared sighand_struct, so locking | |
668 | * sighand_struct is always a proper superset of | |
669 | * the locking of signal_struct. | |
670 | */ | |
671 | struct signal_struct { | |
ea6d290c | 672 | atomic_t sigcnt; |
1da177e4 | 673 | atomic_t live; |
b3ac022c | 674 | int nr_threads; |
f44666b0 | 675 | atomic_t oom_victims; /* # of TIF_MEDIE threads in this thread group */ |
0c740d0a | 676 | struct list_head thread_head; |
1da177e4 LT |
677 | |
678 | wait_queue_head_t wait_chldexit; /* for wait4() */ | |
679 | ||
680 | /* current thread group signal load-balancing target: */ | |
36c8b586 | 681 | struct task_struct *curr_target; |
1da177e4 LT |
682 | |
683 | /* shared signal handling: */ | |
684 | struct sigpending shared_pending; | |
685 | ||
686 | /* thread group exit support */ | |
687 | int group_exit_code; | |
688 | /* overloaded: | |
689 | * - notify group_exit_task when ->count is equal to notify_count | |
690 | * - everyone except group_exit_task is stopped during signal delivery | |
691 | * of fatal signals, group_exit_task processes the signal. | |
692 | */ | |
1da177e4 | 693 | int notify_count; |
07dd20e0 | 694 | struct task_struct *group_exit_task; |
1da177e4 LT |
695 | |
696 | /* thread group stop support, overloads group_exit_code too */ | |
697 | int group_stop_count; | |
698 | unsigned int flags; /* see SIGNAL_* flags below */ | |
699 | ||
ebec18a6 LP |
700 | /* |
701 | * PR_SET_CHILD_SUBREAPER marks a process, like a service | |
702 | * manager, to re-parent orphan (double-forking) child processes | |
703 | * to this process instead of 'init'. The service manager is | |
704 | * able to receive SIGCHLD signals and is able to investigate | |
705 | * the process until it calls wait(). All children of this | |
706 | * process will inherit a flag if they should look for a | |
707 | * child_subreaper process at exit. | |
708 | */ | |
709 | unsigned int is_child_subreaper:1; | |
710 | unsigned int has_child_subreaper:1; | |
711 | ||
1da177e4 | 712 | /* POSIX.1b Interval Timers */ |
5ed67f05 PE |
713 | int posix_timer_id; |
714 | struct list_head posix_timers; | |
1da177e4 LT |
715 | |
716 | /* ITIMER_REAL timer for the process */ | |
2ff678b8 | 717 | struct hrtimer real_timer; |
fea9d175 | 718 | struct pid *leader_pid; |
2ff678b8 | 719 | ktime_t it_real_incr; |
1da177e4 | 720 | |
42c4ab41 SG |
721 | /* |
722 | * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use | |
723 | * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these | |
724 | * values are defined to 0 and 1 respectively | |
725 | */ | |
726 | struct cpu_itimer it[2]; | |
1da177e4 | 727 | |
f06febc9 | 728 | /* |
4cd4c1b4 PZ |
729 | * Thread group totals for process CPU timers. |
730 | * See thread_group_cputimer(), et al, for details. | |
f06febc9 | 731 | */ |
4cd4c1b4 | 732 | struct thread_group_cputimer cputimer; |
f06febc9 FM |
733 | |
734 | /* Earliest-expiration cache. */ | |
735 | struct task_cputime cputime_expires; | |
736 | ||
d027d45d | 737 | #ifdef CONFIG_NO_HZ_FULL |
f009a7a7 | 738 | atomic_t tick_dep_mask; |
d027d45d FW |
739 | #endif |
740 | ||
f06febc9 FM |
741 | struct list_head cpu_timers[3]; |
742 | ||
ab521dc0 | 743 | struct pid *tty_old_pgrp; |
1ec320af | 744 | |
1da177e4 LT |
745 | /* boolean value for session group leader */ |
746 | int leader; | |
747 | ||
748 | struct tty_struct *tty; /* NULL if no tty */ | |
749 | ||
5091faa4 MG |
750 | #ifdef CONFIG_SCHED_AUTOGROUP |
751 | struct autogroup *autogroup; | |
752 | #endif | |
1da177e4 LT |
753 | /* |
754 | * Cumulative resource counters for dead threads in the group, | |
755 | * and for reaped dead child processes forked by this group. | |
756 | * Live threads maintain their own counters and add to these | |
757 | * in __exit_signal, except for the group leader. | |
758 | */ | |
e78c3496 | 759 | seqlock_t stats_lock; |
32bd671d | 760 | cputime_t utime, stime, cutime, cstime; |
9ac52315 LV |
761 | cputime_t gtime; |
762 | cputime_t cgtime; | |
9d7fb042 | 763 | struct prev_cputime prev_cputime; |
1da177e4 LT |
764 | unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; |
765 | unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; | |
6eaeeaba | 766 | unsigned long inblock, oublock, cinblock, coublock; |
1f10206c | 767 | unsigned long maxrss, cmaxrss; |
940389b8 | 768 | struct task_io_accounting ioac; |
1da177e4 | 769 | |
32bd671d PZ |
770 | /* |
771 | * Cumulative ns of schedule CPU time fo dead threads in the | |
772 | * group, not including a zombie group leader, (This only differs | |
773 | * from jiffies_to_ns(utime + stime) if sched_clock uses something | |
774 | * other than jiffies.) | |
775 | */ | |
776 | unsigned long long sum_sched_runtime; | |
777 | ||
1da177e4 LT |
778 | /* |
779 | * We don't bother to synchronize most readers of this at all, | |
780 | * because there is no reader checking a limit that actually needs | |
781 | * to get both rlim_cur and rlim_max atomically, and either one | |
782 | * alone is a single word that can safely be read normally. | |
783 | * getrlimit/setrlimit use task_lock(current->group_leader) to | |
784 | * protect this instead of the siglock, because they really | |
785 | * have no need to disable irqs. | |
786 | */ | |
787 | struct rlimit rlim[RLIM_NLIMITS]; | |
788 | ||
0e464814 KK |
789 | #ifdef CONFIG_BSD_PROCESS_ACCT |
790 | struct pacct_struct pacct; /* per-process accounting information */ | |
791 | #endif | |
ad4ecbcb | 792 | #ifdef CONFIG_TASKSTATS |
ad4ecbcb SN |
793 | struct taskstats *stats; |
794 | #endif | |
522ed776 MT |
795 | #ifdef CONFIG_AUDIT |
796 | unsigned audit_tty; | |
797 | struct tty_audit_buf *tty_audit_buf; | |
798 | #endif | |
28b83c51 | 799 | |
c96fc2d8 TH |
800 | /* |
801 | * Thread is the potential origin of an oom condition; kill first on | |
802 | * oom | |
803 | */ | |
804 | bool oom_flag_origin; | |
a9c58b90 DR |
805 | short oom_score_adj; /* OOM kill score adjustment */ |
806 | short oom_score_adj_min; /* OOM kill score adjustment min value. | |
807 | * Only settable by CAP_SYS_RESOURCE. */ | |
9b1bf12d KM |
808 | |
809 | struct mutex cred_guard_mutex; /* guard against foreign influences on | |
810 | * credential calculations | |
811 | * (notably. ptrace) */ | |
1da177e4 LT |
812 | }; |
813 | ||
814 | /* | |
815 | * Bits in flags field of signal_struct. | |
816 | */ | |
817 | #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ | |
ee77f075 ON |
818 | #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */ |
819 | #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */ | |
403bad72 | 820 | #define SIGNAL_GROUP_COREDUMP 0x00000008 /* coredump in progress */ |
e4420551 ON |
821 | /* |
822 | * Pending notifications to parent. | |
823 | */ | |
824 | #define SIGNAL_CLD_STOPPED 0x00000010 | |
825 | #define SIGNAL_CLD_CONTINUED 0x00000020 | |
826 | #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) | |
1da177e4 | 827 | |
fae5fa44 ON |
828 | #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ |
829 | ||
ed5d2cac ON |
830 | /* If true, all threads except ->group_exit_task have pending SIGKILL */ |
831 | static inline int signal_group_exit(const struct signal_struct *sig) | |
832 | { | |
833 | return (sig->flags & SIGNAL_GROUP_EXIT) || | |
834 | (sig->group_exit_task != NULL); | |
835 | } | |
836 | ||
1da177e4 LT |
837 | /* |
838 | * Some day this will be a full-fledged user tracking system.. | |
839 | */ | |
840 | struct user_struct { | |
841 | atomic_t __count; /* reference count */ | |
842 | atomic_t processes; /* How many processes does this user have? */ | |
1da177e4 | 843 | atomic_t sigpending; /* How many pending signals does this user have? */ |
2d9048e2 | 844 | #ifdef CONFIG_INOTIFY_USER |
0eeca283 RL |
845 | atomic_t inotify_watches; /* How many inotify watches does this user have? */ |
846 | atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ | |
847 | #endif | |
4afeff85 EP |
848 | #ifdef CONFIG_FANOTIFY |
849 | atomic_t fanotify_listeners; | |
850 | #endif | |
7ef9964e | 851 | #ifdef CONFIG_EPOLL |
52bd19f7 | 852 | atomic_long_t epoll_watches; /* The number of file descriptors currently watched */ |
7ef9964e | 853 | #endif |
970a8645 | 854 | #ifdef CONFIG_POSIX_MQUEUE |
1da177e4 LT |
855 | /* protected by mq_lock */ |
856 | unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ | |
970a8645 | 857 | #endif |
1da177e4 | 858 | unsigned long locked_shm; /* How many pages of mlocked shm ? */ |
712f4aad | 859 | unsigned long unix_inflight; /* How many files in flight in unix sockets */ |
759c0114 | 860 | atomic_long_t pipe_bufs; /* how many pages are allocated in pipe buffers */ |
1da177e4 LT |
861 | |
862 | #ifdef CONFIG_KEYS | |
863 | struct key *uid_keyring; /* UID specific keyring */ | |
864 | struct key *session_keyring; /* UID's default session keyring */ | |
865 | #endif | |
866 | ||
867 | /* Hash table maintenance information */ | |
735de223 | 868 | struct hlist_node uidhash_node; |
7b44ab97 | 869 | kuid_t uid; |
24e377a8 | 870 | |
aaac3ba9 | 871 | #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL) |
789f90fc PZ |
872 | atomic_long_t locked_vm; |
873 | #endif | |
1da177e4 LT |
874 | }; |
875 | ||
eb41d946 | 876 | extern int uids_sysfs_init(void); |
5cb350ba | 877 | |
7b44ab97 | 878 | extern struct user_struct *find_user(kuid_t); |
1da177e4 LT |
879 | |
880 | extern struct user_struct root_user; | |
881 | #define INIT_USER (&root_user) | |
882 | ||
b6dff3ec | 883 | |
1da177e4 LT |
884 | struct backing_dev_info; |
885 | struct reclaim_state; | |
886 | ||
f6db8347 | 887 | #ifdef CONFIG_SCHED_INFO |
1da177e4 LT |
888 | struct sched_info { |
889 | /* cumulative counters */ | |
2d72376b | 890 | unsigned long pcount; /* # of times run on this cpu */ |
9c2c4802 | 891 | unsigned long long run_delay; /* time spent waiting on a runqueue */ |
1da177e4 LT |
892 | |
893 | /* timestamps */ | |
172ba844 BS |
894 | unsigned long long last_arrival,/* when we last ran on a cpu */ |
895 | last_queued; /* when we were last queued to run */ | |
1da177e4 | 896 | }; |
f6db8347 | 897 | #endif /* CONFIG_SCHED_INFO */ |
1da177e4 | 898 | |
ca74e92b SN |
899 | #ifdef CONFIG_TASK_DELAY_ACCT |
900 | struct task_delay_info { | |
901 | spinlock_t lock; | |
902 | unsigned int flags; /* Private per-task flags */ | |
903 | ||
904 | /* For each stat XXX, add following, aligned appropriately | |
905 | * | |
906 | * struct timespec XXX_start, XXX_end; | |
907 | * u64 XXX_delay; | |
908 | * u32 XXX_count; | |
909 | * | |
910 | * Atomicity of updates to XXX_delay, XXX_count protected by | |
911 | * single lock above (split into XXX_lock if contention is an issue). | |
912 | */ | |
0ff92245 SN |
913 | |
914 | /* | |
915 | * XXX_count is incremented on every XXX operation, the delay | |
916 | * associated with the operation is added to XXX_delay. | |
917 | * XXX_delay contains the accumulated delay time in nanoseconds. | |
918 | */ | |
9667a23d | 919 | u64 blkio_start; /* Shared by blkio, swapin */ |
0ff92245 SN |
920 | u64 blkio_delay; /* wait for sync block io completion */ |
921 | u64 swapin_delay; /* wait for swapin block io completion */ | |
922 | u32 blkio_count; /* total count of the number of sync block */ | |
923 | /* io operations performed */ | |
924 | u32 swapin_count; /* total count of the number of swapin block */ | |
925 | /* io operations performed */ | |
873b4771 | 926 | |
9667a23d | 927 | u64 freepages_start; |
873b4771 KK |
928 | u64 freepages_delay; /* wait for memory reclaim */ |
929 | u32 freepages_count; /* total count of memory reclaim */ | |
ca74e92b | 930 | }; |
52f17b6c CS |
931 | #endif /* CONFIG_TASK_DELAY_ACCT */ |
932 | ||
933 | static inline int sched_info_on(void) | |
934 | { | |
935 | #ifdef CONFIG_SCHEDSTATS | |
936 | return 1; | |
937 | #elif defined(CONFIG_TASK_DELAY_ACCT) | |
938 | extern int delayacct_on; | |
939 | return delayacct_on; | |
940 | #else | |
941 | return 0; | |
ca74e92b | 942 | #endif |
52f17b6c | 943 | } |
ca74e92b | 944 | |
cb251765 MG |
945 | #ifdef CONFIG_SCHEDSTATS |
946 | void force_schedstat_enabled(void); | |
947 | #endif | |
948 | ||
d15bcfdb IM |
949 | enum cpu_idle_type { |
950 | CPU_IDLE, | |
951 | CPU_NOT_IDLE, | |
952 | CPU_NEWLY_IDLE, | |
953 | CPU_MAX_IDLE_TYPES | |
1da177e4 LT |
954 | }; |
955 | ||
6ecdd749 YD |
956 | /* |
957 | * Integer metrics need fixed point arithmetic, e.g., sched/fair | |
958 | * has a few: load, load_avg, util_avg, freq, and capacity. | |
959 | * | |
960 | * We define a basic fixed point arithmetic range, and then formalize | |
961 | * all these metrics based on that basic range. | |
962 | */ | |
963 | # define SCHED_FIXEDPOINT_SHIFT 10 | |
964 | # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT) | |
965 | ||
1399fa78 | 966 | /* |
ca8ce3d0 | 967 | * Increase resolution of cpu_capacity calculations |
1399fa78 | 968 | */ |
6ecdd749 | 969 | #define SCHED_CAPACITY_SHIFT SCHED_FIXEDPOINT_SHIFT |
ca8ce3d0 | 970 | #define SCHED_CAPACITY_SCALE (1L << SCHED_CAPACITY_SHIFT) |
1da177e4 | 971 | |
76751049 PZ |
972 | /* |
973 | * Wake-queues are lists of tasks with a pending wakeup, whose | |
974 | * callers have already marked the task as woken internally, | |
975 | * and can thus carry on. A common use case is being able to | |
976 | * do the wakeups once the corresponding user lock as been | |
977 | * released. | |
978 | * | |
979 | * We hold reference to each task in the list across the wakeup, | |
980 | * thus guaranteeing that the memory is still valid by the time | |
981 | * the actual wakeups are performed in wake_up_q(). | |
982 | * | |
983 | * One per task suffices, because there's never a need for a task to be | |
984 | * in two wake queues simultaneously; it is forbidden to abandon a task | |
985 | * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is | |
986 | * already in a wake queue, the wakeup will happen soon and the second | |
987 | * waker can just skip it. | |
988 | * | |
989 | * The WAKE_Q macro declares and initializes the list head. | |
990 | * wake_up_q() does NOT reinitialize the list; it's expected to be | |
991 | * called near the end of a function, where the fact that the queue is | |
992 | * not used again will be easy to see by inspection. | |
993 | * | |
994 | * Note that this can cause spurious wakeups. schedule() callers | |
995 | * must ensure the call is done inside a loop, confirming that the | |
996 | * wakeup condition has in fact occurred. | |
997 | */ | |
998 | struct wake_q_node { | |
999 | struct wake_q_node *next; | |
1000 | }; | |
1001 | ||
1002 | struct wake_q_head { | |
1003 | struct wake_q_node *first; | |
1004 | struct wake_q_node **lastp; | |
1005 | }; | |
1006 | ||
1007 | #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01) | |
1008 | ||
1009 | #define WAKE_Q(name) \ | |
1010 | struct wake_q_head name = { WAKE_Q_TAIL, &name.first } | |
1011 | ||
1012 | extern void wake_q_add(struct wake_q_head *head, | |
1013 | struct task_struct *task); | |
1014 | extern void wake_up_q(struct wake_q_head *head); | |
1015 | ||
1399fa78 NR |
1016 | /* |
1017 | * sched-domains (multiprocessor balancing) declarations: | |
1018 | */ | |
2dd73a4f | 1019 | #ifdef CONFIG_SMP |
b5d978e0 PZ |
1020 | #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */ |
1021 | #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */ | |
1022 | #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */ | |
1023 | #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */ | |
c88d5910 | 1024 | #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */ |
b5d978e0 | 1025 | #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */ |
5d4dfddd | 1026 | #define SD_SHARE_CPUCAPACITY 0x0080 /* Domain members share cpu power */ |
d77b3ed5 | 1027 | #define SD_SHARE_POWERDOMAIN 0x0100 /* Domain members share power domain */ |
b5d978e0 PZ |
1028 | #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */ |
1029 | #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ | |
532cb4c4 | 1030 | #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */ |
b5d978e0 | 1031 | #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ |
e3589f6c | 1032 | #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */ |
3a7053b3 | 1033 | #define SD_NUMA 0x4000 /* cross-node balancing */ |
5c45bf27 | 1034 | |
143e1e28 | 1035 | #ifdef CONFIG_SCHED_SMT |
b6220ad6 | 1036 | static inline int cpu_smt_flags(void) |
143e1e28 | 1037 | { |
5d4dfddd | 1038 | return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES; |
143e1e28 VG |
1039 | } |
1040 | #endif | |
1041 | ||
1042 | #ifdef CONFIG_SCHED_MC | |
b6220ad6 | 1043 | static inline int cpu_core_flags(void) |
143e1e28 VG |
1044 | { |
1045 | return SD_SHARE_PKG_RESOURCES; | |
1046 | } | |
1047 | #endif | |
1048 | ||
1049 | #ifdef CONFIG_NUMA | |
b6220ad6 | 1050 | static inline int cpu_numa_flags(void) |
143e1e28 VG |
1051 | { |
1052 | return SD_NUMA; | |
1053 | } | |
1054 | #endif | |
532cb4c4 | 1055 | |
1d3504fc HS |
1056 | struct sched_domain_attr { |
1057 | int relax_domain_level; | |
1058 | }; | |
1059 | ||
1060 | #define SD_ATTR_INIT (struct sched_domain_attr) { \ | |
1061 | .relax_domain_level = -1, \ | |
1062 | } | |
1063 | ||
60495e77 PZ |
1064 | extern int sched_domain_level_max; |
1065 | ||
5e6521ea LZ |
1066 | struct sched_group; |
1067 | ||
1da177e4 LT |
1068 | struct sched_domain { |
1069 | /* These fields must be setup */ | |
1070 | struct sched_domain *parent; /* top domain must be null terminated */ | |
1a848870 | 1071 | struct sched_domain *child; /* bottom domain must be null terminated */ |
1da177e4 | 1072 | struct sched_group *groups; /* the balancing groups of the domain */ |
1da177e4 LT |
1073 | unsigned long min_interval; /* Minimum balance interval ms */ |
1074 | unsigned long max_interval; /* Maximum balance interval ms */ | |
1075 | unsigned int busy_factor; /* less balancing by factor if busy */ | |
1076 | unsigned int imbalance_pct; /* No balance until over watermark */ | |
1da177e4 | 1077 | unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ |
7897986b NP |
1078 | unsigned int busy_idx; |
1079 | unsigned int idle_idx; | |
1080 | unsigned int newidle_idx; | |
1081 | unsigned int wake_idx; | |
147cbb4b | 1082 | unsigned int forkexec_idx; |
a52bfd73 | 1083 | unsigned int smt_gain; |
25f55d9d VG |
1084 | |
1085 | int nohz_idle; /* NOHZ IDLE status */ | |
1da177e4 | 1086 | int flags; /* See SD_* */ |
60495e77 | 1087 | int level; |
1da177e4 LT |
1088 | |
1089 | /* Runtime fields. */ | |
1090 | unsigned long last_balance; /* init to jiffies. units in jiffies */ | |
1091 | unsigned int balance_interval; /* initialise to 1. units in ms. */ | |
1092 | unsigned int nr_balance_failed; /* initialise to 0 */ | |
1093 | ||
f48627e6 | 1094 | /* idle_balance() stats */ |
9bd721c5 | 1095 | u64 max_newidle_lb_cost; |
f48627e6 | 1096 | unsigned long next_decay_max_lb_cost; |
2398f2c6 | 1097 | |
1da177e4 LT |
1098 | #ifdef CONFIG_SCHEDSTATS |
1099 | /* load_balance() stats */ | |
480b9434 KC |
1100 | unsigned int lb_count[CPU_MAX_IDLE_TYPES]; |
1101 | unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; | |
1102 | unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; | |
1103 | unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; | |
1104 | unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; | |
1105 | unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; | |
1106 | unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; | |
1107 | unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; | |
1da177e4 LT |
1108 | |
1109 | /* Active load balancing */ | |
480b9434 KC |
1110 | unsigned int alb_count; |
1111 | unsigned int alb_failed; | |
1112 | unsigned int alb_pushed; | |
1da177e4 | 1113 | |
68767a0a | 1114 | /* SD_BALANCE_EXEC stats */ |
480b9434 KC |
1115 | unsigned int sbe_count; |
1116 | unsigned int sbe_balanced; | |
1117 | unsigned int sbe_pushed; | |
1da177e4 | 1118 | |
68767a0a | 1119 | /* SD_BALANCE_FORK stats */ |
480b9434 KC |
1120 | unsigned int sbf_count; |
1121 | unsigned int sbf_balanced; | |
1122 | unsigned int sbf_pushed; | |
68767a0a | 1123 | |
1da177e4 | 1124 | /* try_to_wake_up() stats */ |
480b9434 KC |
1125 | unsigned int ttwu_wake_remote; |
1126 | unsigned int ttwu_move_affine; | |
1127 | unsigned int ttwu_move_balance; | |
1da177e4 | 1128 | #endif |
a5d8c348 IM |
1129 | #ifdef CONFIG_SCHED_DEBUG |
1130 | char *name; | |
1131 | #endif | |
dce840a0 PZ |
1132 | union { |
1133 | void *private; /* used during construction */ | |
1134 | struct rcu_head rcu; /* used during destruction */ | |
1135 | }; | |
6c99e9ad | 1136 | |
669c55e9 | 1137 | unsigned int span_weight; |
4200efd9 IM |
1138 | /* |
1139 | * Span of all CPUs in this domain. | |
1140 | * | |
1141 | * NOTE: this field is variable length. (Allocated dynamically | |
1142 | * by attaching extra space to the end of the structure, | |
1143 | * depending on how many CPUs the kernel has booted up with) | |
4200efd9 IM |
1144 | */ |
1145 | unsigned long span[0]; | |
1da177e4 LT |
1146 | }; |
1147 | ||
758b2cdc RR |
1148 | static inline struct cpumask *sched_domain_span(struct sched_domain *sd) |
1149 | { | |
6c99e9ad | 1150 | return to_cpumask(sd->span); |
758b2cdc RR |
1151 | } |
1152 | ||
acc3f5d7 | 1153 | extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], |
1d3504fc | 1154 | struct sched_domain_attr *dattr_new); |
029190c5 | 1155 | |
acc3f5d7 RR |
1156 | /* Allocate an array of sched domains, for partition_sched_domains(). */ |
1157 | cpumask_var_t *alloc_sched_domains(unsigned int ndoms); | |
1158 | void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); | |
1159 | ||
39be3501 PZ |
1160 | bool cpus_share_cache(int this_cpu, int that_cpu); |
1161 | ||
143e1e28 | 1162 | typedef const struct cpumask *(*sched_domain_mask_f)(int cpu); |
b6220ad6 | 1163 | typedef int (*sched_domain_flags_f)(void); |
143e1e28 VG |
1164 | |
1165 | #define SDTL_OVERLAP 0x01 | |
1166 | ||
1167 | struct sd_data { | |
1168 | struct sched_domain **__percpu sd; | |
1169 | struct sched_group **__percpu sg; | |
63b2ca30 | 1170 | struct sched_group_capacity **__percpu sgc; |
143e1e28 VG |
1171 | }; |
1172 | ||
1173 | struct sched_domain_topology_level { | |
1174 | sched_domain_mask_f mask; | |
1175 | sched_domain_flags_f sd_flags; | |
1176 | int flags; | |
1177 | int numa_level; | |
1178 | struct sd_data data; | |
1179 | #ifdef CONFIG_SCHED_DEBUG | |
1180 | char *name; | |
1181 | #endif | |
1182 | }; | |
1183 | ||
143e1e28 | 1184 | extern void set_sched_topology(struct sched_domain_topology_level *tl); |
f6be8af1 | 1185 | extern void wake_up_if_idle(int cpu); |
143e1e28 VG |
1186 | |
1187 | #ifdef CONFIG_SCHED_DEBUG | |
1188 | # define SD_INIT_NAME(type) .name = #type | |
1189 | #else | |
1190 | # define SD_INIT_NAME(type) | |
1191 | #endif | |
1192 | ||
1b427c15 | 1193 | #else /* CONFIG_SMP */ |
1da177e4 | 1194 | |
1b427c15 | 1195 | struct sched_domain_attr; |
d02c7a8c | 1196 | |
1b427c15 | 1197 | static inline void |
acc3f5d7 | 1198 | partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], |
1b427c15 IM |
1199 | struct sched_domain_attr *dattr_new) |
1200 | { | |
d02c7a8c | 1201 | } |
39be3501 PZ |
1202 | |
1203 | static inline bool cpus_share_cache(int this_cpu, int that_cpu) | |
1204 | { | |
1205 | return true; | |
1206 | } | |
1207 | ||
1b427c15 | 1208 | #endif /* !CONFIG_SMP */ |
1da177e4 | 1209 | |
47fe38fc | 1210 | |
1da177e4 | 1211 | struct io_context; /* See blkdev.h */ |
1da177e4 | 1212 | |
1da177e4 | 1213 | |
383f2835 | 1214 | #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK |
36c8b586 | 1215 | extern void prefetch_stack(struct task_struct *t); |
383f2835 CK |
1216 | #else |
1217 | static inline void prefetch_stack(struct task_struct *t) { } | |
1218 | #endif | |
1da177e4 LT |
1219 | |
1220 | struct audit_context; /* See audit.c */ | |
1221 | struct mempolicy; | |
b92ce558 | 1222 | struct pipe_inode_info; |
4865ecf1 | 1223 | struct uts_namespace; |
1da177e4 | 1224 | |
20b8a59f | 1225 | struct load_weight { |
9dbdb155 PZ |
1226 | unsigned long weight; |
1227 | u32 inv_weight; | |
20b8a59f IM |
1228 | }; |
1229 | ||
9d89c257 | 1230 | /* |
7b595334 YD |
1231 | * The load_avg/util_avg accumulates an infinite geometric series |
1232 | * (see __update_load_avg() in kernel/sched/fair.c). | |
1233 | * | |
1234 | * [load_avg definition] | |
1235 | * | |
1236 | * load_avg = runnable% * scale_load_down(load) | |
1237 | * | |
1238 | * where runnable% is the time ratio that a sched_entity is runnable. | |
1239 | * For cfs_rq, it is the aggregated load_avg of all runnable and | |
9d89c257 | 1240 | * blocked sched_entities. |
7b595334 YD |
1241 | * |
1242 | * load_avg may also take frequency scaling into account: | |
1243 | * | |
1244 | * load_avg = runnable% * scale_load_down(load) * freq% | |
1245 | * | |
1246 | * where freq% is the CPU frequency normalized to the highest frequency. | |
1247 | * | |
1248 | * [util_avg definition] | |
1249 | * | |
1250 | * util_avg = running% * SCHED_CAPACITY_SCALE | |
1251 | * | |
1252 | * where running% is the time ratio that a sched_entity is running on | |
1253 | * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable | |
1254 | * and blocked sched_entities. | |
1255 | * | |
1256 | * util_avg may also factor frequency scaling and CPU capacity scaling: | |
1257 | * | |
1258 | * util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity% | |
1259 | * | |
1260 | * where freq% is the same as above, and capacity% is the CPU capacity | |
1261 | * normalized to the greatest capacity (due to uarch differences, etc). | |
1262 | * | |
1263 | * N.B., the above ratios (runnable%, running%, freq%, and capacity%) | |
1264 | * themselves are in the range of [0, 1]. To do fixed point arithmetics, | |
1265 | * we therefore scale them to as large a range as necessary. This is for | |
1266 | * example reflected by util_avg's SCHED_CAPACITY_SCALE. | |
1267 | * | |
1268 | * [Overflow issue] | |
1269 | * | |
1270 | * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities | |
1271 | * with the highest load (=88761), always runnable on a single cfs_rq, | |
1272 | * and should not overflow as the number already hits PID_MAX_LIMIT. | |
1273 | * | |
1274 | * For all other cases (including 32-bit kernels), struct load_weight's | |
1275 | * weight will overflow first before we do, because: | |
1276 | * | |
1277 | * Max(load_avg) <= Max(load.weight) | |
1278 | * | |
1279 | * Then it is the load_weight's responsibility to consider overflow | |
1280 | * issues. | |
9d89c257 | 1281 | */ |
9d85f21c | 1282 | struct sched_avg { |
9d89c257 YD |
1283 | u64 last_update_time, load_sum; |
1284 | u32 util_sum, period_contrib; | |
1285 | unsigned long load_avg, util_avg; | |
9d85f21c PT |
1286 | }; |
1287 | ||
94c18227 | 1288 | #ifdef CONFIG_SCHEDSTATS |
41acab88 | 1289 | struct sched_statistics { |
20b8a59f | 1290 | u64 wait_start; |
94c18227 | 1291 | u64 wait_max; |
6d082592 AV |
1292 | u64 wait_count; |
1293 | u64 wait_sum; | |
8f0dfc34 AV |
1294 | u64 iowait_count; |
1295 | u64 iowait_sum; | |
94c18227 | 1296 | |
20b8a59f | 1297 | u64 sleep_start; |
20b8a59f | 1298 | u64 sleep_max; |
94c18227 IM |
1299 | s64 sum_sleep_runtime; |
1300 | ||
1301 | u64 block_start; | |
20b8a59f IM |
1302 | u64 block_max; |
1303 | u64 exec_max; | |
eba1ed4b | 1304 | u64 slice_max; |
cc367732 | 1305 | |
cc367732 IM |
1306 | u64 nr_migrations_cold; |
1307 | u64 nr_failed_migrations_affine; | |
1308 | u64 nr_failed_migrations_running; | |
1309 | u64 nr_failed_migrations_hot; | |
1310 | u64 nr_forced_migrations; | |
cc367732 IM |
1311 | |
1312 | u64 nr_wakeups; | |
1313 | u64 nr_wakeups_sync; | |
1314 | u64 nr_wakeups_migrate; | |
1315 | u64 nr_wakeups_local; | |
1316 | u64 nr_wakeups_remote; | |
1317 | u64 nr_wakeups_affine; | |
1318 | u64 nr_wakeups_affine_attempts; | |
1319 | u64 nr_wakeups_passive; | |
1320 | u64 nr_wakeups_idle; | |
41acab88 LDM |
1321 | }; |
1322 | #endif | |
1323 | ||
1324 | struct sched_entity { | |
1325 | struct load_weight load; /* for load-balancing */ | |
1326 | struct rb_node run_node; | |
1327 | struct list_head group_node; | |
1328 | unsigned int on_rq; | |
1329 | ||
1330 | u64 exec_start; | |
1331 | u64 sum_exec_runtime; | |
1332 | u64 vruntime; | |
1333 | u64 prev_sum_exec_runtime; | |
1334 | ||
41acab88 LDM |
1335 | u64 nr_migrations; |
1336 | ||
41acab88 LDM |
1337 | #ifdef CONFIG_SCHEDSTATS |
1338 | struct sched_statistics statistics; | |
94c18227 IM |
1339 | #endif |
1340 | ||
20b8a59f | 1341 | #ifdef CONFIG_FAIR_GROUP_SCHED |
fed14d45 | 1342 | int depth; |
20b8a59f IM |
1343 | struct sched_entity *parent; |
1344 | /* rq on which this entity is (to be) queued: */ | |
1345 | struct cfs_rq *cfs_rq; | |
1346 | /* rq "owned" by this entity/group: */ | |
1347 | struct cfs_rq *my_q; | |
1348 | #endif | |
8bd75c77 | 1349 | |
141965c7 | 1350 | #ifdef CONFIG_SMP |
5a107804 JO |
1351 | /* |
1352 | * Per entity load average tracking. | |
1353 | * | |
1354 | * Put into separate cache line so it does not | |
1355 | * collide with read-mostly values above. | |
1356 | */ | |
1357 | struct sched_avg avg ____cacheline_aligned_in_smp; | |
9d85f21c | 1358 | #endif |
20b8a59f | 1359 | }; |
70b97a7f | 1360 | |
fa717060 PZ |
1361 | struct sched_rt_entity { |
1362 | struct list_head run_list; | |
78f2c7db | 1363 | unsigned long timeout; |
57d2aa00 | 1364 | unsigned long watchdog_stamp; |
bee367ed | 1365 | unsigned int time_slice; |
ff77e468 PZ |
1366 | unsigned short on_rq; |
1367 | unsigned short on_list; | |
6f505b16 | 1368 | |
58d6c2d7 | 1369 | struct sched_rt_entity *back; |
052f1dc7 | 1370 | #ifdef CONFIG_RT_GROUP_SCHED |
6f505b16 PZ |
1371 | struct sched_rt_entity *parent; |
1372 | /* rq on which this entity is (to be) queued: */ | |
1373 | struct rt_rq *rt_rq; | |
1374 | /* rq "owned" by this entity/group: */ | |
1375 | struct rt_rq *my_q; | |
1376 | #endif | |
fa717060 PZ |
1377 | }; |
1378 | ||
aab03e05 DF |
1379 | struct sched_dl_entity { |
1380 | struct rb_node rb_node; | |
1381 | ||
1382 | /* | |
1383 | * Original scheduling parameters. Copied here from sched_attr | |
4027d080 | 1384 | * during sched_setattr(), they will remain the same until |
1385 | * the next sched_setattr(). | |
aab03e05 DF |
1386 | */ |
1387 | u64 dl_runtime; /* maximum runtime for each instance */ | |
1388 | u64 dl_deadline; /* relative deadline of each instance */ | |
755378a4 | 1389 | u64 dl_period; /* separation of two instances (period) */ |
332ac17e | 1390 | u64 dl_bw; /* dl_runtime / dl_deadline */ |
aab03e05 DF |
1391 | |
1392 | /* | |
1393 | * Actual scheduling parameters. Initialized with the values above, | |
1394 | * they are continously updated during task execution. Note that | |
1395 | * the remaining runtime could be < 0 in case we are in overrun. | |
1396 | */ | |
1397 | s64 runtime; /* remaining runtime for this instance */ | |
1398 | u64 deadline; /* absolute deadline for this instance */ | |
1399 | unsigned int flags; /* specifying the scheduler behaviour */ | |
1400 | ||
1401 | /* | |
1402 | * Some bool flags: | |
1403 | * | |
1404 | * @dl_throttled tells if we exhausted the runtime. If so, the | |
1405 | * task has to wait for a replenishment to be performed at the | |
1406 | * next firing of dl_timer. | |
1407 | * | |
2d3d891d DF |
1408 | * @dl_boosted tells if we are boosted due to DI. If so we are |
1409 | * outside bandwidth enforcement mechanism (but only until we | |
5bfd126e JL |
1410 | * exit the critical section); |
1411 | * | |
1412 | * @dl_yielded tells if task gave up the cpu before consuming | |
1413 | * all its available runtime during the last job. | |
aab03e05 | 1414 | */ |
72f9f3fd | 1415 | int dl_throttled, dl_boosted, dl_yielded; |
aab03e05 DF |
1416 | |
1417 | /* | |
1418 | * Bandwidth enforcement timer. Each -deadline task has its | |
1419 | * own bandwidth to be enforced, thus we need one timer per task. | |
1420 | */ | |
1421 | struct hrtimer dl_timer; | |
1422 | }; | |
8bd75c77 | 1423 | |
1d082fd0 PM |
1424 | union rcu_special { |
1425 | struct { | |
8203d6d0 PM |
1426 | u8 blocked; |
1427 | u8 need_qs; | |
1428 | u8 exp_need_qs; | |
1429 | u8 pad; /* Otherwise the compiler can store garbage here. */ | |
1430 | } b; /* Bits. */ | |
1431 | u32 s; /* Set of bits. */ | |
1d082fd0 | 1432 | }; |
86848966 PM |
1433 | struct rcu_node; |
1434 | ||
8dc85d54 PZ |
1435 | enum perf_event_task_context { |
1436 | perf_invalid_context = -1, | |
1437 | perf_hw_context = 0, | |
89a1e187 | 1438 | perf_sw_context, |
8dc85d54 PZ |
1439 | perf_nr_task_contexts, |
1440 | }; | |
1441 | ||
72b252ae MG |
1442 | /* Track pages that require TLB flushes */ |
1443 | struct tlbflush_unmap_batch { | |
1444 | /* | |
1445 | * Each bit set is a CPU that potentially has a TLB entry for one of | |
1446 | * the PFNs being flushed. See set_tlb_ubc_flush_pending(). | |
1447 | */ | |
1448 | struct cpumask cpumask; | |
1449 | ||
1450 | /* True if any bit in cpumask is set */ | |
1451 | bool flush_required; | |
d950c947 MG |
1452 | |
1453 | /* | |
1454 | * If true then the PTE was dirty when unmapped. The entry must be | |
1455 | * flushed before IO is initiated or a stale TLB entry potentially | |
1456 | * allows an update without redirtying the page. | |
1457 | */ | |
1458 | bool writable; | |
72b252ae MG |
1459 | }; |
1460 | ||
1da177e4 LT |
1461 | struct task_struct { |
1462 | volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ | |
f7e4217b | 1463 | void *stack; |
1da177e4 | 1464 | atomic_t usage; |
97dc32cd WC |
1465 | unsigned int flags; /* per process flags, defined below */ |
1466 | unsigned int ptrace; | |
1da177e4 | 1467 | |
2dd73a4f | 1468 | #ifdef CONFIG_SMP |
fa14ff4a | 1469 | struct llist_node wake_entry; |
3ca7a440 | 1470 | int on_cpu; |
63b0e9ed | 1471 | unsigned int wakee_flips; |
62470419 | 1472 | unsigned long wakee_flip_decay_ts; |
63b0e9ed | 1473 | struct task_struct *last_wakee; |
ac66f547 PZ |
1474 | |
1475 | int wake_cpu; | |
2dd73a4f | 1476 | #endif |
fd2f4419 | 1477 | int on_rq; |
50e645a8 | 1478 | |
b29739f9 | 1479 | int prio, static_prio, normal_prio; |
c7aceaba | 1480 | unsigned int rt_priority; |
5522d5d5 | 1481 | const struct sched_class *sched_class; |
20b8a59f | 1482 | struct sched_entity se; |
fa717060 | 1483 | struct sched_rt_entity rt; |
8323f26c PZ |
1484 | #ifdef CONFIG_CGROUP_SCHED |
1485 | struct task_group *sched_task_group; | |
1486 | #endif | |
aab03e05 | 1487 | struct sched_dl_entity dl; |
1da177e4 | 1488 | |
e107be36 AK |
1489 | #ifdef CONFIG_PREEMPT_NOTIFIERS |
1490 | /* list of struct preempt_notifier: */ | |
1491 | struct hlist_head preempt_notifiers; | |
1492 | #endif | |
1493 | ||
6c5c9341 | 1494 | #ifdef CONFIG_BLK_DEV_IO_TRACE |
2056a782 | 1495 | unsigned int btrace_seq; |
6c5c9341 | 1496 | #endif |
1da177e4 | 1497 | |
97dc32cd | 1498 | unsigned int policy; |
29baa747 | 1499 | int nr_cpus_allowed; |
1da177e4 | 1500 | cpumask_t cpus_allowed; |
1da177e4 | 1501 | |
a57eb940 | 1502 | #ifdef CONFIG_PREEMPT_RCU |
e260be67 | 1503 | int rcu_read_lock_nesting; |
1d082fd0 | 1504 | union rcu_special rcu_read_unlock_special; |
f41d911f | 1505 | struct list_head rcu_node_entry; |
a57eb940 | 1506 | struct rcu_node *rcu_blocked_node; |
28f6569a | 1507 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
8315f422 PM |
1508 | #ifdef CONFIG_TASKS_RCU |
1509 | unsigned long rcu_tasks_nvcsw; | |
1510 | bool rcu_tasks_holdout; | |
1511 | struct list_head rcu_tasks_holdout_list; | |
176f8f7a | 1512 | int rcu_tasks_idle_cpu; |
8315f422 | 1513 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
e260be67 | 1514 | |
f6db8347 | 1515 | #ifdef CONFIG_SCHED_INFO |
1da177e4 LT |
1516 | struct sched_info sched_info; |
1517 | #endif | |
1518 | ||
1519 | struct list_head tasks; | |
806c09a7 | 1520 | #ifdef CONFIG_SMP |
917b627d | 1521 | struct plist_node pushable_tasks; |
1baca4ce | 1522 | struct rb_node pushable_dl_tasks; |
806c09a7 | 1523 | #endif |
1da177e4 LT |
1524 | |
1525 | struct mm_struct *mm, *active_mm; | |
615d6e87 DB |
1526 | /* per-thread vma caching */ |
1527 | u32 vmacache_seqnum; | |
1528 | struct vm_area_struct *vmacache[VMACACHE_SIZE]; | |
34e55232 KH |
1529 | #if defined(SPLIT_RSS_COUNTING) |
1530 | struct task_rss_stat rss_stat; | |
1531 | #endif | |
1da177e4 | 1532 | /* task state */ |
97dc32cd | 1533 | int exit_state; |
1da177e4 LT |
1534 | int exit_code, exit_signal; |
1535 | int pdeath_signal; /* The signal sent when the parent dies */ | |
e7cc4173 | 1536 | unsigned long jobctl; /* JOBCTL_*, siglock protected */ |
9b89f6ba AE |
1537 | |
1538 | /* Used for emulating ABI behavior of previous Linux versions */ | |
97dc32cd | 1539 | unsigned int personality; |
9b89f6ba | 1540 | |
be958bdc | 1541 | /* scheduler bits, serialized by scheduler locks */ |
ca94c442 | 1542 | unsigned sched_reset_on_fork:1; |
a8e4f2ea | 1543 | unsigned sched_contributes_to_load:1; |
ff303e66 | 1544 | unsigned sched_migrated:1; |
b7e7ade3 | 1545 | unsigned sched_remote_wakeup:1; |
be958bdc PZ |
1546 | unsigned :0; /* force alignment to the next boundary */ |
1547 | ||
1548 | /* unserialized, strictly 'current' */ | |
1549 | unsigned in_execve:1; /* bit to tell LSMs we're in execve */ | |
1550 | unsigned in_iowait:1; | |
7e781418 AL |
1551 | #if !defined(TIF_RESTORE_SIGMASK) |
1552 | unsigned restore_sigmask:1; | |
1553 | #endif | |
626ebc41 TH |
1554 | #ifdef CONFIG_MEMCG |
1555 | unsigned memcg_may_oom:1; | |
127424c8 | 1556 | #ifndef CONFIG_SLOB |
6f185c29 VD |
1557 | unsigned memcg_kmem_skip_account:1; |
1558 | #endif | |
127424c8 | 1559 | #endif |
ff303e66 PZ |
1560 | #ifdef CONFIG_COMPAT_BRK |
1561 | unsigned brk_randomized:1; | |
1562 | #endif | |
6f185c29 | 1563 | |
1d4457f9 KC |
1564 | unsigned long atomic_flags; /* Flags needing atomic access. */ |
1565 | ||
f56141e3 AL |
1566 | struct restart_block restart_block; |
1567 | ||
1da177e4 LT |
1568 | pid_t pid; |
1569 | pid_t tgid; | |
0a425405 | 1570 | |
1314562a | 1571 | #ifdef CONFIG_CC_STACKPROTECTOR |
0a425405 AV |
1572 | /* Canary value for the -fstack-protector gcc feature */ |
1573 | unsigned long stack_canary; | |
1314562a | 1574 | #endif |
4d1d61a6 | 1575 | /* |
1da177e4 | 1576 | * pointers to (original) parent process, youngest child, younger sibling, |
4d1d61a6 | 1577 | * older sibling, respectively. (p->father can be replaced with |
f470021a | 1578 | * p->real_parent->pid) |
1da177e4 | 1579 | */ |
abd63bc3 KC |
1580 | struct task_struct __rcu *real_parent; /* real parent process */ |
1581 | struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */ | |
1da177e4 | 1582 | /* |
f470021a | 1583 | * children/sibling forms the list of my natural children |
1da177e4 LT |
1584 | */ |
1585 | struct list_head children; /* list of my children */ | |
1586 | struct list_head sibling; /* linkage in my parent's children list */ | |
1587 | struct task_struct *group_leader; /* threadgroup leader */ | |
1588 | ||
f470021a RM |
1589 | /* |
1590 | * ptraced is the list of tasks this task is using ptrace on. | |
1591 | * This includes both natural children and PTRACE_ATTACH targets. | |
1592 | * p->ptrace_entry is p's link on the p->parent->ptraced list. | |
1593 | */ | |
1594 | struct list_head ptraced; | |
1595 | struct list_head ptrace_entry; | |
1596 | ||
1da177e4 | 1597 | /* PID/PID hash table linkage. */ |
92476d7f | 1598 | struct pid_link pids[PIDTYPE_MAX]; |
47e65328 | 1599 | struct list_head thread_group; |
0c740d0a | 1600 | struct list_head thread_node; |
1da177e4 LT |
1601 | |
1602 | struct completion *vfork_done; /* for vfork() */ | |
1603 | int __user *set_child_tid; /* CLONE_CHILD_SETTID */ | |
1604 | int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ | |
1605 | ||
c66f08be | 1606 | cputime_t utime, stime, utimescaled, stimescaled; |
9ac52315 | 1607 | cputime_t gtime; |
9d7fb042 | 1608 | struct prev_cputime prev_cputime; |
6a61671b | 1609 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
b7ce2277 | 1610 | seqcount_t vtime_seqcount; |
6a61671b FW |
1611 | unsigned long long vtime_snap; |
1612 | enum { | |
7098c1ea FW |
1613 | /* Task is sleeping or running in a CPU with VTIME inactive */ |
1614 | VTIME_INACTIVE = 0, | |
1615 | /* Task runs in userspace in a CPU with VTIME active */ | |
6a61671b | 1616 | VTIME_USER, |
7098c1ea | 1617 | /* Task runs in kernelspace in a CPU with VTIME active */ |
6a61671b FW |
1618 | VTIME_SYS, |
1619 | } vtime_snap_whence; | |
d99ca3b9 | 1620 | #endif |
d027d45d FW |
1621 | |
1622 | #ifdef CONFIG_NO_HZ_FULL | |
f009a7a7 | 1623 | atomic_t tick_dep_mask; |
d027d45d | 1624 | #endif |
1da177e4 | 1625 | unsigned long nvcsw, nivcsw; /* context switch counts */ |
ccbf62d8 | 1626 | u64 start_time; /* monotonic time in nsec */ |
57e0be04 | 1627 | u64 real_start_time; /* boot based time in nsec */ |
1da177e4 LT |
1628 | /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ |
1629 | unsigned long min_flt, maj_flt; | |
1630 | ||
f06febc9 | 1631 | struct task_cputime cputime_expires; |
1da177e4 LT |
1632 | struct list_head cpu_timers[3]; |
1633 | ||
1634 | /* process credentials */ | |
1b0ba1c9 | 1635 | const struct cred __rcu *real_cred; /* objective and real subjective task |
3b11a1de | 1636 | * credentials (COW) */ |
1b0ba1c9 | 1637 | const struct cred __rcu *cred; /* effective (overridable) subjective task |
3b11a1de | 1638 | * credentials (COW) */ |
36772092 PBG |
1639 | char comm[TASK_COMM_LEN]; /* executable name excluding path |
1640 | - access with [gs]et_task_comm (which lock | |
1641 | it with task_lock()) | |
221af7f8 | 1642 | - initialized normally by setup_new_exec */ |
1da177e4 | 1643 | /* file system info */ |
756daf26 | 1644 | struct nameidata *nameidata; |
3d5b6fcc | 1645 | #ifdef CONFIG_SYSVIPC |
1da177e4 LT |
1646 | /* ipc stuff */ |
1647 | struct sysv_sem sysvsem; | |
ab602f79 | 1648 | struct sysv_shm sysvshm; |
3d5b6fcc | 1649 | #endif |
e162b39a | 1650 | #ifdef CONFIG_DETECT_HUNG_TASK |
82a1fcb9 | 1651 | /* hung task detection */ |
82a1fcb9 IM |
1652 | unsigned long last_switch_count; |
1653 | #endif | |
1da177e4 LT |
1654 | /* filesystem information */ |
1655 | struct fs_struct *fs; | |
1656 | /* open file information */ | |
1657 | struct files_struct *files; | |
1651e14e | 1658 | /* namespaces */ |
ab516013 | 1659 | struct nsproxy *nsproxy; |
1da177e4 LT |
1660 | /* signal handlers */ |
1661 | struct signal_struct *signal; | |
1662 | struct sighand_struct *sighand; | |
1663 | ||
1664 | sigset_t blocked, real_blocked; | |
f3de272b | 1665 | sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ |
1da177e4 LT |
1666 | struct sigpending pending; |
1667 | ||
1668 | unsigned long sas_ss_sp; | |
1669 | size_t sas_ss_size; | |
2a742138 | 1670 | unsigned sas_ss_flags; |
2e01fabe | 1671 | |
67d12145 | 1672 | struct callback_head *task_works; |
e73f8959 | 1673 | |
1da177e4 | 1674 | struct audit_context *audit_context; |
bfef93a5 | 1675 | #ifdef CONFIG_AUDITSYSCALL |
e1760bd5 | 1676 | kuid_t loginuid; |
4746ec5b | 1677 | unsigned int sessionid; |
bfef93a5 | 1678 | #endif |
932ecebb | 1679 | struct seccomp seccomp; |
1da177e4 LT |
1680 | |
1681 | /* Thread group tracking */ | |
1682 | u32 parent_exec_id; | |
1683 | u32 self_exec_id; | |
58568d2a MX |
1684 | /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, |
1685 | * mempolicy */ | |
1da177e4 | 1686 | spinlock_t alloc_lock; |
1da177e4 | 1687 | |
b29739f9 | 1688 | /* Protection of the PI data structures: */ |
1d615482 | 1689 | raw_spinlock_t pi_lock; |
b29739f9 | 1690 | |
76751049 PZ |
1691 | struct wake_q_node wake_q; |
1692 | ||
23f78d4a IM |
1693 | #ifdef CONFIG_RT_MUTEXES |
1694 | /* PI waiters blocked on a rt_mutex held by this task */ | |
fb00aca4 PZ |
1695 | struct rb_root pi_waiters; |
1696 | struct rb_node *pi_waiters_leftmost; | |
23f78d4a IM |
1697 | /* Deadlock detection and priority inheritance handling */ |
1698 | struct rt_mutex_waiter *pi_blocked_on; | |
23f78d4a IM |
1699 | #endif |
1700 | ||
408894ee IM |
1701 | #ifdef CONFIG_DEBUG_MUTEXES |
1702 | /* mutex deadlock detection */ | |
1703 | struct mutex_waiter *blocked_on; | |
1704 | #endif | |
de30a2b3 IM |
1705 | #ifdef CONFIG_TRACE_IRQFLAGS |
1706 | unsigned int irq_events; | |
de30a2b3 | 1707 | unsigned long hardirq_enable_ip; |
de30a2b3 | 1708 | unsigned long hardirq_disable_ip; |
fa1452e8 | 1709 | unsigned int hardirq_enable_event; |
de30a2b3 | 1710 | unsigned int hardirq_disable_event; |
fa1452e8 HS |
1711 | int hardirqs_enabled; |
1712 | int hardirq_context; | |
de30a2b3 | 1713 | unsigned long softirq_disable_ip; |
de30a2b3 | 1714 | unsigned long softirq_enable_ip; |
fa1452e8 | 1715 | unsigned int softirq_disable_event; |
de30a2b3 | 1716 | unsigned int softirq_enable_event; |
fa1452e8 | 1717 | int softirqs_enabled; |
de30a2b3 IM |
1718 | int softirq_context; |
1719 | #endif | |
fbb9ce95 | 1720 | #ifdef CONFIG_LOCKDEP |
bdb9441e | 1721 | # define MAX_LOCK_DEPTH 48UL |
fbb9ce95 IM |
1722 | u64 curr_chain_key; |
1723 | int lockdep_depth; | |
fbb9ce95 | 1724 | unsigned int lockdep_recursion; |
c7aceaba | 1725 | struct held_lock held_locks[MAX_LOCK_DEPTH]; |
cf40bd16 | 1726 | gfp_t lockdep_reclaim_gfp; |
fbb9ce95 | 1727 | #endif |
c6d30853 AR |
1728 | #ifdef CONFIG_UBSAN |
1729 | unsigned int in_ubsan; | |
1730 | #endif | |
408894ee | 1731 | |
1da177e4 LT |
1732 | /* journalling filesystem info */ |
1733 | void *journal_info; | |
1734 | ||
d89d8796 | 1735 | /* stacked block device info */ |
bddd87c7 | 1736 | struct bio_list *bio_list; |
d89d8796 | 1737 | |
73c10101 JA |
1738 | #ifdef CONFIG_BLOCK |
1739 | /* stack plugging */ | |
1740 | struct blk_plug *plug; | |
1741 | #endif | |
1742 | ||
1da177e4 LT |
1743 | /* VM state */ |
1744 | struct reclaim_state *reclaim_state; | |
1745 | ||
1da177e4 LT |
1746 | struct backing_dev_info *backing_dev_info; |
1747 | ||
1748 | struct io_context *io_context; | |
1749 | ||
1750 | unsigned long ptrace_message; | |
1751 | siginfo_t *last_siginfo; /* For ptrace use. */ | |
7c3ab738 | 1752 | struct task_io_accounting ioac; |
8f0ab514 | 1753 | #if defined(CONFIG_TASK_XACCT) |
1da177e4 LT |
1754 | u64 acct_rss_mem1; /* accumulated rss usage */ |
1755 | u64 acct_vm_mem1; /* accumulated virtual memory usage */ | |
49b5cf34 | 1756 | cputime_t acct_timexpd; /* stime + utime since last update */ |
1da177e4 LT |
1757 | #endif |
1758 | #ifdef CONFIG_CPUSETS | |
58568d2a | 1759 | nodemask_t mems_allowed; /* Protected by alloc_lock */ |
cc9a6c87 | 1760 | seqcount_t mems_allowed_seq; /* Seqence no to catch updates */ |
825a46af | 1761 | int cpuset_mem_spread_rotor; |
6adef3eb | 1762 | int cpuset_slab_spread_rotor; |
1da177e4 | 1763 | #endif |
ddbcc7e8 | 1764 | #ifdef CONFIG_CGROUPS |
817929ec | 1765 | /* Control Group info protected by css_set_lock */ |
2c392b8c | 1766 | struct css_set __rcu *cgroups; |
817929ec PM |
1767 | /* cg_list protected by css_set_lock and tsk->alloc_lock */ |
1768 | struct list_head cg_list; | |
ddbcc7e8 | 1769 | #endif |
42b2dd0a | 1770 | #ifdef CONFIG_FUTEX |
0771dfef | 1771 | struct robust_list_head __user *robust_list; |
34f192c6 IM |
1772 | #ifdef CONFIG_COMPAT |
1773 | struct compat_robust_list_head __user *compat_robust_list; | |
1774 | #endif | |
c87e2837 IM |
1775 | struct list_head pi_state_list; |
1776 | struct futex_pi_state *pi_state_cache; | |
c7aceaba | 1777 | #endif |
cdd6c482 | 1778 | #ifdef CONFIG_PERF_EVENTS |
8dc85d54 | 1779 | struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; |
cdd6c482 IM |
1780 | struct mutex perf_event_mutex; |
1781 | struct list_head perf_event_list; | |
a63eaf34 | 1782 | #endif |
8f47b187 TG |
1783 | #ifdef CONFIG_DEBUG_PREEMPT |
1784 | unsigned long preempt_disable_ip; | |
1785 | #endif | |
c7aceaba | 1786 | #ifdef CONFIG_NUMA |
58568d2a | 1787 | struct mempolicy *mempolicy; /* Protected by alloc_lock */ |
c7aceaba | 1788 | short il_next; |
207205a2 | 1789 | short pref_node_fork; |
42b2dd0a | 1790 | #endif |
cbee9f88 PZ |
1791 | #ifdef CONFIG_NUMA_BALANCING |
1792 | int numa_scan_seq; | |
cbee9f88 | 1793 | unsigned int numa_scan_period; |
598f0ec0 | 1794 | unsigned int numa_scan_period_max; |
de1c9ce6 | 1795 | int numa_preferred_nid; |
6b9a7460 | 1796 | unsigned long numa_migrate_retry; |
cbee9f88 | 1797 | u64 node_stamp; /* migration stamp */ |
7e2703e6 RR |
1798 | u64 last_task_numa_placement; |
1799 | u64 last_sum_exec_runtime; | |
cbee9f88 | 1800 | struct callback_head numa_work; |
f809ca9a | 1801 | |
8c8a743c PZ |
1802 | struct list_head numa_entry; |
1803 | struct numa_group *numa_group; | |
1804 | ||
745d6147 | 1805 | /* |
44dba3d5 IM |
1806 | * numa_faults is an array split into four regions: |
1807 | * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer | |
1808 | * in this precise order. | |
1809 | * | |
1810 | * faults_memory: Exponential decaying average of faults on a per-node | |
1811 | * basis. Scheduling placement decisions are made based on these | |
1812 | * counts. The values remain static for the duration of a PTE scan. | |
1813 | * faults_cpu: Track the nodes the process was running on when a NUMA | |
1814 | * hinting fault was incurred. | |
1815 | * faults_memory_buffer and faults_cpu_buffer: Record faults per node | |
1816 | * during the current scan window. When the scan completes, the counts | |
1817 | * in faults_memory and faults_cpu decay and these values are copied. | |
745d6147 | 1818 | */ |
44dba3d5 | 1819 | unsigned long *numa_faults; |
83e1d2cd | 1820 | unsigned long total_numa_faults; |
745d6147 | 1821 | |
04bb2f94 RR |
1822 | /* |
1823 | * numa_faults_locality tracks if faults recorded during the last | |
074c2381 MG |
1824 | * scan window were remote/local or failed to migrate. The task scan |
1825 | * period is adapted based on the locality of the faults with different | |
1826 | * weights depending on whether they were shared or private faults | |
04bb2f94 | 1827 | */ |
074c2381 | 1828 | unsigned long numa_faults_locality[3]; |
04bb2f94 | 1829 | |
b32e86b4 | 1830 | unsigned long numa_pages_migrated; |
cbee9f88 PZ |
1831 | #endif /* CONFIG_NUMA_BALANCING */ |
1832 | ||
72b252ae MG |
1833 | #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH |
1834 | struct tlbflush_unmap_batch tlb_ubc; | |
1835 | #endif | |
1836 | ||
e56d0903 | 1837 | struct rcu_head rcu; |
b92ce558 JA |
1838 | |
1839 | /* | |
1840 | * cache last used pipe for splice | |
1841 | */ | |
1842 | struct pipe_inode_info *splice_pipe; | |
5640f768 ED |
1843 | |
1844 | struct page_frag task_frag; | |
1845 | ||
ca74e92b SN |
1846 | #ifdef CONFIG_TASK_DELAY_ACCT |
1847 | struct task_delay_info *delays; | |
f4f154fd AM |
1848 | #endif |
1849 | #ifdef CONFIG_FAULT_INJECTION | |
1850 | int make_it_fail; | |
ca74e92b | 1851 | #endif |
9d823e8f WF |
1852 | /* |
1853 | * when (nr_dirtied >= nr_dirtied_pause), it's time to call | |
1854 | * balance_dirty_pages() for some dirty throttling pause | |
1855 | */ | |
1856 | int nr_dirtied; | |
1857 | int nr_dirtied_pause; | |
83712358 | 1858 | unsigned long dirty_paused_when; /* start of a write-and-pause period */ |
9d823e8f | 1859 | |
9745512c AV |
1860 | #ifdef CONFIG_LATENCYTOP |
1861 | int latency_record_count; | |
1862 | struct latency_record latency_record[LT_SAVECOUNT]; | |
1863 | #endif | |
6976675d AV |
1864 | /* |
1865 | * time slack values; these are used to round up poll() and | |
1866 | * select() etc timeout values. These are in nanoseconds. | |
1867 | */ | |
da8b44d5 JS |
1868 | u64 timer_slack_ns; |
1869 | u64 default_timer_slack_ns; | |
f8d570a4 | 1870 | |
0b24becc AR |
1871 | #ifdef CONFIG_KASAN |
1872 | unsigned int kasan_depth; | |
1873 | #endif | |
fb52607a | 1874 | #ifdef CONFIG_FUNCTION_GRAPH_TRACER |
3ad2f3fb | 1875 | /* Index of current stored address in ret_stack */ |
f201ae23 FW |
1876 | int curr_ret_stack; |
1877 | /* Stack of return addresses for return function tracing */ | |
1878 | struct ftrace_ret_stack *ret_stack; | |
8aef2d28 SR |
1879 | /* time stamp for last schedule */ |
1880 | unsigned long long ftrace_timestamp; | |
f201ae23 FW |
1881 | /* |
1882 | * Number of functions that haven't been traced | |
1883 | * because of depth overrun. | |
1884 | */ | |
1885 | atomic_t trace_overrun; | |
380c4b14 FW |
1886 | /* Pause for the tracing */ |
1887 | atomic_t tracing_graph_pause; | |
f201ae23 | 1888 | #endif |
ea4e2bc4 SR |
1889 | #ifdef CONFIG_TRACING |
1890 | /* state flags for use by tracers */ | |
1891 | unsigned long trace; | |
b1cff0ad | 1892 | /* bitmask and counter of trace recursion */ |
261842b7 SR |
1893 | unsigned long trace_recursion; |
1894 | #endif /* CONFIG_TRACING */ | |
5c9a8750 DV |
1895 | #ifdef CONFIG_KCOV |
1896 | /* Coverage collection mode enabled for this task (0 if disabled). */ | |
1897 | enum kcov_mode kcov_mode; | |
1898 | /* Size of the kcov_area. */ | |
1899 | unsigned kcov_size; | |
1900 | /* Buffer for coverage collection. */ | |
1901 | void *kcov_area; | |
1902 | /* kcov desciptor wired with this task or NULL. */ | |
1903 | struct kcov *kcov; | |
1904 | #endif | |
6f185c29 | 1905 | #ifdef CONFIG_MEMCG |
626ebc41 TH |
1906 | struct mem_cgroup *memcg_in_oom; |
1907 | gfp_t memcg_oom_gfp_mask; | |
1908 | int memcg_oom_order; | |
b23afb93 TH |
1909 | |
1910 | /* number of pages to reclaim on returning to userland */ | |
1911 | unsigned int memcg_nr_pages_over_high; | |
569b846d | 1912 | #endif |
0326f5a9 SD |
1913 | #ifdef CONFIG_UPROBES |
1914 | struct uprobe_task *utask; | |
0326f5a9 | 1915 | #endif |
cafe5635 KO |
1916 | #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE) |
1917 | unsigned int sequential_io; | |
1918 | unsigned int sequential_io_avg; | |
1919 | #endif | |
8eb23b9f PZ |
1920 | #ifdef CONFIG_DEBUG_ATOMIC_SLEEP |
1921 | unsigned long task_state_change; | |
1922 | #endif | |
8bcbde54 | 1923 | int pagefault_disabled; |
03049269 | 1924 | #ifdef CONFIG_MMU |
29c696e1 | 1925 | struct task_struct *oom_reaper_list; |
03049269 | 1926 | #endif |
91a034ed MD |
1927 | #ifdef CONFIG_RSEQ |
1928 | struct rseq __user *rseq; | |
1929 | u32 rseq_event_counter; | |
1930 | unsigned int rseq_refcount; | |
1931 | #endif | |
0c8c0f03 DH |
1932 | /* CPU-specific state of this task */ |
1933 | struct thread_struct thread; | |
1934 | /* | |
1935 | * WARNING: on x86, 'thread_struct' contains a variable-sized | |
1936 | * structure. It *MUST* be at the end of 'task_struct'. | |
1937 | * | |
1938 | * Do not put anything below here! | |
1939 | */ | |
1da177e4 LT |
1940 | }; |
1941 | ||
5aaeb5c0 IM |
1942 | #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT |
1943 | extern int arch_task_struct_size __read_mostly; | |
1944 | #else | |
1945 | # define arch_task_struct_size (sizeof(struct task_struct)) | |
1946 | #endif | |
0c8c0f03 | 1947 | |
76e6eee0 | 1948 | /* Future-safe accessor for struct task_struct's cpus_allowed. */ |
a4636818 | 1949 | #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed) |
76e6eee0 | 1950 | |
50605ffb TG |
1951 | static inline int tsk_nr_cpus_allowed(struct task_struct *p) |
1952 | { | |
1953 | return p->nr_cpus_allowed; | |
1954 | } | |
1955 | ||
6688cc05 PZ |
1956 | #define TNF_MIGRATED 0x01 |
1957 | #define TNF_NO_GROUP 0x02 | |
dabe1d99 | 1958 | #define TNF_SHARED 0x04 |
04bb2f94 | 1959 | #define TNF_FAULT_LOCAL 0x08 |
074c2381 | 1960 | #define TNF_MIGRATE_FAIL 0x10 |
6688cc05 | 1961 | |
b18dc5f2 MH |
1962 | static inline bool in_vfork(struct task_struct *tsk) |
1963 | { | |
1964 | bool ret; | |
1965 | ||
1966 | /* | |
1967 | * need RCU to access ->real_parent if CLONE_VM was used along with | |
1968 | * CLONE_PARENT. | |
1969 | * | |
1970 | * We check real_parent->mm == tsk->mm because CLONE_VFORK does not | |
1971 | * imply CLONE_VM | |
1972 | * | |
1973 | * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus | |
1974 | * ->real_parent is not necessarily the task doing vfork(), so in | |
1975 | * theory we can't rely on task_lock() if we want to dereference it. | |
1976 | * | |
1977 | * And in this case we can't trust the real_parent->mm == tsk->mm | |
1978 | * check, it can be false negative. But we do not care, if init or | |
1979 | * another oom-unkillable task does this it should blame itself. | |
1980 | */ | |
1981 | rcu_read_lock(); | |
1982 | ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm; | |
1983 | rcu_read_unlock(); | |
1984 | ||
1985 | return ret; | |
1986 | } | |
1987 | ||
cbee9f88 | 1988 | #ifdef CONFIG_NUMA_BALANCING |
6688cc05 | 1989 | extern void task_numa_fault(int last_node, int node, int pages, int flags); |
e29cf08b | 1990 | extern pid_t task_numa_group_id(struct task_struct *p); |
1a687c2e | 1991 | extern void set_numabalancing_state(bool enabled); |
82727018 | 1992 | extern void task_numa_free(struct task_struct *p); |
10f39042 RR |
1993 | extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page, |
1994 | int src_nid, int dst_cpu); | |
cbee9f88 | 1995 | #else |
ac8e895b | 1996 | static inline void task_numa_fault(int last_node, int node, int pages, |
6688cc05 | 1997 | int flags) |
cbee9f88 PZ |
1998 | { |
1999 | } | |
e29cf08b MG |
2000 | static inline pid_t task_numa_group_id(struct task_struct *p) |
2001 | { | |
2002 | return 0; | |
2003 | } | |
1a687c2e MG |
2004 | static inline void set_numabalancing_state(bool enabled) |
2005 | { | |
2006 | } | |
82727018 RR |
2007 | static inline void task_numa_free(struct task_struct *p) |
2008 | { | |
2009 | } | |
10f39042 RR |
2010 | static inline bool should_numa_migrate_memory(struct task_struct *p, |
2011 | struct page *page, int src_nid, int dst_cpu) | |
2012 | { | |
2013 | return true; | |
2014 | } | |
cbee9f88 PZ |
2015 | #endif |
2016 | ||
e868171a | 2017 | static inline struct pid *task_pid(struct task_struct *task) |
22c935f4 EB |
2018 | { |
2019 | return task->pids[PIDTYPE_PID].pid; | |
2020 | } | |
2021 | ||
e868171a | 2022 | static inline struct pid *task_tgid(struct task_struct *task) |
22c935f4 EB |
2023 | { |
2024 | return task->group_leader->pids[PIDTYPE_PID].pid; | |
2025 | } | |
2026 | ||
6dda81f4 ON |
2027 | /* |
2028 | * Without tasklist or rcu lock it is not safe to dereference | |
2029 | * the result of task_pgrp/task_session even if task == current, | |
2030 | * we can race with another thread doing sys_setsid/sys_setpgid. | |
2031 | */ | |
e868171a | 2032 | static inline struct pid *task_pgrp(struct task_struct *task) |
22c935f4 EB |
2033 | { |
2034 | return task->group_leader->pids[PIDTYPE_PGID].pid; | |
2035 | } | |
2036 | ||
e868171a | 2037 | static inline struct pid *task_session(struct task_struct *task) |
22c935f4 EB |
2038 | { |
2039 | return task->group_leader->pids[PIDTYPE_SID].pid; | |
2040 | } | |
2041 | ||
7af57294 PE |
2042 | struct pid_namespace; |
2043 | ||
2044 | /* | |
2045 | * the helpers to get the task's different pids as they are seen | |
2046 | * from various namespaces | |
2047 | * | |
2048 | * task_xid_nr() : global id, i.e. the id seen from the init namespace; | |
44c4e1b2 EB |
2049 | * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of |
2050 | * current. | |
7af57294 PE |
2051 | * task_xid_nr_ns() : id seen from the ns specified; |
2052 | * | |
2053 | * set_task_vxid() : assigns a virtual id to a task; | |
2054 | * | |
7af57294 PE |
2055 | * see also pid_nr() etc in include/linux/pid.h |
2056 | */ | |
52ee2dfd ON |
2057 | pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, |
2058 | struct pid_namespace *ns); | |
7af57294 | 2059 | |
e868171a | 2060 | static inline pid_t task_pid_nr(struct task_struct *tsk) |
7af57294 PE |
2061 | { |
2062 | return tsk->pid; | |
2063 | } | |
2064 | ||
52ee2dfd ON |
2065 | static inline pid_t task_pid_nr_ns(struct task_struct *tsk, |
2066 | struct pid_namespace *ns) | |
2067 | { | |
2068 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); | |
2069 | } | |
7af57294 PE |
2070 | |
2071 | static inline pid_t task_pid_vnr(struct task_struct *tsk) | |
2072 | { | |
52ee2dfd | 2073 | return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); |
7af57294 PE |
2074 | } |
2075 | ||
2076 | ||
e868171a | 2077 | static inline pid_t task_tgid_nr(struct task_struct *tsk) |
7af57294 PE |
2078 | { |
2079 | return tsk->tgid; | |
2080 | } | |
2081 | ||
2f2a3a46 | 2082 | pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); |
7af57294 PE |
2083 | |
2084 | static inline pid_t task_tgid_vnr(struct task_struct *tsk) | |
2085 | { | |
2086 | return pid_vnr(task_tgid(tsk)); | |
2087 | } | |
2088 | ||
2089 | ||
80e0b6e8 | 2090 | static inline int pid_alive(const struct task_struct *p); |
ad36d282 RGB |
2091 | static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns) |
2092 | { | |
2093 | pid_t pid = 0; | |
2094 | ||
2095 | rcu_read_lock(); | |
2096 | if (pid_alive(tsk)) | |
2097 | pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns); | |
2098 | rcu_read_unlock(); | |
2099 | ||
2100 | return pid; | |
2101 | } | |
2102 | ||
2103 | static inline pid_t task_ppid_nr(const struct task_struct *tsk) | |
2104 | { | |
2105 | return task_ppid_nr_ns(tsk, &init_pid_ns); | |
2106 | } | |
2107 | ||
52ee2dfd ON |
2108 | static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, |
2109 | struct pid_namespace *ns) | |
7af57294 | 2110 | { |
52ee2dfd | 2111 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); |
7af57294 PE |
2112 | } |
2113 | ||
7af57294 PE |
2114 | static inline pid_t task_pgrp_vnr(struct task_struct *tsk) |
2115 | { | |
52ee2dfd | 2116 | return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); |
7af57294 PE |
2117 | } |
2118 | ||
2119 | ||
52ee2dfd ON |
2120 | static inline pid_t task_session_nr_ns(struct task_struct *tsk, |
2121 | struct pid_namespace *ns) | |
7af57294 | 2122 | { |
52ee2dfd | 2123 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); |
7af57294 PE |
2124 | } |
2125 | ||
7af57294 PE |
2126 | static inline pid_t task_session_vnr(struct task_struct *tsk) |
2127 | { | |
52ee2dfd | 2128 | return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); |
7af57294 PE |
2129 | } |
2130 | ||
1b0f7ffd ON |
2131 | /* obsolete, do not use */ |
2132 | static inline pid_t task_pgrp_nr(struct task_struct *tsk) | |
2133 | { | |
2134 | return task_pgrp_nr_ns(tsk, &init_pid_ns); | |
2135 | } | |
7af57294 | 2136 | |
1da177e4 LT |
2137 | /** |
2138 | * pid_alive - check that a task structure is not stale | |
2139 | * @p: Task structure to be checked. | |
2140 | * | |
2141 | * Test if a process is not yet dead (at most zombie state) | |
2142 | * If pid_alive fails, then pointers within the task structure | |
2143 | * can be stale and must not be dereferenced. | |
e69f6186 YB |
2144 | * |
2145 | * Return: 1 if the process is alive. 0 otherwise. | |
1da177e4 | 2146 | */ |
ad36d282 | 2147 | static inline int pid_alive(const struct task_struct *p) |
1da177e4 | 2148 | { |
92476d7f | 2149 | return p->pids[PIDTYPE_PID].pid != NULL; |
1da177e4 LT |
2150 | } |
2151 | ||
f400e198 | 2152 | /** |
570f5241 SS |
2153 | * is_global_init - check if a task structure is init. Since init |
2154 | * is free to have sub-threads we need to check tgid. | |
3260259f H |
2155 | * @tsk: Task structure to be checked. |
2156 | * | |
2157 | * Check if a task structure is the first user space task the kernel created. | |
e69f6186 YB |
2158 | * |
2159 | * Return: 1 if the task structure is init. 0 otherwise. | |
b460cbc5 | 2160 | */ |
e868171a | 2161 | static inline int is_global_init(struct task_struct *tsk) |
b461cc03 | 2162 | { |
570f5241 | 2163 | return task_tgid_nr(tsk) == 1; |
b461cc03 | 2164 | } |
b460cbc5 | 2165 | |
9ec52099 CLG |
2166 | extern struct pid *cad_pid; |
2167 | ||
1da177e4 | 2168 | extern void free_task(struct task_struct *tsk); |
1da177e4 | 2169 | #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) |
e56d0903 | 2170 | |
158d9ebd | 2171 | extern void __put_task_struct(struct task_struct *t); |
e56d0903 IM |
2172 | |
2173 | static inline void put_task_struct(struct task_struct *t) | |
2174 | { | |
2175 | if (atomic_dec_and_test(&t->usage)) | |
8c7904a0 | 2176 | __put_task_struct(t); |
e56d0903 | 2177 | } |
1da177e4 | 2178 | |
150593bf ON |
2179 | struct task_struct *task_rcu_dereference(struct task_struct **ptask); |
2180 | struct task_struct *try_get_task_struct(struct task_struct **ptask); | |
2181 | ||
6a61671b FW |
2182 | #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN |
2183 | extern void task_cputime(struct task_struct *t, | |
2184 | cputime_t *utime, cputime_t *stime); | |
2185 | extern void task_cputime_scaled(struct task_struct *t, | |
2186 | cputime_t *utimescaled, cputime_t *stimescaled); | |
2187 | extern cputime_t task_gtime(struct task_struct *t); | |
2188 | #else | |
6fac4829 FW |
2189 | static inline void task_cputime(struct task_struct *t, |
2190 | cputime_t *utime, cputime_t *stime) | |
2191 | { | |
2192 | if (utime) | |
2193 | *utime = t->utime; | |
2194 | if (stime) | |
2195 | *stime = t->stime; | |
2196 | } | |
2197 | ||
2198 | static inline void task_cputime_scaled(struct task_struct *t, | |
2199 | cputime_t *utimescaled, | |
2200 | cputime_t *stimescaled) | |
2201 | { | |
2202 | if (utimescaled) | |
2203 | *utimescaled = t->utimescaled; | |
2204 | if (stimescaled) | |
2205 | *stimescaled = t->stimescaled; | |
2206 | } | |
6a61671b FW |
2207 | |
2208 | static inline cputime_t task_gtime(struct task_struct *t) | |
2209 | { | |
2210 | return t->gtime; | |
2211 | } | |
2212 | #endif | |
e80d0a1a FW |
2213 | extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st); |
2214 | extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st); | |
49048622 | 2215 | |
1da177e4 LT |
2216 | /* |
2217 | * Per process flags | |
2218 | */ | |
1da177e4 | 2219 | #define PF_EXITING 0x00000004 /* getting shut down */ |
778e9a9c | 2220 | #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ |
94886b84 | 2221 | #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ |
21aa9af0 | 2222 | #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ |
1da177e4 | 2223 | #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ |
4db96cf0 | 2224 | #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ |
1da177e4 LT |
2225 | #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ |
2226 | #define PF_DUMPCORE 0x00000200 /* dumped core */ | |
2227 | #define PF_SIGNALED 0x00000400 /* killed by a signal */ | |
2228 | #define PF_MEMALLOC 0x00000800 /* Allocating memory */ | |
72fa5997 | 2229 | #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */ |
1da177e4 | 2230 | #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ |
774a1221 | 2231 | #define PF_USED_ASYNC 0x00004000 /* used async_schedule*(), used by module init */ |
1da177e4 LT |
2232 | #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ |
2233 | #define PF_FROZEN 0x00010000 /* frozen for system suspend */ | |
2234 | #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ | |
2235 | #define PF_KSWAPD 0x00040000 /* I am kswapd */ | |
21caf2fc | 2236 | #define PF_MEMALLOC_NOIO 0x00080000 /* Allocating memory without IO involved */ |
1da177e4 | 2237 | #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ |
246bb0b1 | 2238 | #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ |
b31dc66a JA |
2239 | #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ |
2240 | #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ | |
14a40ffc | 2241 | #define PF_NO_SETAFFINITY 0x04000000 /* Userland is not allowed to meddle with cpus_allowed */ |
4db96cf0 | 2242 | #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ |
61a87122 | 2243 | #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ |
58a69cb4 | 2244 | #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ |
2b44c4db | 2245 | #define PF_SUSPEND_TASK 0x80000000 /* this thread called freeze_processes and should not be frozen */ |
1da177e4 LT |
2246 | |
2247 | /* | |
2248 | * Only the _current_ task can read/write to tsk->flags, but other | |
2249 | * tasks can access tsk->flags in readonly mode for example | |
2250 | * with tsk_used_math (like during threaded core dumping). | |
2251 | * There is however an exception to this rule during ptrace | |
2252 | * or during fork: the ptracer task is allowed to write to the | |
2253 | * child->flags of its traced child (same goes for fork, the parent | |
2254 | * can write to the child->flags), because we're guaranteed the | |
2255 | * child is not running and in turn not changing child->flags | |
2256 | * at the same time the parent does it. | |
2257 | */ | |
2258 | #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) | |
2259 | #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) | |
2260 | #define clear_used_math() clear_stopped_child_used_math(current) | |
2261 | #define set_used_math() set_stopped_child_used_math(current) | |
2262 | #define conditional_stopped_child_used_math(condition, child) \ | |
2263 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) | |
2264 | #define conditional_used_math(condition) \ | |
2265 | conditional_stopped_child_used_math(condition, current) | |
2266 | #define copy_to_stopped_child_used_math(child) \ | |
2267 | do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) | |
2268 | /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ | |
2269 | #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) | |
2270 | #define used_math() tsk_used_math(current) | |
2271 | ||
934f3072 JB |
2272 | /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags |
2273 | * __GFP_FS is also cleared as it implies __GFP_IO. | |
2274 | */ | |
21caf2fc ML |
2275 | static inline gfp_t memalloc_noio_flags(gfp_t flags) |
2276 | { | |
2277 | if (unlikely(current->flags & PF_MEMALLOC_NOIO)) | |
934f3072 | 2278 | flags &= ~(__GFP_IO | __GFP_FS); |
21caf2fc ML |
2279 | return flags; |
2280 | } | |
2281 | ||
2282 | static inline unsigned int memalloc_noio_save(void) | |
2283 | { | |
2284 | unsigned int flags = current->flags & PF_MEMALLOC_NOIO; | |
2285 | current->flags |= PF_MEMALLOC_NOIO; | |
2286 | return flags; | |
2287 | } | |
2288 | ||
2289 | static inline void memalloc_noio_restore(unsigned int flags) | |
2290 | { | |
2291 | current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags; | |
2292 | } | |
2293 | ||
1d4457f9 | 2294 | /* Per-process atomic flags. */ |
a2b86f77 | 2295 | #define PFA_NO_NEW_PRIVS 0 /* May not gain new privileges. */ |
2ad654bc ZL |
2296 | #define PFA_SPREAD_PAGE 1 /* Spread page cache over cpuset */ |
2297 | #define PFA_SPREAD_SLAB 2 /* Spread some slab caches over cpuset */ | |
77ed2c57 | 2298 | #define PFA_LMK_WAITING 3 /* Lowmemorykiller is waiting */ |
2ad654bc | 2299 | |
1d4457f9 | 2300 | |
e0e5070b ZL |
2301 | #define TASK_PFA_TEST(name, func) \ |
2302 | static inline bool task_##func(struct task_struct *p) \ | |
2303 | { return test_bit(PFA_##name, &p->atomic_flags); } | |
2304 | #define TASK_PFA_SET(name, func) \ | |
2305 | static inline void task_set_##func(struct task_struct *p) \ | |
2306 | { set_bit(PFA_##name, &p->atomic_flags); } | |
2307 | #define TASK_PFA_CLEAR(name, func) \ | |
2308 | static inline void task_clear_##func(struct task_struct *p) \ | |
2309 | { clear_bit(PFA_##name, &p->atomic_flags); } | |
2310 | ||
2311 | TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs) | |
2312 | TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs) | |
1d4457f9 | 2313 | |
2ad654bc ZL |
2314 | TASK_PFA_TEST(SPREAD_PAGE, spread_page) |
2315 | TASK_PFA_SET(SPREAD_PAGE, spread_page) | |
2316 | TASK_PFA_CLEAR(SPREAD_PAGE, spread_page) | |
2317 | ||
2318 | TASK_PFA_TEST(SPREAD_SLAB, spread_slab) | |
2319 | TASK_PFA_SET(SPREAD_SLAB, spread_slab) | |
2320 | TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab) | |
1d4457f9 | 2321 | |
77ed2c57 TH |
2322 | TASK_PFA_TEST(LMK_WAITING, lmk_waiting) |
2323 | TASK_PFA_SET(LMK_WAITING, lmk_waiting) | |
2324 | ||
e5c1902e | 2325 | /* |
a8f072c1 | 2326 | * task->jobctl flags |
e5c1902e | 2327 | */ |
a8f072c1 | 2328 | #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */ |
e5c1902e | 2329 | |
a8f072c1 TH |
2330 | #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */ |
2331 | #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */ | |
2332 | #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */ | |
73ddff2b | 2333 | #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */ |
fb1d910c | 2334 | #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */ |
a8f072c1 | 2335 | #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */ |
544b2c91 | 2336 | #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */ |
a8f072c1 | 2337 | |
b76808e6 PD |
2338 | #define JOBCTL_STOP_DEQUEUED (1UL << JOBCTL_STOP_DEQUEUED_BIT) |
2339 | #define JOBCTL_STOP_PENDING (1UL << JOBCTL_STOP_PENDING_BIT) | |
2340 | #define JOBCTL_STOP_CONSUME (1UL << JOBCTL_STOP_CONSUME_BIT) | |
2341 | #define JOBCTL_TRAP_STOP (1UL << JOBCTL_TRAP_STOP_BIT) | |
2342 | #define JOBCTL_TRAP_NOTIFY (1UL << JOBCTL_TRAP_NOTIFY_BIT) | |
2343 | #define JOBCTL_TRAPPING (1UL << JOBCTL_TRAPPING_BIT) | |
2344 | #define JOBCTL_LISTENING (1UL << JOBCTL_LISTENING_BIT) | |
a8f072c1 | 2345 | |
fb1d910c | 2346 | #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY) |
73ddff2b | 2347 | #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK) |
3759a0d9 | 2348 | |
7dd3db54 | 2349 | extern bool task_set_jobctl_pending(struct task_struct *task, |
b76808e6 | 2350 | unsigned long mask); |
73ddff2b | 2351 | extern void task_clear_jobctl_trapping(struct task_struct *task); |
3759a0d9 | 2352 | extern void task_clear_jobctl_pending(struct task_struct *task, |
b76808e6 | 2353 | unsigned long mask); |
39efa3ef | 2354 | |
f41d911f PM |
2355 | static inline void rcu_copy_process(struct task_struct *p) |
2356 | { | |
8315f422 | 2357 | #ifdef CONFIG_PREEMPT_RCU |
f41d911f | 2358 | p->rcu_read_lock_nesting = 0; |
1d082fd0 | 2359 | p->rcu_read_unlock_special.s = 0; |
dd5d19ba | 2360 | p->rcu_blocked_node = NULL; |
f41d911f | 2361 | INIT_LIST_HEAD(&p->rcu_node_entry); |
8315f422 PM |
2362 | #endif /* #ifdef CONFIG_PREEMPT_RCU */ |
2363 | #ifdef CONFIG_TASKS_RCU | |
2364 | p->rcu_tasks_holdout = false; | |
2365 | INIT_LIST_HEAD(&p->rcu_tasks_holdout_list); | |
176f8f7a | 2366 | p->rcu_tasks_idle_cpu = -1; |
8315f422 | 2367 | #endif /* #ifdef CONFIG_TASKS_RCU */ |
f41d911f PM |
2368 | } |
2369 | ||
907aed48 MG |
2370 | static inline void tsk_restore_flags(struct task_struct *task, |
2371 | unsigned long orig_flags, unsigned long flags) | |
2372 | { | |
2373 | task->flags &= ~flags; | |
2374 | task->flags |= orig_flags & flags; | |
2375 | } | |
2376 | ||
f82f8042 JL |
2377 | extern int cpuset_cpumask_can_shrink(const struct cpumask *cur, |
2378 | const struct cpumask *trial); | |
7f51412a JL |
2379 | extern int task_can_attach(struct task_struct *p, |
2380 | const struct cpumask *cs_cpus_allowed); | |
1da177e4 | 2381 | #ifdef CONFIG_SMP |
1e1b6c51 KM |
2382 | extern void do_set_cpus_allowed(struct task_struct *p, |
2383 | const struct cpumask *new_mask); | |
2384 | ||
cd8ba7cd | 2385 | extern int set_cpus_allowed_ptr(struct task_struct *p, |
96f874e2 | 2386 | const struct cpumask *new_mask); |
1da177e4 | 2387 | #else |
1e1b6c51 KM |
2388 | static inline void do_set_cpus_allowed(struct task_struct *p, |
2389 | const struct cpumask *new_mask) | |
2390 | { | |
2391 | } | |
cd8ba7cd | 2392 | static inline int set_cpus_allowed_ptr(struct task_struct *p, |
96f874e2 | 2393 | const struct cpumask *new_mask) |
1da177e4 | 2394 | { |
96f874e2 | 2395 | if (!cpumask_test_cpu(0, new_mask)) |
1da177e4 LT |
2396 | return -EINVAL; |
2397 | return 0; | |
2398 | } | |
2399 | #endif | |
e0ad9556 | 2400 | |
3451d024 | 2401 | #ifdef CONFIG_NO_HZ_COMMON |
5167e8d5 PZ |
2402 | void calc_load_enter_idle(void); |
2403 | void calc_load_exit_idle(void); | |
2404 | #else | |
2405 | static inline void calc_load_enter_idle(void) { } | |
2406 | static inline void calc_load_exit_idle(void) { } | |
3451d024 | 2407 | #endif /* CONFIG_NO_HZ_COMMON */ |
5167e8d5 | 2408 | |
b342501c | 2409 | /* |
c676329a PZ |
2410 | * Do not use outside of architecture code which knows its limitations. |
2411 | * | |
2412 | * sched_clock() has no promise of monotonicity or bounded drift between | |
2413 | * CPUs, use (which you should not) requires disabling IRQs. | |
2414 | * | |
2415 | * Please use one of the three interfaces below. | |
b342501c | 2416 | */ |
1bbfa6f2 | 2417 | extern unsigned long long notrace sched_clock(void); |
c676329a | 2418 | /* |
489a71b0 | 2419 | * See the comment in kernel/sched/clock.c |
c676329a | 2420 | */ |
545a2bf7 | 2421 | extern u64 running_clock(void); |
c676329a PZ |
2422 | extern u64 sched_clock_cpu(int cpu); |
2423 | ||
e436d800 | 2424 | |
c1955a3d | 2425 | extern void sched_clock_init(void); |
3e51f33f | 2426 | |
c1955a3d | 2427 | #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK |
3e51f33f PZ |
2428 | static inline void sched_clock_tick(void) |
2429 | { | |
2430 | } | |
2431 | ||
2432 | static inline void sched_clock_idle_sleep_event(void) | |
2433 | { | |
2434 | } | |
2435 | ||
2436 | static inline void sched_clock_idle_wakeup_event(u64 delta_ns) | |
2437 | { | |
2438 | } | |
2c923e94 DL |
2439 | |
2440 | static inline u64 cpu_clock(int cpu) | |
2441 | { | |
2442 | return sched_clock(); | |
2443 | } | |
2444 | ||
2445 | static inline u64 local_clock(void) | |
2446 | { | |
2447 | return sched_clock(); | |
2448 | } | |
3e51f33f | 2449 | #else |
c676329a PZ |
2450 | /* |
2451 | * Architectures can set this to 1 if they have specified | |
2452 | * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, | |
2453 | * but then during bootup it turns out that sched_clock() | |
2454 | * is reliable after all: | |
2455 | */ | |
35af99e6 PZ |
2456 | extern int sched_clock_stable(void); |
2457 | extern void set_sched_clock_stable(void); | |
2458 | extern void clear_sched_clock_stable(void); | |
c676329a | 2459 | |
3e51f33f PZ |
2460 | extern void sched_clock_tick(void); |
2461 | extern void sched_clock_idle_sleep_event(void); | |
2462 | extern void sched_clock_idle_wakeup_event(u64 delta_ns); | |
2c923e94 DL |
2463 | |
2464 | /* | |
2465 | * As outlined in clock.c, provides a fast, high resolution, nanosecond | |
2466 | * time source that is monotonic per cpu argument and has bounded drift | |
2467 | * between cpus. | |
2468 | * | |
2469 | * ######################### BIG FAT WARNING ########################## | |
2470 | * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can # | |
2471 | * # go backwards !! # | |
2472 | * #################################################################### | |
2473 | */ | |
2474 | static inline u64 cpu_clock(int cpu) | |
2475 | { | |
2476 | return sched_clock_cpu(cpu); | |
2477 | } | |
2478 | ||
2479 | static inline u64 local_clock(void) | |
2480 | { | |
2481 | return sched_clock_cpu(raw_smp_processor_id()); | |
2482 | } | |
3e51f33f PZ |
2483 | #endif |
2484 | ||
b52bfee4 VP |
2485 | #ifdef CONFIG_IRQ_TIME_ACCOUNTING |
2486 | /* | |
2487 | * An i/f to runtime opt-in for irq time accounting based off of sched_clock. | |
2488 | * The reason for this explicit opt-in is not to have perf penalty with | |
2489 | * slow sched_clocks. | |
2490 | */ | |
2491 | extern void enable_sched_clock_irqtime(void); | |
2492 | extern void disable_sched_clock_irqtime(void); | |
2493 | #else | |
2494 | static inline void enable_sched_clock_irqtime(void) {} | |
2495 | static inline void disable_sched_clock_irqtime(void) {} | |
2496 | #endif | |
2497 | ||
36c8b586 | 2498 | extern unsigned long long |
41b86e9c | 2499 | task_sched_runtime(struct task_struct *task); |
1da177e4 LT |
2500 | |
2501 | /* sched_exec is called by processes performing an exec */ | |
2502 | #ifdef CONFIG_SMP | |
2503 | extern void sched_exec(void); | |
2504 | #else | |
2505 | #define sched_exec() {} | |
2506 | #endif | |
2507 | ||
2aa44d05 IM |
2508 | extern void sched_clock_idle_sleep_event(void); |
2509 | extern void sched_clock_idle_wakeup_event(u64 delta_ns); | |
bb29ab26 | 2510 | |
1da177e4 LT |
2511 | #ifdef CONFIG_HOTPLUG_CPU |
2512 | extern void idle_task_exit(void); | |
2513 | #else | |
2514 | static inline void idle_task_exit(void) {} | |
2515 | #endif | |
2516 | ||
3451d024 | 2517 | #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP) |
1c20091e | 2518 | extern void wake_up_nohz_cpu(int cpu); |
06d8308c | 2519 | #else |
1c20091e | 2520 | static inline void wake_up_nohz_cpu(int cpu) { } |
06d8308c TG |
2521 | #endif |
2522 | ||
ce831b38 | 2523 | #ifdef CONFIG_NO_HZ_FULL |
265f22a9 | 2524 | extern u64 scheduler_tick_max_deferment(void); |
06d8308c TG |
2525 | #endif |
2526 | ||
5091faa4 | 2527 | #ifdef CONFIG_SCHED_AUTOGROUP |
5091faa4 MG |
2528 | extern void sched_autogroup_create_attach(struct task_struct *p); |
2529 | extern void sched_autogroup_detach(struct task_struct *p); | |
2530 | extern void sched_autogroup_fork(struct signal_struct *sig); | |
2531 | extern void sched_autogroup_exit(struct signal_struct *sig); | |
2532 | #ifdef CONFIG_PROC_FS | |
2533 | extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m); | |
2e5b5b3a | 2534 | extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice); |
5091faa4 MG |
2535 | #endif |
2536 | #else | |
2537 | static inline void sched_autogroup_create_attach(struct task_struct *p) { } | |
2538 | static inline void sched_autogroup_detach(struct task_struct *p) { } | |
2539 | static inline void sched_autogroup_fork(struct signal_struct *sig) { } | |
2540 | static inline void sched_autogroup_exit(struct signal_struct *sig) { } | |
2541 | #endif | |
2542 | ||
fa93384f | 2543 | extern int yield_to(struct task_struct *p, bool preempt); |
36c8b586 IM |
2544 | extern void set_user_nice(struct task_struct *p, long nice); |
2545 | extern int task_prio(const struct task_struct *p); | |
d0ea0268 DY |
2546 | /** |
2547 | * task_nice - return the nice value of a given task. | |
2548 | * @p: the task in question. | |
2549 | * | |
2550 | * Return: The nice value [ -20 ... 0 ... 19 ]. | |
2551 | */ | |
2552 | static inline int task_nice(const struct task_struct *p) | |
2553 | { | |
2554 | return PRIO_TO_NICE((p)->static_prio); | |
2555 | } | |
36c8b586 IM |
2556 | extern int can_nice(const struct task_struct *p, const int nice); |
2557 | extern int task_curr(const struct task_struct *p); | |
1da177e4 | 2558 | extern int idle_cpu(int cpu); |
fe7de49f KM |
2559 | extern int sched_setscheduler(struct task_struct *, int, |
2560 | const struct sched_param *); | |
961ccddd | 2561 | extern int sched_setscheduler_nocheck(struct task_struct *, int, |
fe7de49f | 2562 | const struct sched_param *); |
d50dde5a DF |
2563 | extern int sched_setattr(struct task_struct *, |
2564 | const struct sched_attr *); | |
36c8b586 | 2565 | extern struct task_struct *idle_task(int cpu); |
c4f30608 PM |
2566 | /** |
2567 | * is_idle_task - is the specified task an idle task? | |
fa757281 | 2568 | * @p: the task in question. |
e69f6186 YB |
2569 | * |
2570 | * Return: 1 if @p is an idle task. 0 otherwise. | |
c4f30608 | 2571 | */ |
7061ca3b | 2572 | static inline bool is_idle_task(const struct task_struct *p) |
c4f30608 PM |
2573 | { |
2574 | return p->pid == 0; | |
2575 | } | |
36c8b586 IM |
2576 | extern struct task_struct *curr_task(int cpu); |
2577 | extern void set_curr_task(int cpu, struct task_struct *p); | |
1da177e4 LT |
2578 | |
2579 | void yield(void); | |
2580 | ||
1da177e4 LT |
2581 | union thread_union { |
2582 | struct thread_info thread_info; | |
2583 | unsigned long stack[THREAD_SIZE/sizeof(long)]; | |
2584 | }; | |
2585 | ||
2586 | #ifndef __HAVE_ARCH_KSTACK_END | |
2587 | static inline int kstack_end(void *addr) | |
2588 | { | |
2589 | /* Reliable end of stack detection: | |
2590 | * Some APM bios versions misalign the stack | |
2591 | */ | |
2592 | return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); | |
2593 | } | |
2594 | #endif | |
2595 | ||
2596 | extern union thread_union init_thread_union; | |
2597 | extern struct task_struct init_task; | |
2598 | ||
2599 | extern struct mm_struct init_mm; | |
2600 | ||
198fe21b PE |
2601 | extern struct pid_namespace init_pid_ns; |
2602 | ||
2603 | /* | |
2604 | * find a task by one of its numerical ids | |
2605 | * | |
198fe21b PE |
2606 | * find_task_by_pid_ns(): |
2607 | * finds a task by its pid in the specified namespace | |
228ebcbe PE |
2608 | * find_task_by_vpid(): |
2609 | * finds a task by its virtual pid | |
198fe21b | 2610 | * |
e49859e7 | 2611 | * see also find_vpid() etc in include/linux/pid.h |
198fe21b PE |
2612 | */ |
2613 | ||
228ebcbe PE |
2614 | extern struct task_struct *find_task_by_vpid(pid_t nr); |
2615 | extern struct task_struct *find_task_by_pid_ns(pid_t nr, | |
2616 | struct pid_namespace *ns); | |
198fe21b | 2617 | |
1da177e4 | 2618 | /* per-UID process charging. */ |
7b44ab97 | 2619 | extern struct user_struct * alloc_uid(kuid_t); |
1da177e4 LT |
2620 | static inline struct user_struct *get_uid(struct user_struct *u) |
2621 | { | |
2622 | atomic_inc(&u->__count); | |
2623 | return u; | |
2624 | } | |
2625 | extern void free_uid(struct user_struct *); | |
1da177e4 LT |
2626 | |
2627 | #include <asm/current.h> | |
2628 | ||
f0af911a | 2629 | extern void xtime_update(unsigned long ticks); |
1da177e4 | 2630 | |
b3c97528 HH |
2631 | extern int wake_up_state(struct task_struct *tsk, unsigned int state); |
2632 | extern int wake_up_process(struct task_struct *tsk); | |
3e51e3ed | 2633 | extern void wake_up_new_task(struct task_struct *tsk); |
1da177e4 LT |
2634 | #ifdef CONFIG_SMP |
2635 | extern void kick_process(struct task_struct *tsk); | |
2636 | #else | |
2637 | static inline void kick_process(struct task_struct *tsk) { } | |
2638 | #endif | |
aab03e05 | 2639 | extern int sched_fork(unsigned long clone_flags, struct task_struct *p); |
ad46c2c4 | 2640 | extern void sched_dead(struct task_struct *p); |
1da177e4 | 2641 | |
1da177e4 LT |
2642 | extern void proc_caches_init(void); |
2643 | extern void flush_signals(struct task_struct *); | |
10ab825b | 2644 | extern void ignore_signals(struct task_struct *); |
1da177e4 LT |
2645 | extern void flush_signal_handlers(struct task_struct *, int force_default); |
2646 | extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); | |
2647 | ||
be0e6f29 | 2648 | static inline int kernel_dequeue_signal(siginfo_t *info) |
1da177e4 | 2649 | { |
be0e6f29 ON |
2650 | struct task_struct *tsk = current; |
2651 | siginfo_t __info; | |
1da177e4 LT |
2652 | int ret; |
2653 | ||
be0e6f29 ON |
2654 | spin_lock_irq(&tsk->sighand->siglock); |
2655 | ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info); | |
2656 | spin_unlock_irq(&tsk->sighand->siglock); | |
1da177e4 LT |
2657 | |
2658 | return ret; | |
53c8f9f1 | 2659 | } |
1da177e4 | 2660 | |
9a13049e ON |
2661 | static inline void kernel_signal_stop(void) |
2662 | { | |
2663 | spin_lock_irq(¤t->sighand->siglock); | |
2664 | if (current->jobctl & JOBCTL_STOP_DEQUEUED) | |
2665 | __set_current_state(TASK_STOPPED); | |
2666 | spin_unlock_irq(¤t->sighand->siglock); | |
2667 | ||
2668 | schedule(); | |
2669 | } | |
2670 | ||
1da177e4 LT |
2671 | extern void release_task(struct task_struct * p); |
2672 | extern int send_sig_info(int, struct siginfo *, struct task_struct *); | |
1da177e4 LT |
2673 | extern int force_sigsegv(int, struct task_struct *); |
2674 | extern int force_sig_info(int, struct siginfo *, struct task_struct *); | |
c4b92fc1 | 2675 | extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); |
c4b92fc1 | 2676 | extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); |
d178bc3a SH |
2677 | extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *, |
2678 | const struct cred *, u32); | |
c4b92fc1 EB |
2679 | extern int kill_pgrp(struct pid *pid, int sig, int priv); |
2680 | extern int kill_pid(struct pid *pid, int sig, int priv); | |
c3de4b38 | 2681 | extern int kill_proc_info(int, struct siginfo *, pid_t); |
86773473 | 2682 | extern __must_check bool do_notify_parent(struct task_struct *, int); |
a7f0765e | 2683 | extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent); |
1da177e4 | 2684 | extern void force_sig(int, struct task_struct *); |
1da177e4 | 2685 | extern int send_sig(int, struct task_struct *, int); |
09faef11 | 2686 | extern int zap_other_threads(struct task_struct *p); |
1da177e4 LT |
2687 | extern struct sigqueue *sigqueue_alloc(void); |
2688 | extern void sigqueue_free(struct sigqueue *); | |
ac5c2153 | 2689 | extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); |
9ac95f2f | 2690 | extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); |
1da177e4 | 2691 | |
7e781418 AL |
2692 | #ifdef TIF_RESTORE_SIGMASK |
2693 | /* | |
2694 | * Legacy restore_sigmask accessors. These are inefficient on | |
2695 | * SMP architectures because they require atomic operations. | |
2696 | */ | |
2697 | ||
2698 | /** | |
2699 | * set_restore_sigmask() - make sure saved_sigmask processing gets done | |
2700 | * | |
2701 | * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code | |
2702 | * will run before returning to user mode, to process the flag. For | |
2703 | * all callers, TIF_SIGPENDING is already set or it's no harm to set | |
2704 | * it. TIF_RESTORE_SIGMASK need not be in the set of bits that the | |
2705 | * arch code will notice on return to user mode, in case those bits | |
2706 | * are scarce. We set TIF_SIGPENDING here to ensure that the arch | |
2707 | * signal code always gets run when TIF_RESTORE_SIGMASK is set. | |
2708 | */ | |
2709 | static inline void set_restore_sigmask(void) | |
2710 | { | |
2711 | set_thread_flag(TIF_RESTORE_SIGMASK); | |
2712 | WARN_ON(!test_thread_flag(TIF_SIGPENDING)); | |
2713 | } | |
2714 | static inline void clear_restore_sigmask(void) | |
2715 | { | |
2716 | clear_thread_flag(TIF_RESTORE_SIGMASK); | |
2717 | } | |
2718 | static inline bool test_restore_sigmask(void) | |
2719 | { | |
2720 | return test_thread_flag(TIF_RESTORE_SIGMASK); | |
2721 | } | |
2722 | static inline bool test_and_clear_restore_sigmask(void) | |
2723 | { | |
2724 | return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK); | |
2725 | } | |
2726 | ||
2727 | #else /* TIF_RESTORE_SIGMASK */ | |
2728 | ||
2729 | /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */ | |
2730 | static inline void set_restore_sigmask(void) | |
2731 | { | |
2732 | current->restore_sigmask = true; | |
2733 | WARN_ON(!test_thread_flag(TIF_SIGPENDING)); | |
2734 | } | |
2735 | static inline void clear_restore_sigmask(void) | |
2736 | { | |
2737 | current->restore_sigmask = false; | |
2738 | } | |
2739 | static inline bool test_restore_sigmask(void) | |
2740 | { | |
2741 | return current->restore_sigmask; | |
2742 | } | |
2743 | static inline bool test_and_clear_restore_sigmask(void) | |
2744 | { | |
2745 | if (!current->restore_sigmask) | |
2746 | return false; | |
2747 | current->restore_sigmask = false; | |
2748 | return true; | |
2749 | } | |
2750 | #endif | |
2751 | ||
51a7b448 AV |
2752 | static inline void restore_saved_sigmask(void) |
2753 | { | |
2754 | if (test_and_clear_restore_sigmask()) | |
77097ae5 | 2755 | __set_current_blocked(¤t->saved_sigmask); |
51a7b448 AV |
2756 | } |
2757 | ||
b7f9a11a AV |
2758 | static inline sigset_t *sigmask_to_save(void) |
2759 | { | |
2760 | sigset_t *res = ¤t->blocked; | |
2761 | if (unlikely(test_restore_sigmask())) | |
2762 | res = ¤t->saved_sigmask; | |
2763 | return res; | |
2764 | } | |
2765 | ||
9ec52099 CLG |
2766 | static inline int kill_cad_pid(int sig, int priv) |
2767 | { | |
2768 | return kill_pid(cad_pid, sig, priv); | |
2769 | } | |
2770 | ||
1da177e4 LT |
2771 | /* These can be the second arg to send_sig_info/send_group_sig_info. */ |
2772 | #define SEND_SIG_NOINFO ((struct siginfo *) 0) | |
2773 | #define SEND_SIG_PRIV ((struct siginfo *) 1) | |
2774 | #define SEND_SIG_FORCED ((struct siginfo *) 2) | |
2775 | ||
2a855dd0 SAS |
2776 | /* |
2777 | * True if we are on the alternate signal stack. | |
2778 | */ | |
1da177e4 LT |
2779 | static inline int on_sig_stack(unsigned long sp) |
2780 | { | |
c876eeab AL |
2781 | /* |
2782 | * If the signal stack is SS_AUTODISARM then, by construction, we | |
2783 | * can't be on the signal stack unless user code deliberately set | |
2784 | * SS_AUTODISARM when we were already on it. | |
2785 | * | |
2786 | * This improves reliability: if user state gets corrupted such that | |
2787 | * the stack pointer points very close to the end of the signal stack, | |
2788 | * then this check will enable the signal to be handled anyway. | |
2789 | */ | |
2790 | if (current->sas_ss_flags & SS_AUTODISARM) | |
2791 | return 0; | |
2792 | ||
2a855dd0 SAS |
2793 | #ifdef CONFIG_STACK_GROWSUP |
2794 | return sp >= current->sas_ss_sp && | |
2795 | sp - current->sas_ss_sp < current->sas_ss_size; | |
2796 | #else | |
2797 | return sp > current->sas_ss_sp && | |
2798 | sp - current->sas_ss_sp <= current->sas_ss_size; | |
2799 | #endif | |
1da177e4 LT |
2800 | } |
2801 | ||
2802 | static inline int sas_ss_flags(unsigned long sp) | |
2803 | { | |
72f15c03 RW |
2804 | if (!current->sas_ss_size) |
2805 | return SS_DISABLE; | |
2806 | ||
2807 | return on_sig_stack(sp) ? SS_ONSTACK : 0; | |
1da177e4 LT |
2808 | } |
2809 | ||
2a742138 SS |
2810 | static inline void sas_ss_reset(struct task_struct *p) |
2811 | { | |
2812 | p->sas_ss_sp = 0; | |
2813 | p->sas_ss_size = 0; | |
2814 | p->sas_ss_flags = SS_DISABLE; | |
2815 | } | |
2816 | ||
5a1b98d3 AV |
2817 | static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig) |
2818 | { | |
2819 | if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp)) | |
2820 | #ifdef CONFIG_STACK_GROWSUP | |
2821 | return current->sas_ss_sp; | |
2822 | #else | |
2823 | return current->sas_ss_sp + current->sas_ss_size; | |
2824 | #endif | |
2825 | return sp; | |
2826 | } | |
2827 | ||
1da177e4 LT |
2828 | /* |
2829 | * Routines for handling mm_structs | |
2830 | */ | |
2831 | extern struct mm_struct * mm_alloc(void); | |
2832 | ||
2833 | /* mmdrop drops the mm and the page tables */ | |
b3c97528 | 2834 | extern void __mmdrop(struct mm_struct *); |
d2005e3f | 2835 | static inline void mmdrop(struct mm_struct *mm) |
1da177e4 | 2836 | { |
6fb43d7b | 2837 | if (unlikely(atomic_dec_and_test(&mm->mm_count))) |
1da177e4 LT |
2838 | __mmdrop(mm); |
2839 | } | |
2840 | ||
d2005e3f ON |
2841 | static inline bool mmget_not_zero(struct mm_struct *mm) |
2842 | { | |
2843 | return atomic_inc_not_zero(&mm->mm_users); | |
2844 | } | |
2845 | ||
1da177e4 LT |
2846 | /* mmput gets rid of the mappings and all user-space */ |
2847 | extern void mmput(struct mm_struct *); | |
7ef949d7 MH |
2848 | #ifdef CONFIG_MMU |
2849 | /* same as above but performs the slow path from the async context. Can | |
ec8d7c14 MH |
2850 | * be called from the atomic context as well |
2851 | */ | |
2852 | extern void mmput_async(struct mm_struct *); | |
7ef949d7 | 2853 | #endif |
ec8d7c14 | 2854 | |
1da177e4 LT |
2855 | /* Grab a reference to a task's mm, if it is not already going away */ |
2856 | extern struct mm_struct *get_task_mm(struct task_struct *task); | |
8cdb878d CY |
2857 | /* |
2858 | * Grab a reference to a task's mm, if it is not already going away | |
2859 | * and ptrace_may_access with the mode parameter passed to it | |
2860 | * succeeds. | |
2861 | */ | |
2862 | extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode); | |
1da177e4 LT |
2863 | /* Remove the current tasks stale references to the old mm_struct */ |
2864 | extern void mm_release(struct task_struct *, struct mm_struct *); | |
2865 | ||
3033f14a JT |
2866 | #ifdef CONFIG_HAVE_COPY_THREAD_TLS |
2867 | extern int copy_thread_tls(unsigned long, unsigned long, unsigned long, | |
2868 | struct task_struct *, unsigned long); | |
2869 | #else | |
6f2c55b8 | 2870 | extern int copy_thread(unsigned long, unsigned long, unsigned long, |
afa86fc4 | 2871 | struct task_struct *); |
3033f14a JT |
2872 | |
2873 | /* Architectures that haven't opted into copy_thread_tls get the tls argument | |
2874 | * via pt_regs, so ignore the tls argument passed via C. */ | |
2875 | static inline int copy_thread_tls( | |
2876 | unsigned long clone_flags, unsigned long sp, unsigned long arg, | |
2877 | struct task_struct *p, unsigned long tls) | |
2878 | { | |
2879 | return copy_thread(clone_flags, sp, arg, p); | |
2880 | } | |
2881 | #endif | |
1da177e4 | 2882 | extern void flush_thread(void); |
5f56a5df JS |
2883 | |
2884 | #ifdef CONFIG_HAVE_EXIT_THREAD | |
e6464694 | 2885 | extern void exit_thread(struct task_struct *tsk); |
5f56a5df | 2886 | #else |
e6464694 | 2887 | static inline void exit_thread(struct task_struct *tsk) |
5f56a5df JS |
2888 | { |
2889 | } | |
2890 | #endif | |
1da177e4 | 2891 | |
1da177e4 | 2892 | extern void exit_files(struct task_struct *); |
a7e5328a | 2893 | extern void __cleanup_sighand(struct sighand_struct *); |
cbaffba1 | 2894 | |
1da177e4 | 2895 | extern void exit_itimers(struct signal_struct *); |
cbaffba1 | 2896 | extern void flush_itimer_signals(void); |
1da177e4 | 2897 | |
9402c95f | 2898 | extern void do_group_exit(int); |
1da177e4 | 2899 | |
c4ad8f98 | 2900 | extern int do_execve(struct filename *, |
d7627467 | 2901 | const char __user * const __user *, |
da3d4c5f | 2902 | const char __user * const __user *); |
51f39a1f DD |
2903 | extern int do_execveat(int, struct filename *, |
2904 | const char __user * const __user *, | |
2905 | const char __user * const __user *, | |
2906 | int); | |
3033f14a | 2907 | extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long); |
e80d6661 | 2908 | extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *); |
36c8b586 | 2909 | struct task_struct *fork_idle(int); |
2aa3a7f8 | 2910 | extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags); |
1da177e4 | 2911 | |
82b89778 AH |
2912 | extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec); |
2913 | static inline void set_task_comm(struct task_struct *tsk, const char *from) | |
2914 | { | |
2915 | __set_task_comm(tsk, from, false); | |
2916 | } | |
59714d65 | 2917 | extern char *get_task_comm(char *to, struct task_struct *tsk); |
1da177e4 LT |
2918 | |
2919 | #ifdef CONFIG_SMP | |
317f3941 | 2920 | void scheduler_ipi(void); |
85ba2d86 | 2921 | extern unsigned long wait_task_inactive(struct task_struct *, long match_state); |
1da177e4 | 2922 | #else |
184748cc | 2923 | static inline void scheduler_ipi(void) { } |
85ba2d86 RM |
2924 | static inline unsigned long wait_task_inactive(struct task_struct *p, |
2925 | long match_state) | |
2926 | { | |
2927 | return 1; | |
2928 | } | |
1da177e4 LT |
2929 | #endif |
2930 | ||
fafe870f FW |
2931 | #define tasklist_empty() \ |
2932 | list_empty(&init_task.tasks) | |
2933 | ||
05725f7e JP |
2934 | #define next_task(p) \ |
2935 | list_entry_rcu((p)->tasks.next, struct task_struct, tasks) | |
1da177e4 LT |
2936 | |
2937 | #define for_each_process(p) \ | |
2938 | for (p = &init_task ; (p = next_task(p)) != &init_task ; ) | |
2939 | ||
5bb459bb | 2940 | extern bool current_is_single_threaded(void); |
d84f4f99 | 2941 | |
1da177e4 LT |
2942 | /* |
2943 | * Careful: do_each_thread/while_each_thread is a double loop so | |
2944 | * 'break' will not work as expected - use goto instead. | |
2945 | */ | |
2946 | #define do_each_thread(g, t) \ | |
2947 | for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do | |
2948 | ||
2949 | #define while_each_thread(g, t) \ | |
2950 | while ((t = next_thread(t)) != g) | |
2951 | ||
0c740d0a ON |
2952 | #define __for_each_thread(signal, t) \ |
2953 | list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node) | |
2954 | ||
2955 | #define for_each_thread(p, t) \ | |
2956 | __for_each_thread((p)->signal, t) | |
2957 | ||
2958 | /* Careful: this is a double loop, 'break' won't work as expected. */ | |
2959 | #define for_each_process_thread(p, t) \ | |
2960 | for_each_process(p) for_each_thread(p, t) | |
2961 | ||
7e49827c ON |
2962 | static inline int get_nr_threads(struct task_struct *tsk) |
2963 | { | |
b3ac022c | 2964 | return tsk->signal->nr_threads; |
7e49827c ON |
2965 | } |
2966 | ||
087806b1 ON |
2967 | static inline bool thread_group_leader(struct task_struct *p) |
2968 | { | |
2969 | return p->exit_signal >= 0; | |
2970 | } | |
1da177e4 | 2971 | |
0804ef4b EB |
2972 | /* Do to the insanities of de_thread it is possible for a process |
2973 | * to have the pid of the thread group leader without actually being | |
2974 | * the thread group leader. For iteration through the pids in proc | |
2975 | * all we care about is that we have a task with the appropriate | |
2976 | * pid, we don't actually care if we have the right task. | |
2977 | */ | |
e1403b8e | 2978 | static inline bool has_group_leader_pid(struct task_struct *p) |
0804ef4b | 2979 | { |
e1403b8e | 2980 | return task_pid(p) == p->signal->leader_pid; |
0804ef4b EB |
2981 | } |
2982 | ||
bac0abd6 | 2983 | static inline |
e1403b8e | 2984 | bool same_thread_group(struct task_struct *p1, struct task_struct *p2) |
bac0abd6 | 2985 | { |
e1403b8e | 2986 | return p1->signal == p2->signal; |
bac0abd6 PE |
2987 | } |
2988 | ||
36c8b586 | 2989 | static inline struct task_struct *next_thread(const struct task_struct *p) |
47e65328 | 2990 | { |
05725f7e JP |
2991 | return list_entry_rcu(p->thread_group.next, |
2992 | struct task_struct, thread_group); | |
47e65328 ON |
2993 | } |
2994 | ||
e868171a | 2995 | static inline int thread_group_empty(struct task_struct *p) |
1da177e4 | 2996 | { |
47e65328 | 2997 | return list_empty(&p->thread_group); |
1da177e4 LT |
2998 | } |
2999 | ||
3000 | #define delay_group_leader(p) \ | |
3001 | (thread_group_leader(p) && !thread_group_empty(p)) | |
3002 | ||
1da177e4 | 3003 | /* |
260ea101 | 3004 | * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring |
22e2c507 | 3005 | * subscriptions and synchronises with wait4(). Also used in procfs. Also |
ddbcc7e8 | 3006 | * pins the final release of task.io_context. Also protects ->cpuset and |
d68b46fe | 3007 | * ->cgroup.subsys[]. And ->vfork_done. |
1da177e4 LT |
3008 | * |
3009 | * Nests both inside and outside of read_lock(&tasklist_lock). | |
3010 | * It must not be nested with write_lock_irq(&tasklist_lock), | |
3011 | * neither inside nor outside. | |
3012 | */ | |
3013 | static inline void task_lock(struct task_struct *p) | |
3014 | { | |
3015 | spin_lock(&p->alloc_lock); | |
3016 | } | |
3017 | ||
3018 | static inline void task_unlock(struct task_struct *p) | |
3019 | { | |
3020 | spin_unlock(&p->alloc_lock); | |
3021 | } | |
3022 | ||
b8ed374e | 3023 | extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk, |
f63ee72e ON |
3024 | unsigned long *flags); |
3025 | ||
9388dc30 AV |
3026 | static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk, |
3027 | unsigned long *flags) | |
3028 | { | |
3029 | struct sighand_struct *ret; | |
3030 | ||
3031 | ret = __lock_task_sighand(tsk, flags); | |
3032 | (void)__cond_lock(&tsk->sighand->siglock, ret); | |
3033 | return ret; | |
3034 | } | |
b8ed374e | 3035 | |
f63ee72e ON |
3036 | static inline void unlock_task_sighand(struct task_struct *tsk, |
3037 | unsigned long *flags) | |
3038 | { | |
3039 | spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); | |
3040 | } | |
3041 | ||
77e4ef99 | 3042 | /** |
7d7efec3 TH |
3043 | * threadgroup_change_begin - mark the beginning of changes to a threadgroup |
3044 | * @tsk: task causing the changes | |
77e4ef99 | 3045 | * |
7d7efec3 TH |
3046 | * All operations which modify a threadgroup - a new thread joining the |
3047 | * group, death of a member thread (the assertion of PF_EXITING) and | |
3048 | * exec(2) dethreading the process and replacing the leader - are wrapped | |
3049 | * by threadgroup_change_{begin|end}(). This is to provide a place which | |
3050 | * subsystems needing threadgroup stability can hook into for | |
3051 | * synchronization. | |
77e4ef99 | 3052 | */ |
7d7efec3 | 3053 | static inline void threadgroup_change_begin(struct task_struct *tsk) |
4714d1d3 | 3054 | { |
7d7efec3 TH |
3055 | might_sleep(); |
3056 | cgroup_threadgroup_change_begin(tsk); | |
4714d1d3 | 3057 | } |
77e4ef99 TH |
3058 | |
3059 | /** | |
7d7efec3 TH |
3060 | * threadgroup_change_end - mark the end of changes to a threadgroup |
3061 | * @tsk: task causing the changes | |
77e4ef99 | 3062 | * |
7d7efec3 | 3063 | * See threadgroup_change_begin(). |
77e4ef99 | 3064 | */ |
7d7efec3 | 3065 | static inline void threadgroup_change_end(struct task_struct *tsk) |
4714d1d3 | 3066 | { |
7d7efec3 | 3067 | cgroup_threadgroup_change_end(tsk); |
4714d1d3 | 3068 | } |
4714d1d3 | 3069 | |
f037360f AV |
3070 | #ifndef __HAVE_THREAD_FUNCTIONS |
3071 | ||
f7e4217b RZ |
3072 | #define task_thread_info(task) ((struct thread_info *)(task)->stack) |
3073 | #define task_stack_page(task) ((task)->stack) | |
a1261f54 | 3074 | |
10ebffde AV |
3075 | static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) |
3076 | { | |
3077 | *task_thread_info(p) = *task_thread_info(org); | |
3078 | task_thread_info(p)->task = p; | |
3079 | } | |
3080 | ||
6a40281a CE |
3081 | /* |
3082 | * Return the address of the last usable long on the stack. | |
3083 | * | |
3084 | * When the stack grows down, this is just above the thread | |
3085 | * info struct. Going any lower will corrupt the threadinfo. | |
3086 | * | |
3087 | * When the stack grows up, this is the highest address. | |
3088 | * Beyond that position, we corrupt data on the next page. | |
3089 | */ | |
10ebffde AV |
3090 | static inline unsigned long *end_of_stack(struct task_struct *p) |
3091 | { | |
6a40281a CE |
3092 | #ifdef CONFIG_STACK_GROWSUP |
3093 | return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1; | |
3094 | #else | |
f7e4217b | 3095 | return (unsigned long *)(task_thread_info(p) + 1); |
6a40281a | 3096 | #endif |
10ebffde AV |
3097 | } |
3098 | ||
f037360f | 3099 | #endif |
a70857e4 AT |
3100 | #define task_stack_end_corrupted(task) \ |
3101 | (*(end_of_stack(task)) != STACK_END_MAGIC) | |
f037360f | 3102 | |
8b05c7e6 FT |
3103 | static inline int object_is_on_stack(void *obj) |
3104 | { | |
3105 | void *stack = task_stack_page(current); | |
3106 | ||
3107 | return (obj >= stack) && (obj < (stack + THREAD_SIZE)); | |
3108 | } | |
3109 | ||
b235beea | 3110 | extern void thread_stack_cache_init(void); |
8c9843e5 | 3111 | |
7c9f8861 ES |
3112 | #ifdef CONFIG_DEBUG_STACK_USAGE |
3113 | static inline unsigned long stack_not_used(struct task_struct *p) | |
3114 | { | |
3115 | unsigned long *n = end_of_stack(p); | |
3116 | ||
3117 | do { /* Skip over canary */ | |
6c31da34 HD |
3118 | # ifdef CONFIG_STACK_GROWSUP |
3119 | n--; | |
3120 | # else | |
7c9f8861 | 3121 | n++; |
6c31da34 | 3122 | # endif |
7c9f8861 ES |
3123 | } while (!*n); |
3124 | ||
6c31da34 HD |
3125 | # ifdef CONFIG_STACK_GROWSUP |
3126 | return (unsigned long)end_of_stack(p) - (unsigned long)n; | |
3127 | # else | |
7c9f8861 | 3128 | return (unsigned long)n - (unsigned long)end_of_stack(p); |
6c31da34 | 3129 | # endif |
7c9f8861 ES |
3130 | } |
3131 | #endif | |
d4311ff1 | 3132 | extern void set_task_stack_end_magic(struct task_struct *tsk); |
7c9f8861 | 3133 | |
1da177e4 LT |
3134 | /* set thread flags in other task's structures |
3135 | * - see asm/thread_info.h for TIF_xxxx flags available | |
3136 | */ | |
3137 | static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3138 | { | |
a1261f54 | 3139 | set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3140 | } |
3141 | ||
3142 | static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3143 | { | |
a1261f54 | 3144 | clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3145 | } |
3146 | ||
3147 | static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3148 | { | |
a1261f54 | 3149 | return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3150 | } |
3151 | ||
3152 | static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3153 | { | |
a1261f54 | 3154 | return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3155 | } |
3156 | ||
3157 | static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) | |
3158 | { | |
a1261f54 | 3159 | return test_ti_thread_flag(task_thread_info(tsk), flag); |
1da177e4 LT |
3160 | } |
3161 | ||
3162 | static inline void set_tsk_need_resched(struct task_struct *tsk) | |
3163 | { | |
3164 | set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
3165 | } | |
3166 | ||
3167 | static inline void clear_tsk_need_resched(struct task_struct *tsk) | |
3168 | { | |
3169 | clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); | |
3170 | } | |
3171 | ||
8ae121ac GH |
3172 | static inline int test_tsk_need_resched(struct task_struct *tsk) |
3173 | { | |
3174 | return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); | |
3175 | } | |
3176 | ||
690cc3ff EB |
3177 | static inline int restart_syscall(void) |
3178 | { | |
3179 | set_tsk_thread_flag(current, TIF_SIGPENDING); | |
3180 | return -ERESTARTNOINTR; | |
3181 | } | |
3182 | ||
1da177e4 LT |
3183 | static inline int signal_pending(struct task_struct *p) |
3184 | { | |
3185 | return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); | |
3186 | } | |
f776d12d | 3187 | |
d9588725 RM |
3188 | static inline int __fatal_signal_pending(struct task_struct *p) |
3189 | { | |
3190 | return unlikely(sigismember(&p->pending.signal, SIGKILL)); | |
3191 | } | |
f776d12d MW |
3192 | |
3193 | static inline int fatal_signal_pending(struct task_struct *p) | |
3194 | { | |
3195 | return signal_pending(p) && __fatal_signal_pending(p); | |
3196 | } | |
3197 | ||
16882c1e ON |
3198 | static inline int signal_pending_state(long state, struct task_struct *p) |
3199 | { | |
3200 | if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) | |
3201 | return 0; | |
3202 | if (!signal_pending(p)) | |
3203 | return 0; | |
3204 | ||
16882c1e ON |
3205 | return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); |
3206 | } | |
3207 | ||
1da177e4 LT |
3208 | /* |
3209 | * cond_resched() and cond_resched_lock(): latency reduction via | |
3210 | * explicit rescheduling in places that are safe. The return | |
3211 | * value indicates whether a reschedule was done in fact. | |
3212 | * cond_resched_lock() will drop the spinlock before scheduling, | |
3213 | * cond_resched_softirq() will enable bhs before scheduling. | |
3214 | */ | |
c3921ab7 | 3215 | extern int _cond_resched(void); |
6f80bd98 | 3216 | |
613afbf8 | 3217 | #define cond_resched() ({ \ |
3427445a | 3218 | ___might_sleep(__FILE__, __LINE__, 0); \ |
613afbf8 FW |
3219 | _cond_resched(); \ |
3220 | }) | |
6f80bd98 | 3221 | |
613afbf8 FW |
3222 | extern int __cond_resched_lock(spinlock_t *lock); |
3223 | ||
3224 | #define cond_resched_lock(lock) ({ \ | |
3427445a | 3225 | ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\ |
613afbf8 FW |
3226 | __cond_resched_lock(lock); \ |
3227 | }) | |
3228 | ||
3229 | extern int __cond_resched_softirq(void); | |
3230 | ||
75e1056f | 3231 | #define cond_resched_softirq() ({ \ |
3427445a | 3232 | ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \ |
75e1056f | 3233 | __cond_resched_softirq(); \ |
613afbf8 | 3234 | }) |
1da177e4 | 3235 | |
f6f3c437 SH |
3236 | static inline void cond_resched_rcu(void) |
3237 | { | |
3238 | #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU) | |
3239 | rcu_read_unlock(); | |
3240 | cond_resched(); | |
3241 | rcu_read_lock(); | |
3242 | #endif | |
3243 | } | |
3244 | ||
1da177e4 LT |
3245 | /* |
3246 | * Does a critical section need to be broken due to another | |
95c354fe NP |
3247 | * task waiting?: (technically does not depend on CONFIG_PREEMPT, |
3248 | * but a general need for low latency) | |
1da177e4 | 3249 | */ |
95c354fe | 3250 | static inline int spin_needbreak(spinlock_t *lock) |
1da177e4 | 3251 | { |
95c354fe NP |
3252 | #ifdef CONFIG_PREEMPT |
3253 | return spin_is_contended(lock); | |
3254 | #else | |
1da177e4 | 3255 | return 0; |
95c354fe | 3256 | #endif |
1da177e4 LT |
3257 | } |
3258 | ||
ee761f62 TG |
3259 | /* |
3260 | * Idle thread specific functions to determine the need_resched | |
69dd0f84 | 3261 | * polling state. |
ee761f62 | 3262 | */ |
69dd0f84 | 3263 | #ifdef TIF_POLLING_NRFLAG |
ee761f62 TG |
3264 | static inline int tsk_is_polling(struct task_struct *p) |
3265 | { | |
3266 | return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG); | |
3267 | } | |
ea811747 PZ |
3268 | |
3269 | static inline void __current_set_polling(void) | |
3a98f871 TG |
3270 | { |
3271 | set_thread_flag(TIF_POLLING_NRFLAG); | |
3272 | } | |
3273 | ||
ea811747 PZ |
3274 | static inline bool __must_check current_set_polling_and_test(void) |
3275 | { | |
3276 | __current_set_polling(); | |
3277 | ||
3278 | /* | |
3279 | * Polling state must be visible before we test NEED_RESCHED, | |
8875125e | 3280 | * paired by resched_curr() |
ea811747 | 3281 | */ |
4e857c58 | 3282 | smp_mb__after_atomic(); |
ea811747 PZ |
3283 | |
3284 | return unlikely(tif_need_resched()); | |
3285 | } | |
3286 | ||
3287 | static inline void __current_clr_polling(void) | |
3a98f871 TG |
3288 | { |
3289 | clear_thread_flag(TIF_POLLING_NRFLAG); | |
3290 | } | |
ea811747 PZ |
3291 | |
3292 | static inline bool __must_check current_clr_polling_and_test(void) | |
3293 | { | |
3294 | __current_clr_polling(); | |
3295 | ||
3296 | /* | |
3297 | * Polling state must be visible before we test NEED_RESCHED, | |
8875125e | 3298 | * paired by resched_curr() |
ea811747 | 3299 | */ |
4e857c58 | 3300 | smp_mb__after_atomic(); |
ea811747 PZ |
3301 | |
3302 | return unlikely(tif_need_resched()); | |
3303 | } | |
3304 | ||
ee761f62 TG |
3305 | #else |
3306 | static inline int tsk_is_polling(struct task_struct *p) { return 0; } | |
ea811747 PZ |
3307 | static inline void __current_set_polling(void) { } |
3308 | static inline void __current_clr_polling(void) { } | |
3309 | ||
3310 | static inline bool __must_check current_set_polling_and_test(void) | |
3311 | { | |
3312 | return unlikely(tif_need_resched()); | |
3313 | } | |
3314 | static inline bool __must_check current_clr_polling_and_test(void) | |
3315 | { | |
3316 | return unlikely(tif_need_resched()); | |
3317 | } | |
ee761f62 TG |
3318 | #endif |
3319 | ||
8cb75e0c PZ |
3320 | static inline void current_clr_polling(void) |
3321 | { | |
3322 | __current_clr_polling(); | |
3323 | ||
3324 | /* | |
3325 | * Ensure we check TIF_NEED_RESCHED after we clear the polling bit. | |
3326 | * Once the bit is cleared, we'll get IPIs with every new | |
3327 | * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also | |
3328 | * fold. | |
3329 | */ | |
8875125e | 3330 | smp_mb(); /* paired with resched_curr() */ |
8cb75e0c PZ |
3331 | |
3332 | preempt_fold_need_resched(); | |
3333 | } | |
3334 | ||
75f93fed PZ |
3335 | static __always_inline bool need_resched(void) |
3336 | { | |
3337 | return unlikely(tif_need_resched()); | |
3338 | } | |
3339 | ||
f06febc9 FM |
3340 | /* |
3341 | * Thread group CPU time accounting. | |
3342 | */ | |
4cd4c1b4 | 3343 | void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); |
4da94d49 | 3344 | void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); |
f06febc9 | 3345 | |
7bb44ade RM |
3346 | /* |
3347 | * Reevaluate whether the task has signals pending delivery. | |
3348 | * Wake the task if so. | |
3349 | * This is required every time the blocked sigset_t changes. | |
3350 | * callers must hold sighand->siglock. | |
3351 | */ | |
3352 | extern void recalc_sigpending_and_wake(struct task_struct *t); | |
1da177e4 LT |
3353 | extern void recalc_sigpending(void); |
3354 | ||
910ffdb1 ON |
3355 | extern void signal_wake_up_state(struct task_struct *t, unsigned int state); |
3356 | ||
3357 | static inline void signal_wake_up(struct task_struct *t, bool resume) | |
3358 | { | |
3359 | signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0); | |
3360 | } | |
3361 | static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume) | |
3362 | { | |
3363 | signal_wake_up_state(t, resume ? __TASK_TRACED : 0); | |
3364 | } | |
1da177e4 LT |
3365 | |
3366 | /* | |
3367 | * Wrappers for p->thread_info->cpu access. No-op on UP. | |
3368 | */ | |
3369 | #ifdef CONFIG_SMP | |
3370 | ||
3371 | static inline unsigned int task_cpu(const struct task_struct *p) | |
3372 | { | |
a1261f54 | 3373 | return task_thread_info(p)->cpu; |
1da177e4 LT |
3374 | } |
3375 | ||
b32e86b4 IM |
3376 | static inline int task_node(const struct task_struct *p) |
3377 | { | |
3378 | return cpu_to_node(task_cpu(p)); | |
3379 | } | |
3380 | ||
c65cc870 | 3381 | extern void set_task_cpu(struct task_struct *p, unsigned int cpu); |
1da177e4 LT |
3382 | |
3383 | #else | |
3384 | ||
3385 | static inline unsigned int task_cpu(const struct task_struct *p) | |
3386 | { | |
3387 | return 0; | |
3388 | } | |
3389 | ||
3390 | static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) | |
3391 | { | |
3392 | } | |
3393 | ||
3394 | #endif /* CONFIG_SMP */ | |
3395 | ||
96f874e2 RR |
3396 | extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); |
3397 | extern long sched_getaffinity(pid_t pid, struct cpumask *mask); | |
5c45bf27 | 3398 | |
7c941438 | 3399 | #ifdef CONFIG_CGROUP_SCHED |
07e06b01 | 3400 | extern struct task_group root_task_group; |
8323f26c | 3401 | #endif /* CONFIG_CGROUP_SCHED */ |
9b5b7751 | 3402 | |
54e99124 DG |
3403 | extern int task_can_switch_user(struct user_struct *up, |
3404 | struct task_struct *tsk); | |
3405 | ||
4b98d11b AD |
3406 | #ifdef CONFIG_TASK_XACCT |
3407 | static inline void add_rchar(struct task_struct *tsk, ssize_t amt) | |
3408 | { | |
940389b8 | 3409 | tsk->ioac.rchar += amt; |
4b98d11b AD |
3410 | } |
3411 | ||
3412 | static inline void add_wchar(struct task_struct *tsk, ssize_t amt) | |
3413 | { | |
940389b8 | 3414 | tsk->ioac.wchar += amt; |
4b98d11b AD |
3415 | } |
3416 | ||
3417 | static inline void inc_syscr(struct task_struct *tsk) | |
3418 | { | |
940389b8 | 3419 | tsk->ioac.syscr++; |
4b98d11b AD |
3420 | } |
3421 | ||
3422 | static inline void inc_syscw(struct task_struct *tsk) | |
3423 | { | |
940389b8 | 3424 | tsk->ioac.syscw++; |
4b98d11b AD |
3425 | } |
3426 | #else | |
3427 | static inline void add_rchar(struct task_struct *tsk, ssize_t amt) | |
3428 | { | |
3429 | } | |
3430 | ||
3431 | static inline void add_wchar(struct task_struct *tsk, ssize_t amt) | |
3432 | { | |
3433 | } | |
3434 | ||
3435 | static inline void inc_syscr(struct task_struct *tsk) | |
3436 | { | |
3437 | } | |
3438 | ||
3439 | static inline void inc_syscw(struct task_struct *tsk) | |
3440 | { | |
3441 | } | |
3442 | #endif | |
3443 | ||
82455257 DH |
3444 | #ifndef TASK_SIZE_OF |
3445 | #define TASK_SIZE_OF(tsk) TASK_SIZE | |
3446 | #endif | |
3447 | ||
f98bafa0 | 3448 | #ifdef CONFIG_MEMCG |
cf475ad2 | 3449 | extern void mm_update_next_owner(struct mm_struct *mm); |
cf475ad2 BS |
3450 | #else |
3451 | static inline void mm_update_next_owner(struct mm_struct *mm) | |
3452 | { | |
3453 | } | |
f98bafa0 | 3454 | #endif /* CONFIG_MEMCG */ |
cf475ad2 | 3455 | |
3e10e716 JS |
3456 | static inline unsigned long task_rlimit(const struct task_struct *tsk, |
3457 | unsigned int limit) | |
3458 | { | |
316c1608 | 3459 | return READ_ONCE(tsk->signal->rlim[limit].rlim_cur); |
3e10e716 JS |
3460 | } |
3461 | ||
3462 | static inline unsigned long task_rlimit_max(const struct task_struct *tsk, | |
3463 | unsigned int limit) | |
3464 | { | |
316c1608 | 3465 | return READ_ONCE(tsk->signal->rlim[limit].rlim_max); |
3e10e716 JS |
3466 | } |
3467 | ||
3468 | static inline unsigned long rlimit(unsigned int limit) | |
3469 | { | |
3470 | return task_rlimit(current, limit); | |
3471 | } | |
3472 | ||
3473 | static inline unsigned long rlimit_max(unsigned int limit) | |
3474 | { | |
3475 | return task_rlimit_max(current, limit); | |
3476 | } | |
3477 | ||
adaf9fcd RW |
3478 | #ifdef CONFIG_CPU_FREQ |
3479 | struct update_util_data { | |
3480 | void (*func)(struct update_util_data *data, | |
3481 | u64 time, unsigned long util, unsigned long max); | |
3482 | }; | |
3483 | ||
0bed612b RW |
3484 | void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data, |
3485 | void (*func)(struct update_util_data *data, u64 time, | |
3486 | unsigned long util, unsigned long max)); | |
3487 | void cpufreq_remove_update_util_hook(int cpu); | |
adaf9fcd RW |
3488 | #endif /* CONFIG_CPU_FREQ */ |
3489 | ||
91a034ed MD |
3490 | #ifdef CONFIG_RSEQ |
3491 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
3492 | { | |
3493 | if (t->rseq) | |
3494 | set_tsk_thread_flag(t, TIF_NOTIFY_RESUME); | |
3495 | } | |
3496 | void __rseq_handle_notify_resume(struct pt_regs *regs); | |
3497 | static inline void rseq_handle_notify_resume(struct pt_regs *regs) | |
3498 | { | |
3499 | if (current->rseq) | |
3500 | __rseq_handle_notify_resume(regs); | |
3501 | } | |
3502 | /* | |
3503 | * If parent process has a registered restartable sequences area, the | |
3504 | * child inherits. Only applies when forking a process, not a thread. In | |
3505 | * case a parent fork() in the middle of a restartable sequence, set the | |
3506 | * resume notifier to force the child to retry. | |
3507 | */ | |
3508 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
3509 | { | |
3510 | if (clone_flags & CLONE_THREAD) { | |
3511 | t->rseq = NULL; | |
3512 | t->rseq_event_counter = 0; | |
3513 | t->rseq_refcount = 0; | |
3514 | } else { | |
3515 | t->rseq = current->rseq; | |
3516 | t->rseq_event_counter = current->rseq_event_counter; | |
3517 | t->rseq_refcount = current->rseq_refcount; | |
3518 | rseq_set_notify_resume(t); | |
3519 | } | |
3520 | } | |
3521 | static inline void rseq_execve(struct task_struct *t) | |
3522 | { | |
3523 | t->rseq = NULL; | |
3524 | t->rseq_event_counter = 0; | |
3525 | t->rseq_refcount = 0; | |
3526 | } | |
3527 | static inline void rseq_sched_out(struct task_struct *t) | |
3528 | { | |
3529 | rseq_set_notify_resume(t); | |
3530 | } | |
3531 | static inline void rseq_signal_deliver(struct pt_regs *regs) | |
3532 | { | |
3533 | rseq_handle_notify_resume(regs); | |
3534 | } | |
3535 | #else | |
3536 | static inline void rseq_set_notify_resume(struct task_struct *t) | |
3537 | { | |
3538 | } | |
3539 | static inline void rseq_handle_notify_resume(struct pt_regs *regs) | |
3540 | { | |
3541 | } | |
3542 | static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) | |
3543 | { | |
3544 | } | |
3545 | static inline void rseq_execve(struct task_struct *t) | |
3546 | { | |
3547 | } | |
3548 | static inline void rseq_sched_out(struct task_struct *t) | |
3549 | { | |
3550 | } | |
3551 | static inline void rseq_signal_deliver(struct pt_regs *regs) | |
3552 | { | |
3553 | } | |
3554 | #endif | |
3555 | ||
1da177e4 | 3556 | #endif |