~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

TOMOYO Linux Cross Reference
Linux/include/linux/sched.h

Version: ~ [ linux-5.19-rc3 ] ~ [ linux-5.18.5 ] ~ [ linux-5.17.15 ] ~ [ linux-5.16.20 ] ~ [ linux-5.15.48 ] ~ [ linux-5.14.21 ] ~ [ linux-5.13.19 ] ~ [ linux-5.12.19 ] ~ [ linux-5.11.22 ] ~ [ linux-5.10.123 ] ~ [ linux-5.9.16 ] ~ [ linux-5.8.18 ] ~ [ linux-5.7.19 ] ~ [ linux-5.6.19 ] ~ [ linux-5.5.19 ] ~ [ linux-5.4.199 ] ~ [ linux-5.3.18 ] ~ [ linux-5.2.21 ] ~ [ linux-5.1.21 ] ~ [ linux-5.0.21 ] ~ [ linux-4.20.17 ] ~ [ linux-4.19.248 ] ~ [ linux-4.18.20 ] ~ [ linux-4.17.19 ] ~ [ linux-4.16.18 ] ~ [ linux-4.15.18 ] ~ [ linux-4.14.284 ] ~ [ linux-4.13.16 ] ~ [ linux-4.12.14 ] ~ [ linux-4.11.12 ] ~ [ linux-4.10.17 ] ~ [ linux-4.9.319 ] ~ [ linux-4.8.17 ] ~ [ linux-4.7.10 ] ~ [ linux-4.6.7 ] ~ [ linux-4.5.7 ] ~ [ linux-4.4.302 ] ~ [ linux-4.3.6 ] ~ [ linux-4.2.8 ] ~ [ linux-4.1.52 ] ~ [ linux-4.0.9 ] ~ [ linux-3.10.108 ] ~ [ linux-2.6.32.71 ] ~ [ linux-2.6.0 ] ~ [ linux-2.4.37.11 ] ~ [ unix-v6-master ] ~ [ ccs-tools-1.8.9 ] ~ [ policy-sample ] ~
Architecture: ~ [ i386 ] ~ [ alpha ] ~ [ m68k ] ~ [ mips ] ~ [ ppc ] ~ [ sparc ] ~ [ sparc64 ] ~

  1 #ifndef _LINUX_SCHED_H
  2 #define _LINUX_SCHED_H
  3 
  4 #include <uapi/linux/sched.h>
  5 
  6 #include <linux/sched/prio.h>
  7 
  8 
  9 struct ccs_domain_info;
 10 
 11 struct sched_param {
 12         int sched_priority;
 13 };
 14 
 15 #include <asm/param.h>  /* for HZ */
 16 
 17 #include <linux/capability.h>
 18 #include <linux/threads.h>
 19 #include <linux/kernel.h>
 20 #include <linux/types.h>
 21 #include <linux/timex.h>
 22 #include <linux/jiffies.h>
 23 #include <linux/plist.h>
 24 #include <linux/rbtree.h>
 25 #include <linux/thread_info.h>
 26 #include <linux/cpumask.h>
 27 #include <linux/errno.h>
 28 #include <linux/nodemask.h>
 29 #include <linux/mm_types.h>
 30 #include <linux/preempt.h>
 31 
 32 #include <asm/page.h>
 33 #include <asm/ptrace.h>
 34 #include <linux/cputime.h>
 35 
 36 #include <linux/smp.h>
 37 #include <linux/sem.h>
 38 #include <linux/shm.h>
 39 #include <linux/signal.h>
 40 #include <linux/compiler.h>
 41 #include <linux/completion.h>
 42 #include <linux/pid.h>
 43 #include <linux/percpu.h>
 44 #include <linux/topology.h>
 45 #include <linux/seccomp.h>
 46 #include <linux/rcupdate.h>
 47 #include <linux/rculist.h>
 48 #include <linux/rtmutex.h>
 49 
 50 #include <linux/time.h>
 51 #include <linux/param.h>
 52 #include <linux/resource.h>
 53 #include <linux/timer.h>
 54 #include <linux/hrtimer.h>
 55 #include <linux/kcov.h>
 56 #include <linux/task_io_accounting.h>
 57 #include <linux/latencytop.h>
 58 #include <linux/cred.h>
 59 #include <linux/llist.h>
 60 #include <linux/uidgid.h>
 61 #include <linux/gfp.h>
 62 #include <linux/magic.h>
 63 #include <linux/cgroup-defs.h>
 64 
 65 #include <asm/processor.h>
 66 
 67 #define SCHED_ATTR_SIZE_VER0    48      /* sizeof first published struct */
 68 
 69 /*
 70  * Extended scheduling parameters data structure.
 71  *
 72  * This is needed because the original struct sched_param can not be
 73  * altered without introducing ABI issues with legacy applications
 74  * (e.g., in sched_getparam()).
 75  *
 76  * However, the possibility of specifying more than just a priority for
 77  * the tasks may be useful for a wide variety of application fields, e.g.,
 78  * multimedia, streaming, automation and control, and many others.
 79  *
 80  * This variant (sched_attr) is meant at describing a so-called
 81  * sporadic time-constrained task. In such model a task is specified by:
 82  *  - the activation period or minimum instance inter-arrival time;
 83  *  - the maximum (or average, depending on the actual scheduling
 84  *    discipline) computation time of all instances, a.k.a. runtime;
 85  *  - the deadline (relative to the actual activation time) of each
 86  *    instance.
 87  * Very briefly, a periodic (sporadic) task asks for the execution of
 88  * some specific computation --which is typically called an instance--
 89  * (at most) every period. Moreover, each instance typically lasts no more
 90  * than the runtime and must be completed by time instant t equal to
 91  * the instance activation time + the deadline.
 92  *
 93  * This is reflected by the actual fields of the sched_attr structure:
 94  *
 95  *  @size               size of the structure, for fwd/bwd compat.
 96  *
 97  *  @sched_policy       task's scheduling policy
 98  *  @sched_flags        for customizing the scheduler behaviour
 99  *  @sched_nice         task's nice value      (SCHED_NORMAL/BATCH)
100  *  @sched_priority     task's static priority (SCHED_FIFO/RR)
101  *  @sched_deadline     representative of the task's deadline
102  *  @sched_runtime      representative of the task's runtime
103  *  @sched_period       representative of the task's period
104  *
105  * Given this task model, there are a multiplicity of scheduling algorithms
106  * and policies, that can be used to ensure all the tasks will make their
107  * timing constraints.
108  *
109  * As of now, the SCHED_DEADLINE policy (sched_dl scheduling class) is the
110  * only user of this new interface. More information about the algorithm
111  * available in the scheduling class file or in Documentation/.
112  */
113 struct sched_attr {
114         u32 size;
115 
116         u32 sched_policy;
117         u64 sched_flags;
118 
119         /* SCHED_NORMAL, SCHED_BATCH */
120         s32 sched_nice;
121 
122         /* SCHED_FIFO, SCHED_RR */
123         u32 sched_priority;
124 
125         /* SCHED_DEADLINE */
126         u64 sched_runtime;
127         u64 sched_deadline;
128         u64 sched_period;
129 };
130 
131 struct futex_pi_state;
132 struct robust_list_head;
133 struct bio_list;
134 struct fs_struct;
135 struct perf_event_context;
136 struct blk_plug;
137 struct filename;
138 struct nameidata;
139 
140 #define VMACACHE_BITS 2
141 #define VMACACHE_SIZE (1U << VMACACHE_BITS)
142 #define VMACACHE_MASK (VMACACHE_SIZE - 1)
143 
144 /*
145  * These are the constant used to fake the fixed-point load-average
146  * counting. Some notes:
147  *  - 11 bit fractions expand to 22 bits by the multiplies: this gives
148  *    a load-average precision of 10 bits integer + 11 bits fractional
149  *  - if you want to count load-averages more often, you need more
150  *    precision, or rounding will get you. With 2-second counting freq,
151  *    the EXP_n values would be 1981, 2034 and 2043 if still using only
152  *    11 bit fractions.
153  */
154 extern unsigned long avenrun[];         /* Load averages */
155 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift);
156 
157 #define FSHIFT          11              /* nr of bits of precision */
158 #define FIXED_1         (1<<FSHIFT)     /* 1.0 as fixed-point */
159 #define LOAD_FREQ       (5*HZ+1)        /* 5 sec intervals */
160 #define EXP_1           1884            /* 1/exp(5sec/1min) as fixed-point */
161 #define EXP_5           2014            /* 1/exp(5sec/5min) */
162 #define EXP_15          2037            /* 1/exp(5sec/15min) */
163 
164 #define CALC_LOAD(load,exp,n) \
165         load *= exp; \
166         load += n*(FIXED_1-exp); \
167         load >>= FSHIFT;
168 
169 extern unsigned long total_forks;
170 extern int nr_threads;
171 DECLARE_PER_CPU(unsigned long, process_counts);
172 extern int nr_processes(void);
173 extern unsigned long nr_running(void);
174 extern bool single_task_running(void);
175 extern unsigned long nr_iowait(void);
176 extern unsigned long nr_iowait_cpu(int cpu);
177 extern void get_iowait_load(unsigned long *nr_waiters, unsigned long *load);
178 
179 extern void calc_global_load(unsigned long ticks);
180 
181 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
182 extern void cpu_load_update_nohz_start(void);
183 extern void cpu_load_update_nohz_stop(void);
184 #else
185 static inline void cpu_load_update_nohz_start(void) { }
186 static inline void cpu_load_update_nohz_stop(void) { }
187 #endif
188 
189 extern void dump_cpu_task(int cpu);
190 
191 struct seq_file;
192 struct cfs_rq;
193 struct task_group;
194 #ifdef CONFIG_SCHED_DEBUG
195 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m);
196 extern void proc_sched_set_task(struct task_struct *p);
197 #endif
198 
199 /*
200  * Task state bitmask. NOTE! These bits are also
201  * encoded in fs/proc/array.c: get_task_state().
202  *
203  * We have two separate sets of flags: task->state
204  * is about runnability, while task->exit_state are
205  * about the task exiting. Confusing, but this way
206  * modifying one set can't modify the other one by
207  * mistake.
208  */
209 #define TASK_RUNNING            0
210 #define TASK_INTERRUPTIBLE      1
211 #define TASK_UNINTERRUPTIBLE    2
212 #define __TASK_STOPPED          4
213 #define __TASK_TRACED           8
214 /* in tsk->exit_state */
215 #define EXIT_DEAD               16
216 #define EXIT_ZOMBIE             32
217 #define EXIT_TRACE              (EXIT_ZOMBIE | EXIT_DEAD)
218 /* in tsk->state again */
219 #define TASK_DEAD               64
220 #define TASK_WAKEKILL           128
221 #define TASK_WAKING             256
222 #define TASK_PARKED             512
223 #define TASK_NOLOAD             1024
224 #define TASK_NEW                2048
225 #define TASK_STATE_MAX          4096
226 
227 #define TASK_STATE_TO_CHAR_STR "RSDTtXZxKWPNn"
228 
229 extern char ___assert_task_state[1 - 2*!!(
230                 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)];
231 
232 /* Convenience macros for the sake of set_task_state */
233 #define TASK_KILLABLE           (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE)
234 #define TASK_STOPPED            (TASK_WAKEKILL | __TASK_STOPPED)
235 #define TASK_TRACED             (TASK_WAKEKILL | __TASK_TRACED)
236 
237 #define TASK_IDLE               (TASK_UNINTERRUPTIBLE | TASK_NOLOAD)
238 
239 /* Convenience macros for the sake of wake_up */
240 #define TASK_NORMAL             (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE)
241 #define TASK_ALL                (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED)
242 
243 /* get_task_state() */
244 #define TASK_REPORT             (TASK_RUNNING | TASK_INTERRUPTIBLE | \
245                                  TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \
246                                  __TASK_TRACED | EXIT_ZOMBIE | EXIT_DEAD)
247 
248 #define task_is_traced(task)    ((task->state & __TASK_TRACED) != 0)
249 #define task_is_stopped(task)   ((task->state & __TASK_STOPPED) != 0)
250 #define task_is_stopped_or_traced(task) \
251                         ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0)
252 #define task_contributes_to_load(task)  \
253                                 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \
254                                  (task->flags & PF_FROZEN) == 0 && \
255                                  (task->state & TASK_NOLOAD) == 0)
256 
257 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
258 
259 #define __set_task_state(tsk, state_value)                      \
260         do {                                                    \
261                 (tsk)->task_state_change = _THIS_IP_;           \
262                 (tsk)->state = (state_value);                   \
263         } while (0)
264 #define set_task_state(tsk, state_value)                        \
265         do {                                                    \
266                 (tsk)->task_state_change = _THIS_IP_;           \
267                 smp_store_mb((tsk)->state, (state_value));              \
268         } while (0)
269 
270 /*
271  * set_current_state() includes a barrier so that the write of current->state
272  * is correctly serialised wrt the caller's subsequent test of whether to
273  * actually sleep:
274  *
275  *      set_current_state(TASK_UNINTERRUPTIBLE);
276  *      if (do_i_need_to_sleep())
277  *              schedule();
278  *
279  * If the caller does not need such serialisation then use __set_current_state()
280  */
281 #define __set_current_state(state_value)                        \
282         do {                                                    \
283                 current->task_state_change = _THIS_IP_;         \
284                 current->state = (state_value);                 \
285         } while (0)
286 #define set_current_state(state_value)                          \
287         do {                                                    \
288                 current->task_state_change = _THIS_IP_;         \
289                 smp_store_mb(current->state, (state_value));            \
290         } while (0)
291 
292 #else
293 
294 #define __set_task_state(tsk, state_value)              \
295         do { (tsk)->state = (state_value); } while (0)
296 #define set_task_state(tsk, state_value)                \
297         smp_store_mb((tsk)->state, (state_value))
298 
299 /*
300  * set_current_state() includes a barrier so that the write of current->state
301  * is correctly serialised wrt the caller's subsequent test of whether to
302  * actually sleep:
303  *
304  *      set_current_state(TASK_UNINTERRUPTIBLE);
305  *      if (do_i_need_to_sleep())
306  *              schedule();
307  *
308  * If the caller does not need such serialisation then use __set_current_state()
309  */
310 #define __set_current_state(state_value)                \
311         do { current->state = (state_value); } while (0)
312 #define set_current_state(state_value)                  \
313         smp_store_mb(current->state, (state_value))
314 
315 #endif
316 
317 /* Task command name length */
318 #define TASK_COMM_LEN 16
319 
320 #include <linux/spinlock.h>
321 
322 /*
323  * This serializes "schedule()" and also protects
324  * the run-queue from deletions/modifications (but
325  * _adding_ to the beginning of the run-queue has
326  * a separate lock).
327  */
328 extern rwlock_t tasklist_lock;
329 extern spinlock_t mmlist_lock;
330 
331 struct task_struct;
332 
333 #ifdef CONFIG_PROVE_RCU
334 extern int lockdep_tasklist_lock_is_held(void);
335 #endif /* #ifdef CONFIG_PROVE_RCU */
336 
337 extern void sched_init(void);
338 extern void sched_init_smp(void);
339 extern asmlinkage void schedule_tail(struct task_struct *prev);
340 extern void init_idle(struct task_struct *idle, int cpu);
341 extern void init_idle_bootup_task(struct task_struct *idle);
342 
343 extern cpumask_var_t cpu_isolated_map;
344 
345 extern int runqueue_is_locked(int cpu);
346 
347 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ_COMMON)
348 extern void nohz_balance_enter_idle(int cpu);
349 extern void set_cpu_sd_state_idle(void);
350 extern int get_nohz_timer_target(void);
351 #else
352 static inline void nohz_balance_enter_idle(int cpu) { }
353 static inline void set_cpu_sd_state_idle(void) { }
354 #endif
355 
356 /*
357  * Only dump TASK_* tasks. (0 for all tasks)
358  */
359 extern void show_state_filter(unsigned long state_filter);
360 
361 static inline void show_state(void)
362 {
363         show_state_filter(0);
364 }
365 
366 extern void show_regs(struct pt_regs *);
367 
368 /*
369  * TASK is a pointer to the task whose backtrace we want to see (or NULL for current
370  * task), SP is the stack pointer of the first frame that should be shown in the back
371  * trace (or NULL if the entire call-chain of the task should be shown).
372  */
373 extern void show_stack(struct task_struct *task, unsigned long *sp);
374 
375 extern void cpu_init (void);
376 extern void trap_init(void);
377 extern void update_process_times(int user);
378 extern void scheduler_tick(void);
379 extern int sched_cpu_starting(unsigned int cpu);
380 extern int sched_cpu_activate(unsigned int cpu);
381 extern int sched_cpu_deactivate(unsigned int cpu);
382 
383 #ifdef CONFIG_HOTPLUG_CPU
384 extern int sched_cpu_dying(unsigned int cpu);
385 #else
386 # define sched_cpu_dying        NULL
387 #endif
388 
389 extern void sched_show_task(struct task_struct *p);
390 
391 #ifdef CONFIG_LOCKUP_DETECTOR
392 extern void touch_softlockup_watchdog_sched(void);
393 extern void touch_softlockup_watchdog(void);
394 extern void touch_softlockup_watchdog_sync(void);
395 extern void touch_all_softlockup_watchdogs(void);
396 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write,
397                                   void __user *buffer,
398                                   size_t *lenp, loff_t *ppos);
399 extern unsigned int  softlockup_panic;
400 extern unsigned int  hardlockup_panic;
401 void lockup_detector_init(void);
402 #else
403 static inline void touch_softlockup_watchdog_sched(void)
404 {
405 }
406 static inline void touch_softlockup_watchdog(void)
407 {
408 }
409 static inline void touch_softlockup_watchdog_sync(void)
410 {
411 }
412 static inline void touch_all_softlockup_watchdogs(void)
413 {
414 }
415 static inline void lockup_detector_init(void)
416 {
417 }
418 #endif
419 
420 #ifdef CONFIG_DETECT_HUNG_TASK
421 void reset_hung_task_detector(void);
422 #else
423 static inline void reset_hung_task_detector(void)
424 {
425 }
426 #endif
427 
428 /* Attach to any functions which should be ignored in wchan output. */
429 #define __sched         __attribute__((__section__(".sched.text")))
430 
431 /* Linker adds these: start and end of __sched functions */
432 extern char __sched_text_start[], __sched_text_end[];
433 
434 /* Is this address in the __sched functions? */
435 extern int in_sched_functions(unsigned long addr);
436 
437 #define MAX_SCHEDULE_TIMEOUT    LONG_MAX
438 extern signed long schedule_timeout(signed long timeout);
439 extern signed long schedule_timeout_interruptible(signed long timeout);
440 extern signed long schedule_timeout_killable(signed long timeout);
441 extern signed long schedule_timeout_uninterruptible(signed long timeout);
442 extern signed long schedule_timeout_idle(signed long timeout);
443 asmlinkage void schedule(void);
444 extern void schedule_preempt_disabled(void);
445 
446 extern long io_schedule_timeout(long timeout);
447 
448 static inline void io_schedule(void)
449 {
450         io_schedule_timeout(MAX_SCHEDULE_TIMEOUT);
451 }
452 
453 struct nsproxy;
454 struct user_namespace;
455 
456 #ifdef CONFIG_MMU
457 extern void arch_pick_mmap_layout(struct mm_struct *mm);
458 extern unsigned long
459 arch_get_unmapped_area(struct file *, unsigned long, unsigned long,
460                        unsigned long, unsigned long);
461 extern unsigned long
462 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr,
463                           unsigned long len, unsigned long pgoff,
464                           unsigned long flags);
465 #else
466 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {}
467 #endif
468 
469 #define SUID_DUMP_DISABLE       0       /* No setuid dumping */
470 #define SUID_DUMP_USER          1       /* Dump as user of process */
471 #define SUID_DUMP_ROOT          2       /* Dump as root */
472 
473 /* mm flags */
474 
475 /* for SUID_DUMP_* above */
476 #define MMF_DUMPABLE_BITS 2
477 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1)
478 
479 extern void set_dumpable(struct mm_struct *mm, int value);
480 /*
481  * This returns the actual value of the suid_dumpable flag. For things
482  * that are using this for checking for privilege transitions, it must
483  * test against SUID_DUMP_USER rather than treating it as a boolean
484  * value.
485  */
486 static inline int __get_dumpable(unsigned long mm_flags)
487 {
488         return mm_flags & MMF_DUMPABLE_MASK;
489 }
490 
491 static inline int get_dumpable(struct mm_struct *mm)
492 {
493         return __get_dumpable(mm->flags);
494 }
495 
496 /* coredump filter bits */
497 #define MMF_DUMP_ANON_PRIVATE   2
498 #define MMF_DUMP_ANON_SHARED    3
499 #define MMF_DUMP_MAPPED_PRIVATE 4
500 #define MMF_DUMP_MAPPED_SHARED  5
501 #define MMF_DUMP_ELF_HEADERS    6
502 #define MMF_DUMP_HUGETLB_PRIVATE 7
503 #define MMF_DUMP_HUGETLB_SHARED  8
504 #define MMF_DUMP_DAX_PRIVATE    9
505 #define MMF_DUMP_DAX_SHARED     10
506 
507 #define MMF_DUMP_FILTER_SHIFT   MMF_DUMPABLE_BITS
508 #define MMF_DUMP_FILTER_BITS    9
509 #define MMF_DUMP_FILTER_MASK \
510         (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT)
511 #define MMF_DUMP_FILTER_DEFAULT \
512         ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\
513          (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF)
514 
515 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS
516 # define MMF_DUMP_MASK_DEFAULT_ELF      (1 << MMF_DUMP_ELF_HEADERS)
517 #else
518 # define MMF_DUMP_MASK_DEFAULT_ELF      0
519 #endif
520                                         /* leave room for more dump flags */
521 #define MMF_VM_MERGEABLE        16      /* KSM may merge identical pages */
522 #define MMF_VM_HUGEPAGE         17      /* set when VM_HUGEPAGE is set on vma */
523 #define MMF_EXE_FILE_CHANGED    18      /* see prctl_set_mm_exe_file() */
524 
525 #define MMF_HAS_UPROBES         19      /* has uprobes */
526 #define MMF_RECALC_UPROBES      20      /* MMF_HAS_UPROBES can be wrong */
527 #define MMF_OOM_REAPED          21      /* mm has been already reaped */
528 #define MMF_OOM_NOT_REAPABLE    22      /* mm couldn't be reaped */
529 
530 #define MMF_INIT_MASK           (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK)
531 
532 struct sighand_struct {
533         atomic_t                count;
534         struct k_sigaction      action[_NSIG];
535         spinlock_t              siglock;
536         wait_queue_head_t       signalfd_wqh;
537 };
538 
539 struct pacct_struct {
540         int                     ac_flag;
541         long                    ac_exitcode;
542         unsigned long           ac_mem;
543         cputime_t               ac_utime, ac_stime;
544         unsigned long           ac_minflt, ac_majflt;
545 };
546 
547 struct cpu_itimer {
548         cputime_t expires;
549         cputime_t incr;
550         u32 error;
551         u32 incr_error;
552 };
553 
554 /**
555  * struct prev_cputime - snaphsot of system and user cputime
556  * @utime: time spent in user mode
557  * @stime: time spent in system mode
558  * @lock: protects the above two fields
559  *
560  * Stores previous user/system time values such that we can guarantee
561  * monotonicity.
562  */
563 struct prev_cputime {
564 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
565         cputime_t utime;
566         cputime_t stime;
567         raw_spinlock_t lock;
568 #endif
569 };
570 
571 static inline void prev_cputime_init(struct prev_cputime *prev)
572 {
573 #ifndef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE
574         prev->utime = prev->stime = 0;
575         raw_spin_lock_init(&prev->lock);
576 #endif
577 }
578 
579 /**
580  * struct task_cputime - collected CPU time counts
581  * @utime:              time spent in user mode, in &cputime_t units
582  * @stime:              time spent in kernel mode, in &cputime_t units
583  * @sum_exec_runtime:   total time spent on the CPU, in nanoseconds
584  *
585  * This structure groups together three kinds of CPU time that are tracked for
586  * threads and thread groups.  Most things considering CPU time want to group
587  * these counts together and treat all three of them in parallel.
588  */
589 struct task_cputime {
590         cputime_t utime;
591         cputime_t stime;
592         unsigned long long sum_exec_runtime;
593 };
594 
595 /* Alternate field names when used to cache expirations. */
596 #define virt_exp        utime
597 #define prof_exp        stime
598 #define sched_exp       sum_exec_runtime
599 
600 #define INIT_CPUTIME    \
601         (struct task_cputime) {                                 \
602                 .utime = 0,                                     \
603                 .stime = 0,                                     \
604                 .sum_exec_runtime = 0,                          \
605         }
606 
607 /*
608  * This is the atomic variant of task_cputime, which can be used for
609  * storing and updating task_cputime statistics without locking.
610  */
611 struct task_cputime_atomic {
612         atomic64_t utime;
613         atomic64_t stime;
614         atomic64_t sum_exec_runtime;
615 };
616 
617 #define INIT_CPUTIME_ATOMIC \
618         (struct task_cputime_atomic) {                          \
619                 .utime = ATOMIC64_INIT(0),                      \
620                 .stime = ATOMIC64_INIT(0),                      \
621                 .sum_exec_runtime = ATOMIC64_INIT(0),           \
622         }
623 
624 #define PREEMPT_DISABLED        (PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
625 
626 /*
627  * Disable preemption until the scheduler is running -- use an unconditional
628  * value so that it also works on !PREEMPT_COUNT kernels.
629  *
630  * Reset by start_kernel()->sched_init()->init_idle()->init_idle_preempt_count().
631  */
632 #define INIT_PREEMPT_COUNT      PREEMPT_OFFSET
633 
634 /*
635  * Initial preempt_count value; reflects the preempt_count schedule invariant
636  * which states that during context switches:
637  *
638  *    preempt_count() == 2*PREEMPT_DISABLE_OFFSET
639  *
640  * Note: PREEMPT_DISABLE_OFFSET is 0 for !PREEMPT_COUNT kernels.
641  * Note: See finish_task_switch().
642  */
643 #define FORK_PREEMPT_COUNT      (2*PREEMPT_DISABLE_OFFSET + PREEMPT_ENABLED)
644 
645 /**
646  * struct thread_group_cputimer - thread group interval timer counts
647  * @cputime_atomic:     atomic thread group interval timers.
648  * @running:            true when there are timers running and
649  *                      @cputime_atomic receives updates.
650  * @checking_timer:     true when a thread in the group is in the
651  *                      process of checking for thread group timers.
652  *
653  * This structure contains the version of task_cputime, above, that is
654  * used for thread group CPU timer calculations.
655  */
656 struct thread_group_cputimer {
657         struct task_cputime_atomic cputime_atomic;
658         bool running;
659         bool checking_timer;
660 };
661 
662 #include <linux/rwsem.h>
663 struct autogroup;
664 
665 /*
666  * NOTE! "signal_struct" does not have its own
667  * locking, because a shared signal_struct always
668  * implies a shared sighand_struct, so locking
669  * sighand_struct is always a proper superset of
670  * the locking of signal_struct.
671  */
672 struct signal_struct {
673         atomic_t                sigcnt;
674         atomic_t                live;
675         int                     nr_threads;
676         atomic_t oom_victims; /* # of TIF_MEDIE threads in this thread group */
677         struct list_head        thread_head;
678 
679         wait_queue_head_t       wait_chldexit;  /* for wait4() */
680 
681         /* current thread group signal load-balancing target: */
682         struct task_struct      *curr_target;
683 
684         /* shared signal handling: */
685         struct sigpending       shared_pending;
686 
687         /* thread group exit support */
688         int                     group_exit_code;
689         /* overloaded:
690          * - notify group_exit_task when ->count is equal to notify_count
691          * - everyone except group_exit_task is stopped during signal delivery
692          *   of fatal signals, group_exit_task processes the signal.
693          */
694         int                     notify_count;
695         struct task_struct      *group_exit_task;
696 
697         /* thread group stop support, overloads group_exit_code too */
698         int                     group_stop_count;
699         unsigned int            flags; /* see SIGNAL_* flags below */
700 
701         /*
702          * PR_SET_CHILD_SUBREAPER marks a process, like a service
703          * manager, to re-parent orphan (double-forking) child processes
704          * to this process instead of 'init'. The service manager is
705          * able to receive SIGCHLD signals and is able to investigate
706          * the process until it calls wait(). All children of this
707          * process will inherit a flag if they should look for a
708          * child_subreaper process at exit.
709          */
710         unsigned int            is_child_subreaper:1;
711         unsigned int            has_child_subreaper:1;
712 
713         /* POSIX.1b Interval Timers */
714         int                     posix_timer_id;
715         struct list_head        posix_timers;
716 
717         /* ITIMER_REAL timer for the process */
718         struct hrtimer real_timer;
719         struct pid *leader_pid;
720         ktime_t it_real_incr;
721 
722         /*
723          * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use
724          * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these
725          * values are defined to 0 and 1 respectively
726          */
727         struct cpu_itimer it[2];
728 
729         /*
730          * Thread group totals for process CPU timers.
731          * See thread_group_cputimer(), et al, for details.
732          */
733         struct thread_group_cputimer cputimer;
734 
735         /* Earliest-expiration cache. */
736         struct task_cputime cputime_expires;
737 
738 #ifdef CONFIG_NO_HZ_FULL
739         atomic_t tick_dep_mask;
740 #endif
741 
742         struct list_head cpu_timers[3];
743 
744         struct pid *tty_old_pgrp;
745 
746         /* boolean value for session group leader */
747         int leader;
748 
749         struct tty_struct *tty; /* NULL if no tty */
750 
751 #ifdef CONFIG_SCHED_AUTOGROUP
752         struct autogroup *autogroup;
753 #endif
754         /*
755          * Cumulative resource counters for dead threads in the group,
756          * and for reaped dead child processes forked by this group.
757          * Live threads maintain their own counters and add to these
758          * in __exit_signal, except for the group leader.
759          */
760         seqlock_t stats_lock;
761         cputime_t utime, stime, cutime, cstime;
762         cputime_t gtime;
763         cputime_t cgtime;
764         struct prev_cputime prev_cputime;
765         unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw;
766         unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt;
767         unsigned long inblock, oublock, cinblock, coublock;
768         unsigned long maxrss, cmaxrss;
769         struct task_io_accounting ioac;
770 
771         /*
772          * Cumulative ns of schedule CPU time fo dead threads in the
773          * group, not including a zombie group leader, (This only differs
774          * from jiffies_to_ns(utime + stime) if sched_clock uses something
775          * other than jiffies.)
776          */
777         unsigned long long sum_sched_runtime;
778 
779         /*
780          * We don't bother to synchronize most readers of this at all,
781          * because there is no reader checking a limit that actually needs
782          * to get both rlim_cur and rlim_max atomically, and either one
783          * alone is a single word that can safely be read normally.
784          * getrlimit/setrlimit use task_lock(current->group_leader) to
785          * protect this instead of the siglock, because they really
786          * have no need to disable irqs.
787          */
788         struct rlimit rlim[RLIM_NLIMITS];
789 
790 #ifdef CONFIG_BSD_PROCESS_ACCT
791         struct pacct_struct pacct;      /* per-process accounting information */
792 #endif
793 #ifdef CONFIG_TASKSTATS
794         struct taskstats *stats;
795 #endif
796 #ifdef CONFIG_AUDIT
797         unsigned audit_tty;
798         struct tty_audit_buf *tty_audit_buf;
799 #endif
800 
801         /*
802          * Thread is the potential origin of an oom condition; kill first on
803          * oom
804          */
805         bool oom_flag_origin;
806         short oom_score_adj;            /* OOM kill score adjustment */
807         short oom_score_adj_min;        /* OOM kill score adjustment min value.
808                                          * Only settable by CAP_SYS_RESOURCE. */
809 
810         struct mutex cred_guard_mutex;  /* guard against foreign influences on
811                                          * credential calculations
812                                          * (notably. ptrace) */
813 };
814 
815 /*
816  * Bits in flags field of signal_struct.
817  */
818 #define SIGNAL_STOP_STOPPED     0x00000001 /* job control stop in effect */
819 #define SIGNAL_STOP_CONTINUED   0x00000002 /* SIGCONT since WCONTINUED reap */
820 #define SIGNAL_GROUP_EXIT       0x00000004 /* group exit in progress */
821 #define SIGNAL_GROUP_COREDUMP   0x00000008 /* coredump in progress */
822 /*
823  * Pending notifications to parent.
824  */
825 #define SIGNAL_CLD_STOPPED      0x00000010
826 #define SIGNAL_CLD_CONTINUED    0x00000020
827 #define SIGNAL_CLD_MASK         (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED)
828 
829 #define SIGNAL_UNKILLABLE       0x00000040 /* for init: ignore fatal signals */
830 
831 /* If true, all threads except ->group_exit_task have pending SIGKILL */
832 static inline int signal_group_exit(const struct signal_struct *sig)
833 {
834         return  (sig->flags & SIGNAL_GROUP_EXIT) ||
835                 (sig->group_exit_task != NULL);
836 }
837 
838 /*
839  * Some day this will be a full-fledged user tracking system..
840  */
841 struct user_struct {
842         atomic_t __count;       /* reference count */
843         atomic_t processes;     /* How many processes does this user have? */
844         atomic_t sigpending;    /* How many pending signals does this user have? */
845 #ifdef CONFIG_INOTIFY_USER
846         atomic_t inotify_watches; /* How many inotify watches does this user have? */
847         atomic_t inotify_devs;  /* How many inotify devs does this user have opened? */
848 #endif
849 #ifdef CONFIG_FANOTIFY
850         atomic_t fanotify_listeners;
851 #endif
852 #ifdef CONFIG_EPOLL
853         atomic_long_t epoll_watches; /* The number of file descriptors currently watched */
854 #endif
855 #ifdef CONFIG_POSIX_MQUEUE
856         /* protected by mq_lock */
857         unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */
858 #endif
859         unsigned long locked_shm; /* How many pages of mlocked shm ? */
860         unsigned long unix_inflight;    /* How many files in flight in unix sockets */
861         atomic_long_t pipe_bufs;  /* how many pages are allocated in pipe buffers */
862 
863 #ifdef CONFIG_KEYS
864         struct key *uid_keyring;        /* UID specific keyring */
865         struct key *session_keyring;    /* UID's default session keyring */
866 #endif
867 
868         /* Hash table maintenance information */
869         struct hlist_node uidhash_node;
870         kuid_t uid;
871 
872 #if defined(CONFIG_PERF_EVENTS) || defined(CONFIG_BPF_SYSCALL)
873         atomic_long_t locked_vm;
874 #endif
875 };
876 
877 extern int uids_sysfs_init(void);
878 
879 extern struct user_struct *find_user(kuid_t);
880 
881 extern struct user_struct root_user;
882 #define INIT_USER (&root_user)
883 
884 
885 struct backing_dev_info;
886 struct reclaim_state;
887 
888 #ifdef CONFIG_SCHED_INFO
889 struct sched_info {
890         /* cumulative counters */
891         unsigned long pcount;         /* # of times run on this cpu */
892         unsigned long long run_delay; /* time spent waiting on a runqueue */
893 
894         /* timestamps */
895         unsigned long long last_arrival,/* when we last ran on a cpu */
896                            last_queued; /* when we were last queued to run */
897 };
898 #endif /* CONFIG_SCHED_INFO */
899 
900 #ifdef CONFIG_TASK_DELAY_ACCT
901 struct task_delay_info {
902         spinlock_t      lock;
903         unsigned int    flags;  /* Private per-task flags */
904 
905         /* For each stat XXX, add following, aligned appropriately
906          *
907          * struct timespec XXX_start, XXX_end;
908          * u64 XXX_delay;
909          * u32 XXX_count;
910          *
911          * Atomicity of updates to XXX_delay, XXX_count protected by
912          * single lock above (split into XXX_lock if contention is an issue).
913          */
914 
915         /*
916          * XXX_count is incremented on every XXX operation, the delay
917          * associated with the operation is added to XXX_delay.
918          * XXX_delay contains the accumulated delay time in nanoseconds.
919          */
920         u64 blkio_start;        /* Shared by blkio, swapin */
921         u64 blkio_delay;        /* wait for sync block io completion */
922         u64 swapin_delay;       /* wait for swapin block io completion */
923         u32 blkio_count;        /* total count of the number of sync block */
924                                 /* io operations performed */
925         u32 swapin_count;       /* total count of the number of swapin block */
926                                 /* io operations performed */
927 
928         u64 freepages_start;
929         u64 freepages_delay;    /* wait for memory reclaim */
930         u32 freepages_count;    /* total count of memory reclaim */
931 };
932 #endif  /* CONFIG_TASK_DELAY_ACCT */
933 
934 static inline int sched_info_on(void)
935 {
936 #ifdef CONFIG_SCHEDSTATS
937         return 1;
938 #elif defined(CONFIG_TASK_DELAY_ACCT)
939         extern int delayacct_on;
940         return delayacct_on;
941 #else
942         return 0;
943 #endif
944 }
945 
946 #ifdef CONFIG_SCHEDSTATS
947 void force_schedstat_enabled(void);
948 #endif
949 
950 enum cpu_idle_type {
951         CPU_IDLE,
952         CPU_NOT_IDLE,
953         CPU_NEWLY_IDLE,
954         CPU_MAX_IDLE_TYPES
955 };
956 
957 /*
958  * Integer metrics need fixed point arithmetic, e.g., sched/fair
959  * has a few: load, load_avg, util_avg, freq, and capacity.
960  *
961  * We define a basic fixed point arithmetic range, and then formalize
962  * all these metrics based on that basic range.
963  */
964 # define SCHED_FIXEDPOINT_SHIFT 10
965 # define SCHED_FIXEDPOINT_SCALE (1L << SCHED_FIXEDPOINT_SHIFT)
966 
967 /*
968  * Increase resolution of cpu_capacity calculations
969  */
970 #define SCHED_CAPACITY_SHIFT    SCHED_FIXEDPOINT_SHIFT
971 #define SCHED_CAPACITY_SCALE    (1L << SCHED_CAPACITY_SHIFT)
972 
973 /*
974  * Wake-queues are lists of tasks with a pending wakeup, whose
975  * callers have already marked the task as woken internally,
976  * and can thus carry on. A common use case is being able to
977  * do the wakeups once the corresponding user lock as been
978  * released.
979  *
980  * We hold reference to each task in the list across the wakeup,
981  * thus guaranteeing that the memory is still valid by the time
982  * the actual wakeups are performed in wake_up_q().
983  *
984  * One per task suffices, because there's never a need for a task to be
985  * in two wake queues simultaneously; it is forbidden to abandon a task
986  * in a wake queue (a call to wake_up_q() _must_ follow), so if a task is
987  * already in a wake queue, the wakeup will happen soon and the second
988  * waker can just skip it.
989  *
990  * The WAKE_Q macro declares and initializes the list head.
991  * wake_up_q() does NOT reinitialize the list; it's expected to be
992  * called near the end of a function, where the fact that the queue is
993  * not used again will be easy to see by inspection.
994  *
995  * Note that this can cause spurious wakeups. schedule() callers
996  * must ensure the call is done inside a loop, confirming that the
997  * wakeup condition has in fact occurred.
998  */
999 struct wake_q_node {
1000         struct wake_q_node *next;
1001 };
1002 
1003 struct wake_q_head {
1004         struct wake_q_node *first;
1005         struct wake_q_node **lastp;
1006 };
1007 
1008 #define WAKE_Q_TAIL ((struct wake_q_node *) 0x01)
1009 
1010 #define WAKE_Q(name)                                    \
1011         struct wake_q_head name = { WAKE_Q_TAIL, &name.first }
1012 
1013 extern void wake_q_add(struct wake_q_head *head,
1014                        struct task_struct *task);
1015 extern void wake_up_q(struct wake_q_head *head);
1016 
1017 /*
1018  * sched-domains (multiprocessor balancing) declarations:
1019  */
1020 #ifdef CONFIG_SMP
1021 #define SD_LOAD_BALANCE         0x0001  /* Do load balancing on this domain. */
1022 #define SD_BALANCE_NEWIDLE      0x0002  /* Balance when about to become idle */
1023 #define SD_BALANCE_EXEC         0x0004  /* Balance on exec */
1024 #define SD_BALANCE_FORK         0x0008  /* Balance on fork, clone */
1025 #define SD_BALANCE_WAKE         0x0010  /* Balance on wakeup */
1026 #define SD_WAKE_AFFINE          0x0020  /* Wake task to waking CPU */
1027 #define SD_SHARE_CPUCAPACITY    0x0080  /* Domain members share cpu power */
1028 #define SD_SHARE_POWERDOMAIN    0x0100  /* Domain members share power domain */
1029 #define SD_SHARE_PKG_RESOURCES  0x0200  /* Domain members share cpu pkg resources */
1030 #define SD_SERIALIZE            0x0400  /* Only a single load balancing instance */
1031 #define SD_ASYM_PACKING         0x0800  /* Place busy groups earlier in the domain */
1032 #define SD_PREFER_SIBLING       0x1000  /* Prefer to place tasks in a sibling domain */
1033 #define SD_OVERLAP              0x2000  /* sched_domains of this level overlap */
1034 #define SD_NUMA                 0x4000  /* cross-node balancing */
1035 
1036 #ifdef CONFIG_SCHED_SMT
1037 static inline int cpu_smt_flags(void)
1038 {
1039         return SD_SHARE_CPUCAPACITY | SD_SHARE_PKG_RESOURCES;
1040 }
1041 #endif
1042 
1043 #ifdef CONFIG_SCHED_MC
1044 static inline int cpu_core_flags(void)
1045 {
1046         return SD_SHARE_PKG_RESOURCES;
1047 }
1048 #endif
1049 
1050 #ifdef CONFIG_NUMA
1051 static inline int cpu_numa_flags(void)
1052 {
1053         return SD_NUMA;
1054 }
1055 #endif
1056 
1057 struct sched_domain_attr {
1058         int relax_domain_level;
1059 };
1060 
1061 #define SD_ATTR_INIT    (struct sched_domain_attr) {    \
1062         .relax_domain_level = -1,                       \
1063 }
1064 
1065 extern int sched_domain_level_max;
1066 
1067 struct sched_group;
1068 
1069 struct sched_domain {
1070         /* These fields must be setup */
1071         struct sched_domain *parent;    /* top domain must be null terminated */
1072         struct sched_domain *child;     /* bottom domain must be null terminated */
1073         struct sched_group *groups;     /* the balancing groups of the domain */
1074         unsigned long min_interval;     /* Minimum balance interval ms */
1075         unsigned long max_interval;     /* Maximum balance interval ms */
1076         unsigned int busy_factor;       /* less balancing by factor if busy */
1077         unsigned int imbalance_pct;     /* No balance until over watermark */
1078         unsigned int cache_nice_tries;  /* Leave cache hot tasks for # tries */
1079         unsigned int busy_idx;
1080         unsigned int idle_idx;
1081         unsigned int newidle_idx;
1082         unsigned int wake_idx;
1083         unsigned int forkexec_idx;
1084         unsigned int smt_gain;
1085 
1086         int nohz_idle;                  /* NOHZ IDLE status */
1087         int flags;                      /* See SD_* */
1088         int level;
1089 
1090         /* Runtime fields. */
1091         unsigned long last_balance;     /* init to jiffies. units in jiffies */
1092         unsigned int balance_interval;  /* initialise to 1. units in ms. */
1093         unsigned int nr_balance_failed; /* initialise to 0 */
1094 
1095         /* idle_balance() stats */
1096         u64 max_newidle_lb_cost;
1097         unsigned long next_decay_max_lb_cost;
1098 
1099 #ifdef CONFIG_SCHEDSTATS
1100         /* load_balance() stats */
1101         unsigned int lb_count[CPU_MAX_IDLE_TYPES];
1102         unsigned int lb_failed[CPU_MAX_IDLE_TYPES];
1103         unsigned int lb_balanced[CPU_MAX_IDLE_TYPES];
1104         unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES];
1105         unsigned int lb_gained[CPU_MAX_IDLE_TYPES];
1106         unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES];
1107         unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES];
1108         unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES];
1109 
1110         /* Active load balancing */
1111         unsigned int alb_count;
1112         unsigned int alb_failed;
1113         unsigned int alb_pushed;
1114 
1115         /* SD_BALANCE_EXEC stats */
1116         unsigned int sbe_count;
1117         unsigned int sbe_balanced;
1118         unsigned int sbe_pushed;
1119 
1120         /* SD_BALANCE_FORK stats */
1121         unsigned int sbf_count;
1122         unsigned int sbf_balanced;
1123         unsigned int sbf_pushed;
1124 
1125         /* try_to_wake_up() stats */
1126         unsigned int ttwu_wake_remote;
1127         unsigned int ttwu_move_affine;
1128         unsigned int ttwu_move_balance;
1129 #endif
1130 #ifdef CONFIG_SCHED_DEBUG
1131         char *name;
1132 #endif
1133         union {
1134                 void *private;          /* used during construction */
1135                 struct rcu_head rcu;    /* used during destruction */
1136         };
1137 
1138         unsigned int span_weight;
1139         /*
1140          * Span of all CPUs in this domain.
1141          *
1142          * NOTE: this field is variable length. (Allocated dynamically
1143          * by attaching extra space to the end of the structure,
1144          * depending on how many CPUs the kernel has booted up with)
1145          */
1146         unsigned long span[0];
1147 };
1148 
1149 static inline struct cpumask *sched_domain_span(struct sched_domain *sd)
1150 {
1151         return to_cpumask(sd->span);
1152 }
1153 
1154 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1155                                     struct sched_domain_attr *dattr_new);
1156 
1157 /* Allocate an array of sched domains, for partition_sched_domains(). */
1158 cpumask_var_t *alloc_sched_domains(unsigned int ndoms);
1159 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms);
1160 
1161 bool cpus_share_cache(int this_cpu, int that_cpu);
1162 
1163 typedef const struct cpumask *(*sched_domain_mask_f)(int cpu);
1164 typedef int (*sched_domain_flags_f)(void);
1165 
1166 #define SDTL_OVERLAP    0x01
1167 
1168 struct sd_data {
1169         struct sched_domain **__percpu sd;
1170         struct sched_group **__percpu sg;
1171         struct sched_group_capacity **__percpu sgc;
1172 };
1173 
1174 struct sched_domain_topology_level {
1175         sched_domain_mask_f mask;
1176         sched_domain_flags_f sd_flags;
1177         int                 flags;
1178         int                 numa_level;
1179         struct sd_data      data;
1180 #ifdef CONFIG_SCHED_DEBUG
1181         char                *name;
1182 #endif
1183 };
1184 
1185 extern void set_sched_topology(struct sched_domain_topology_level *tl);
1186 extern void wake_up_if_idle(int cpu);
1187 
1188 #ifdef CONFIG_SCHED_DEBUG
1189 # define SD_INIT_NAME(type)             .name = #type
1190 #else
1191 # define SD_INIT_NAME(type)
1192 #endif
1193 
1194 #else /* CONFIG_SMP */
1195 
1196 struct sched_domain_attr;
1197 
1198 static inline void
1199 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[],
1200                         struct sched_domain_attr *dattr_new)
1201 {
1202 }
1203 
1204 static inline bool cpus_share_cache(int this_cpu, int that_cpu)
1205 {
1206         return true;
1207 }
1208 
1209 #endif  /* !CONFIG_SMP */
1210 
1211 
1212 struct io_context;                      /* See blkdev.h */
1213 
1214 
1215 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK
1216 extern void prefetch_stack(struct task_struct *t);
1217 #else
1218 static inline void prefetch_stack(struct task_struct *t) { }
1219 #endif
1220 
1221 struct audit_context;           /* See audit.c */
1222 struct mempolicy;
1223 struct pipe_inode_info;
1224 struct uts_namespace;
1225 
1226 struct load_weight {
1227         unsigned long weight;
1228         u32 inv_weight;
1229 };
1230 
1231 /*
1232  * The load_avg/util_avg accumulates an infinite geometric series
1233  * (see __update_load_avg() in kernel/sched/fair.c).
1234  *
1235  * [load_avg definition]
1236  *
1237  *   load_avg = runnable% * scale_load_down(load)
1238  *
1239  * where runnable% is the time ratio that a sched_entity is runnable.
1240  * For cfs_rq, it is the aggregated load_avg of all runnable and
1241  * blocked sched_entities.
1242  *
1243  * load_avg may also take frequency scaling into account:
1244  *
1245  *   load_avg = runnable% * scale_load_down(load) * freq%
1246  *
1247  * where freq% is the CPU frequency normalized to the highest frequency.
1248  *
1249  * [util_avg definition]
1250  *
1251  *   util_avg = running% * SCHED_CAPACITY_SCALE
1252  *
1253  * where running% is the time ratio that a sched_entity is running on
1254  * a CPU. For cfs_rq, it is the aggregated util_avg of all runnable
1255  * and blocked sched_entities.
1256  *
1257  * util_avg may also factor frequency scaling and CPU capacity scaling:
1258  *
1259  *   util_avg = running% * SCHED_CAPACITY_SCALE * freq% * capacity%
1260  *
1261  * where freq% is the same as above, and capacity% is the CPU capacity
1262  * normalized to the greatest capacity (due to uarch differences, etc).
1263  *
1264  * N.B., the above ratios (runnable%, running%, freq%, and capacity%)
1265  * themselves are in the range of [0, 1]. To do fixed point arithmetics,
1266  * we therefore scale them to as large a range as necessary. This is for
1267  * example reflected by util_avg's SCHED_CAPACITY_SCALE.
1268  *
1269  * [Overflow issue]
1270  *
1271  * The 64-bit load_sum can have 4353082796 (=2^64/47742/88761) entities
1272  * with the highest load (=88761), always runnable on a single cfs_rq,
1273  * and should not overflow as the number already hits PID_MAX_LIMIT.
1274  *
1275  * For all other cases (including 32-bit kernels), struct load_weight's
1276  * weight will overflow first before we do, because:
1277  *
1278  *    Max(load_avg) <= Max(load.weight)
1279  *
1280  * Then it is the load_weight's responsibility to consider overflow
1281  * issues.
1282  */
1283 struct sched_avg {
1284         u64 last_update_time, load_sum;
1285         u32 util_sum, period_contrib;
1286         unsigned long load_avg, util_avg;
1287 };
1288 
1289 #ifdef CONFIG_SCHEDSTATS
1290 struct sched_statistics {
1291         u64                     wait_start;
1292         u64                     wait_max;
1293         u64                     wait_count;
1294         u64                     wait_sum;
1295         u64                     iowait_count;
1296         u64                     iowait_sum;
1297 
1298         u64                     sleep_start;
1299         u64                     sleep_max;
1300         s64                     sum_sleep_runtime;
1301 
1302         u64                     block_start;
1303         u64                     block_max;
1304         u64                     exec_max;
1305         u64                     slice_max;
1306 
1307         u64                     nr_migrations_cold;
1308         u64                     nr_failed_migrations_affine;
1309         u64                     nr_failed_migrations_running;
1310         u64                     nr_failed_migrations_hot;
1311         u64                     nr_forced_migrations;
1312 
1313         u64                     nr_wakeups;
1314         u64                     nr_wakeups_sync;
1315         u64                     nr_wakeups_migrate;
1316         u64                     nr_wakeups_local;
1317         u64                     nr_wakeups_remote;
1318         u64                     nr_wakeups_affine;
1319         u64                     nr_wakeups_affine_attempts;
1320         u64                     nr_wakeups_passive;
1321         u64                     nr_wakeups_idle;
1322 };
1323 #endif
1324 
1325 struct sched_entity {
1326         struct load_weight      load;           /* for load-balancing */
1327         struct rb_node          run_node;
1328         struct list_head        group_node;
1329         unsigned int            on_rq;
1330 
1331         u64                     exec_start;
1332         u64                     sum_exec_runtime;
1333         u64                     vruntime;
1334         u64                     prev_sum_exec_runtime;
1335 
1336         u64                     nr_migrations;
1337 
1338 #ifdef CONFIG_SCHEDSTATS
1339         struct sched_statistics statistics;
1340 #endif
1341 
1342 #ifdef CONFIG_FAIR_GROUP_SCHED
1343         int                     depth;
1344         struct sched_entity     *parent;
1345         /* rq on which this entity is (to be) queued: */
1346         struct cfs_rq           *cfs_rq;
1347         /* rq "owned" by this entity/group: */
1348         struct cfs_rq           *my_q;
1349 #endif
1350 
1351 #ifdef CONFIG_SMP
1352         /*
1353          * Per entity load average tracking.
1354          *
1355          * Put into separate cache line so it does not
1356          * collide with read-mostly values above.
1357          */
1358         struct sched_avg        avg ____cacheline_aligned_in_smp;
1359 #endif
1360 };
1361 
1362 struct sched_rt_entity {
1363         struct list_head run_list;
1364         unsigned long timeout;
1365         unsigned long watchdog_stamp;
1366         unsigned int time_slice;
1367         unsigned short on_rq;
1368         unsigned short on_list;
1369 
1370         struct sched_rt_entity *back;
1371 #ifdef CONFIG_RT_GROUP_SCHED
1372         struct sched_rt_entity  *parent;
1373         /* rq on which this entity is (to be) queued: */
1374         struct rt_rq            *rt_rq;
1375         /* rq "owned" by this entity/group: */
1376         struct rt_rq            *my_q;
1377 #endif
1378 };
1379 
1380 struct sched_dl_entity {
1381         struct rb_node  rb_node;
1382 
1383         /*
1384          * Original scheduling parameters. Copied here from sched_attr
1385          * during sched_setattr(), they will remain the same until
1386          * the next sched_setattr().
1387          */
1388         u64 dl_runtime;         /* maximum runtime for each instance    */
1389         u64 dl_deadline;        /* relative deadline of each instance   */
1390         u64 dl_period;          /* separation of two instances (period) */
1391         u64 dl_bw;              /* dl_runtime / dl_deadline             */
1392 
1393         /*
1394          * Actual scheduling parameters. Initialized with the values above,
1395          * they are continously updated during task execution. Note that
1396          * the remaining runtime could be < 0 in case we are in overrun.
1397          */
1398         s64 runtime;            /* remaining runtime for this instance  */
1399         u64 deadline;           /* absolute deadline for this instance  */
1400         unsigned int flags;     /* specifying the scheduler behaviour   */
1401 
1402         /*
1403          * Some bool flags:
1404          *
1405          * @dl_throttled tells if we exhausted the runtime. If so, the
1406          * task has to wait for a replenishment to be performed at the
1407          * next firing of dl_timer.
1408          *
1409          * @dl_boosted tells if we are boosted due to DI. If so we are
1410          * outside bandwidth enforcement mechanism (but only until we
1411          * exit the critical section);
1412          *
1413          * @dl_yielded tells if task gave up the cpu before consuming
1414          * all its available runtime during the last job.
1415          */
1416         int dl_throttled, dl_boosted, dl_yielded;
1417 
1418         /*
1419          * Bandwidth enforcement timer. Each -deadline task has its
1420          * own bandwidth to be enforced, thus we need one timer per task.
1421          */
1422         struct hrtimer dl_timer;
1423 };
1424 
1425 union rcu_special {
1426         struct {
1427                 u8 blocked;
1428                 u8 need_qs;
1429                 u8 exp_need_qs;
1430                 u8 pad; /* Otherwise the compiler can store garbage here. */
1431         } b; /* Bits. */
1432         u32 s; /* Set of bits. */
1433 };
1434 struct rcu_node;
1435 
1436 enum perf_event_task_context {
1437         perf_invalid_context = -1,
1438         perf_hw_context = 0,
1439         perf_sw_context,
1440         perf_nr_task_contexts,
1441 };
1442 
1443 /* Track pages that require TLB flushes */
1444 struct tlbflush_unmap_batch {
1445         /*
1446          * Each bit set is a CPU that potentially has a TLB entry for one of
1447          * the PFNs being flushed. See set_tlb_ubc_flush_pending().
1448          */
1449         struct cpumask cpumask;
1450 
1451         /* True if any bit in cpumask is set */
1452         bool flush_required;
1453 
1454         /*
1455          * If true then the PTE was dirty when unmapped. The entry must be
1456          * flushed before IO is initiated or a stale TLB entry potentially
1457          * allows an update without redirtying the page.
1458          */
1459         bool writable;
1460 };
1461 
1462 struct task_struct {
1463         volatile long state;    /* -1 unrunnable, 0 runnable, >0 stopped */
1464         void *stack;
1465         atomic_t usage;
1466         unsigned int flags;     /* per process flags, defined below */
1467         unsigned int ptrace;
1468 
1469 #ifdef CONFIG_SMP
1470         struct llist_node wake_entry;
1471         int on_cpu;
1472         unsigned int wakee_flips;
1473         unsigned long wakee_flip_decay_ts;
1474         struct task_struct *last_wakee;
1475 
1476         int wake_cpu;
1477 #endif
1478         int on_rq;
1479 
1480         int prio, static_prio, normal_prio;
1481         unsigned int rt_priority;
1482         const struct sched_class *sched_class;
1483         struct sched_entity se;
1484         struct sched_rt_entity rt;
1485 #ifdef CONFIG_CGROUP_SCHED
1486         struct task_group *sched_task_group;
1487 #endif
1488         struct sched_dl_entity dl;
1489 
1490 #ifdef CONFIG_PREEMPT_NOTIFIERS
1491         /* list of struct preempt_notifier: */
1492         struct hlist_head preempt_notifiers;
1493 #endif
1494 
1495 #ifdef CONFIG_BLK_DEV_IO_TRACE
1496         unsigned int btrace_seq;
1497 #endif
1498 
1499         unsigned int policy;
1500         int nr_cpus_allowed;
1501         cpumask_t cpus_allowed;
1502 
1503 #ifdef CONFIG_PREEMPT_RCU
1504         int rcu_read_lock_nesting;
1505         union rcu_special rcu_read_unlock_special;
1506         struct list_head rcu_node_entry;
1507         struct rcu_node *rcu_blocked_node;
1508 #endif /* #ifdef CONFIG_PREEMPT_RCU */
1509 #ifdef CONFIG_TASKS_RCU
1510         unsigned long rcu_tasks_nvcsw;
1511         bool rcu_tasks_holdout;
1512         struct list_head rcu_tasks_holdout_list;
1513         int rcu_tasks_idle_cpu;
1514 #endif /* #ifdef CONFIG_TASKS_RCU */
1515 
1516 #ifdef CONFIG_SCHED_INFO
1517         struct sched_info sched_info;
1518 #endif
1519 
1520         struct list_head tasks;
1521 #ifdef CONFIG_SMP
1522         struct plist_node pushable_tasks;
1523         struct rb_node pushable_dl_tasks;
1524 #endif
1525 
1526         struct mm_struct *mm, *active_mm;
1527         /* per-thread vma caching */
1528         u32 vmacache_seqnum;
1529         struct vm_area_struct *vmacache[VMACACHE_SIZE];
1530 #if defined(SPLIT_RSS_COUNTING)
1531         struct task_rss_stat    rss_stat;
1532 #endif
1533 /* task state */
1534         int exit_state;
1535         int exit_code, exit_signal;
1536         int pdeath_signal;  /*  The signal sent when the parent dies  */
1537         unsigned long jobctl;   /* JOBCTL_*, siglock protected */
1538 
1539         /* Used for emulating ABI behavior of previous Linux versions */
1540         unsigned int personality;
1541 
1542         /* scheduler bits, serialized by scheduler locks */
1543         unsigned sched_reset_on_fork:1;
1544         unsigned sched_contributes_to_load:1;
1545         unsigned sched_migrated:1;
1546         unsigned sched_remote_wakeup:1;
1547         unsigned :0; /* force alignment to the next boundary */
1548 
1549         /* unserialized, strictly 'current' */
1550         unsigned in_execve:1; /* bit to tell LSMs we're in execve */
1551         unsigned in_iowait:1;
1552 #if !defined(TIF_RESTORE_SIGMASK)
1553         unsigned restore_sigmask:1;
1554 #endif
1555 #ifdef CONFIG_MEMCG
1556         unsigned memcg_may_oom:1;
1557 #ifndef CONFIG_SLOB
1558         unsigned memcg_kmem_skip_account:1;
1559 #endif
1560 #endif
1561 #ifdef CONFIG_COMPAT_BRK
1562         unsigned brk_randomized:1;
1563 #endif
1564 
1565         unsigned long atomic_flags; /* Flags needing atomic access. */
1566 
1567         struct restart_block restart_block;
1568 
1569         pid_t pid;
1570         pid_t tgid;
1571 
1572 #ifdef CONFIG_CC_STACKPROTECTOR
1573         /* Canary value for the -fstack-protector gcc feature */
1574         unsigned long stack_canary;
1575 #endif
1576         /*
1577          * pointers to (original) parent process, youngest child, younger sibling,
1578          * older sibling, respectively.  (p->father can be replaced with
1579          * p->real_parent->pid)
1580          */
1581         struct task_struct __rcu *real_parent; /* real parent process */
1582         struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */
1583         /*
1584          * children/sibling forms the list of my natural children
1585          */
1586         struct list_head children;      /* list of my children */
1587         struct list_head sibling;       /* linkage in my parent's children list */
1588         struct task_struct *group_leader;       /* threadgroup leader */
1589 
1590         /*
1591          * ptraced is the list of tasks this task is using ptrace on.
1592          * This includes both natural children and PTRACE_ATTACH targets.
1593          * p->ptrace_entry is p's link on the p->parent->ptraced list.
1594          */
1595         struct list_head ptraced;
1596         struct list_head ptrace_entry;
1597 
1598         /* PID/PID hash table linkage. */
1599         struct pid_link pids[PIDTYPE_MAX];
1600         struct list_head thread_group;
1601         struct list_head thread_node;
1602 
1603         struct completion *vfork_done;          /* for vfork() */
1604         int __user *set_child_tid;              /* CLONE_CHILD_SETTID */
1605         int __user *clear_child_tid;            /* CLONE_CHILD_CLEARTID */
1606 
1607         cputime_t utime, stime, utimescaled, stimescaled;
1608         cputime_t gtime;
1609         struct prev_cputime prev_cputime;
1610 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
1611         seqcount_t vtime_seqcount;
1612         unsigned long long vtime_snap;
1613         enum {
1614                 /* Task is sleeping or running in a CPU with VTIME inactive */
1615                 VTIME_INACTIVE = 0,
1616                 /* Task runs in userspace in a CPU with VTIME active */
1617                 VTIME_USER,
1618                 /* Task runs in kernelspace in a CPU with VTIME active */
1619                 VTIME_SYS,
1620         } vtime_snap_whence;
1621 #endif
1622 
1623 #ifdef CONFIG_NO_HZ_FULL
1624         atomic_t tick_dep_mask;
1625 #endif
1626         unsigned long nvcsw, nivcsw; /* context switch counts */
1627         u64 start_time;         /* monotonic time in nsec */
1628         u64 real_start_time;    /* boot based time in nsec */
1629 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */
1630         unsigned long min_flt, maj_flt;
1631 
1632         struct task_cputime cputime_expires;
1633         struct list_head cpu_timers[3];
1634 
1635 /* process credentials */
1636         const struct cred __rcu *ptracer_cred; /* Tracer's credentials at attach */
1637         const struct cred __rcu *real_cred; /* objective and real subjective task
1638                                          * credentials (COW) */
1639         const struct cred __rcu *cred;  /* effective (overridable) subjective task
1640                                          * credentials (COW) */
1641         char comm[TASK_COMM_LEN]; /* executable name excluding path
1642                                      - access with [gs]et_task_comm (which lock
1643                                        it with task_lock())
1644                                      - initialized normally by setup_new_exec */
1645 /* file system info */
1646         struct nameidata *nameidata;
1647 #ifdef CONFIG_SYSVIPC
1648 /* ipc stuff */
1649         struct sysv_sem sysvsem;
1650         struct sysv_shm sysvshm;
1651 #endif
1652 #ifdef CONFIG_DETECT_HUNG_TASK
1653 /* hung task detection */
1654         unsigned long last_switch_count;
1655 #endif
1656 /* filesystem information */
1657         struct fs_struct *fs;
1658 /* open file information */
1659         struct files_struct *files;
1660 /* namespaces */
1661         struct nsproxy *nsproxy;
1662 /* signal handlers */
1663         struct signal_struct *signal;
1664         struct sighand_struct *sighand;
1665 
1666         sigset_t blocked, real_blocked;
1667         sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */
1668         struct sigpending pending;
1669 
1670         unsigned long sas_ss_sp;
1671         size_t sas_ss_size;
1672         unsigned sas_ss_flags;
1673 
1674         struct callback_head *task_works;
1675 
1676         struct audit_context *audit_context;
1677 #ifdef CONFIG_AUDITSYSCALL
1678         kuid_t loginuid;
1679         unsigned int sessionid;
1680 #endif
1681         struct seccomp seccomp;
1682 
1683 /* Thread group tracking */
1684         u32 parent_exec_id;
1685         u32 self_exec_id;
1686 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed,
1687  * mempolicy */
1688         spinlock_t alloc_lock;
1689 
1690         /* Protection of the PI data structures: */
1691         raw_spinlock_t pi_lock;
1692 
1693         struct wake_q_node wake_q;
1694 
1695 #ifdef CONFIG_RT_MUTEXES
1696         /* PI waiters blocked on a rt_mutex held by this task */
1697         struct rb_root pi_waiters;
1698         struct rb_node *pi_waiters_leftmost;
1699         /* Deadlock detection and priority inheritance handling */
1700         struct rt_mutex_waiter *pi_blocked_on;
1701 #endif
1702 
1703 #ifdef CONFIG_DEBUG_MUTEXES
1704         /* mutex deadlock detection */
1705         struct mutex_waiter *blocked_on;
1706 #endif
1707 #ifdef CONFIG_TRACE_IRQFLAGS
1708         unsigned int irq_events;
1709         unsigned long hardirq_enable_ip;
1710         unsigned long hardirq_disable_ip;
1711         unsigned int hardirq_enable_event;
1712         unsigned int hardirq_disable_event;
1713         int hardirqs_enabled;
1714         int hardirq_context;
1715         unsigned long softirq_disable_ip;
1716         unsigned long softirq_enable_ip;
1717         unsigned int softirq_disable_event;
1718         unsigned int softirq_enable_event;
1719         int softirqs_enabled;
1720         int softirq_context;
1721 #endif
1722 #ifdef CONFIG_LOCKDEP
1723 # define MAX_LOCK_DEPTH 48UL
1724         u64 curr_chain_key;
1725         int lockdep_depth;
1726         unsigned int lockdep_recursion;
1727         struct held_lock held_locks[MAX_LOCK_DEPTH];
1728         gfp_t lockdep_reclaim_gfp;
1729 #endif
1730 #ifdef CONFIG_UBSAN
1731         unsigned int in_ubsan;
1732 #endif
1733 
1734 /* journalling filesystem info */
1735         void *journal_info;
1736 
1737 /* stacked block device info */
1738         struct bio_list *bio_list;
1739 
1740 #ifdef CONFIG_BLOCK
1741 /* stack plugging */
1742         struct blk_plug *plug;
1743 #endif
1744 
1745 /* VM state */
1746         struct reclaim_state *reclaim_state;
1747 
1748         struct backing_dev_info *backing_dev_info;
1749 
1750         struct io_context *io_context;
1751 
1752         unsigned long ptrace_message;
1753         siginfo_t *last_siginfo; /* For ptrace use.  */
1754         struct task_io_accounting ioac;
1755 #if defined(CONFIG_TASK_XACCT)
1756         u64 acct_rss_mem1;      /* accumulated rss usage */
1757         u64 acct_vm_mem1;       /* accumulated virtual memory usage */
1758         cputime_t acct_timexpd; /* stime + utime since last update */
1759 #endif
1760 #ifdef CONFIG_CPUSETS
1761         nodemask_t mems_allowed;        /* Protected by alloc_lock */
1762         seqcount_t mems_allowed_seq;    /* Seqence no to catch updates */
1763         int cpuset_mem_spread_rotor;
1764         int cpuset_slab_spread_rotor;
1765 #endif
1766 #ifdef CONFIG_CGROUPS
1767         /* Control Group info protected by css_set_lock */
1768         struct css_set __rcu *cgroups;
1769         /* cg_list protected by css_set_lock and tsk->alloc_lock */
1770         struct list_head cg_list;
1771 #endif
1772 #ifdef CONFIG_FUTEX
1773         struct robust_list_head __user *robust_list;
1774 #ifdef CONFIG_COMPAT
1775         struct compat_robust_list_head __user *compat_robust_list;
1776 #endif
1777         struct list_head pi_state_list;
1778         struct futex_pi_state *pi_state_cache;
1779 #endif
1780 #ifdef CONFIG_PERF_EVENTS
1781         struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts];
1782         struct mutex perf_event_mutex;
1783         struct list_head perf_event_list;
1784 #endif
1785 #ifdef CONFIG_DEBUG_PREEMPT
1786         unsigned long preempt_disable_ip;
1787 #endif
1788 #ifdef CONFIG_NUMA
1789         struct mempolicy *mempolicy;    /* Protected by alloc_lock */
1790         short il_next;
1791         short pref_node_fork;
1792 #endif
1793 #ifdef CONFIG_NUMA_BALANCING
1794         int numa_scan_seq;
1795         unsigned int numa_scan_period;
1796         unsigned int numa_scan_period_max;
1797         int numa_preferred_nid;
1798         unsigned long numa_migrate_retry;
1799         u64 node_stamp;                 /* migration stamp  */
1800         u64 last_task_numa_placement;
1801         u64 last_sum_exec_runtime;
1802         struct callback_head numa_work;
1803 
1804         struct list_head numa_entry;
1805         struct numa_group *numa_group;
1806 
1807         /*
1808          * numa_faults is an array split into four regions:
1809          * faults_memory, faults_cpu, faults_memory_buffer, faults_cpu_buffer
1810          * in this precise order.
1811          *
1812          * faults_memory: Exponential decaying average of faults on a per-node
1813          * basis. Scheduling placement decisions are made based on these
1814          * counts. The values remain static for the duration of a PTE scan.
1815          * faults_cpu: Track the nodes the process was running on when a NUMA
1816          * hinting fault was incurred.
1817          * faults_memory_buffer and faults_cpu_buffer: Record faults per node
1818          * during the current scan window. When the scan completes, the counts
1819          * in faults_memory and faults_cpu decay and these values are copied.
1820          */
1821         unsigned long *numa_faults;
1822         unsigned long total_numa_faults;
1823 
1824         /*
1825          * numa_faults_locality tracks if faults recorded during the last
1826          * scan window were remote/local or failed to migrate. The task scan
1827          * period is adapted based on the locality of the faults with different
1828          * weights depending on whether they were shared or private faults
1829          */
1830         unsigned long numa_faults_locality[3];
1831 
1832         unsigned long numa_pages_migrated;
1833 #endif /* CONFIG_NUMA_BALANCING */
1834 
1835 #ifdef CONFIG_ARCH_WANT_BATCHED_UNMAP_TLB_FLUSH
1836         struct tlbflush_unmap_batch tlb_ubc;
1837 #endif
1838 
1839         struct rcu_head rcu;
1840 
1841         /*
1842          * cache last used pipe for splice
1843          */
1844         struct pipe_inode_info *splice_pipe;
1845 
1846         struct page_frag task_frag;
1847 
1848 #ifdef  CONFIG_TASK_DELAY_ACCT
1849         struct task_delay_info *delays;
1850 #endif
1851 #ifdef CONFIG_FAULT_INJECTION
1852         int make_it_fail;
1853 #endif
1854         /*
1855          * when (nr_dirtied >= nr_dirtied_pause), it's time to call
1856          * balance_dirty_pages() for some dirty throttling pause
1857          */
1858         int nr_dirtied;
1859         int nr_dirtied_pause;
1860         unsigned long dirty_paused_when; /* start of a write-and-pause period */
1861 
1862 #ifdef CONFIG_LATENCYTOP
1863         int latency_record_count;
1864         struct latency_record latency_record[LT_SAVECOUNT];
1865 #endif
1866         /*
1867          * time slack values; these are used to round up poll() and
1868          * select() etc timeout values. These are in nanoseconds.
1869          */
1870         u64 timer_slack_ns;
1871         u64 default_timer_slack_ns;
1872 
1873 #ifdef CONFIG_KASAN
1874         unsigned int kasan_depth;
1875 #endif
1876 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
1877         /* Index of current stored address in ret_stack */
1878         int curr_ret_stack;
1879         /* Stack of return addresses for return function tracing */
1880         struct ftrace_ret_stack *ret_stack;
1881         /* time stamp for last schedule */
1882         unsigned long long ftrace_timestamp;
1883         /*
1884          * Number of functions that haven't been traced
1885          * because of depth overrun.
1886          */
1887         atomic_t trace_overrun;
1888         /* Pause for the tracing */
1889         atomic_t tracing_graph_pause;
1890 #endif
1891 #ifdef CONFIG_TRACING
1892         /* state flags for use by tracers */
1893         unsigned long trace;
1894         /* bitmask and counter of trace recursion */
1895         unsigned long trace_recursion;
1896 #endif /* CONFIG_TRACING */
1897 #ifdef CONFIG_KCOV
1898         /* Coverage collection mode enabled for this task (0 if disabled). */
1899         enum kcov_mode kcov_mode;
1900         /* Size of the kcov_area. */
1901         unsigned        kcov_size;
1902         /* Buffer for coverage collection. */
1903         void            *kcov_area;
1904         /* kcov desciptor wired with this task or NULL. */
1905         struct kcov     *kcov;
1906 #endif
1907 #ifdef CONFIG_MEMCG
1908         struct mem_cgroup *memcg_in_oom;
1909         gfp_t memcg_oom_gfp_mask;
1910         int memcg_oom_order;
1911 
1912         /* number of pages to reclaim on returning to userland */
1913         unsigned int memcg_nr_pages_over_high;
1914 #endif
1915 #ifdef CONFIG_UPROBES
1916         struct uprobe_task *utask;
1917 #endif
1918 #if defined(CONFIG_BCACHE) || defined(CONFIG_BCACHE_MODULE)
1919         unsigned int    sequential_io;
1920         unsigned int    sequential_io_avg;
1921 #endif
1922 #ifdef CONFIG_DEBUG_ATOMIC_SLEEP
1923         unsigned long   task_state_change;
1924 #endif
1925         int pagefault_disabled;
1926 #ifdef CONFIG_MMU
1927         struct task_struct *oom_reaper_list;
1928 #endif
1929 #if defined(CONFIG_CCSECURITY) && !defined(CONFIG_CCSECURITY_USE_EXTERNAL_TASK_SECURITY)
1930         struct ccs_domain_info *ccs_domain_info;
1931         u32 ccs_flags;
1932 #endif
1933 /* CPU-specific state of this task */
1934         struct thread_struct thread;
1935 /*
1936  * WARNING: on x86, 'thread_struct' contains a variable-sized
1937  * structure.  It *MUST* be at the end of 'task_struct'.
1938  *
1939  * Do not put anything below here!
1940  */
1941 };
1942 
1943 #ifdef CONFIG_ARCH_WANTS_DYNAMIC_TASK_STRUCT
1944 extern int arch_task_struct_size __read_mostly;
1945 #else
1946 # define arch_task_struct_size (sizeof(struct task_struct))
1947 #endif
1948 
1949 /* Future-safe accessor for struct task_struct's cpus_allowed. */
1950 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed)
1951 
1952 static inline int tsk_nr_cpus_allowed(struct task_struct *p)
1953 {
1954         return p->nr_cpus_allowed;
1955 }
1956 
1957 #define TNF_MIGRATED    0x01
1958 #define TNF_NO_GROUP    0x02
1959 #define TNF_SHARED      0x04
1960 #define TNF_FAULT_LOCAL 0x08
1961 #define TNF_MIGRATE_FAIL 0x10
1962 
1963 static inline bool in_vfork(struct task_struct *tsk)
1964 {
1965         bool ret;
1966 
1967         /*
1968          * need RCU to access ->real_parent if CLONE_VM was used along with
1969          * CLONE_PARENT.
1970          *
1971          * We check real_parent->mm == tsk->mm because CLONE_VFORK does not
1972          * imply CLONE_VM
1973          *
1974          * CLONE_VFORK can be used with CLONE_PARENT/CLONE_THREAD and thus
1975          * ->real_parent is not necessarily the task doing vfork(), so in
1976          * theory we can't rely on task_lock() if we want to dereference it.
1977          *
1978          * And in this case we can't trust the real_parent->mm == tsk->mm
1979          * check, it can be false negative. But we do not care, if init or
1980          * another oom-unkillable task does this it should blame itself.
1981          */
1982         rcu_read_lock();
1983         ret = tsk->vfork_done && tsk->real_parent->mm == tsk->mm;
1984         rcu_read_unlock();
1985 
1986         return ret;
1987 }
1988 
1989 #ifdef CONFIG_NUMA_BALANCING
1990 extern void task_numa_fault(int last_node, int node, int pages, int flags);
1991 extern pid_t task_numa_group_id(struct task_struct *p);
1992 extern void set_numabalancing_state(bool enabled);
1993 extern void task_numa_free(struct task_struct *p);
1994 extern bool should_numa_migrate_memory(struct task_struct *p, struct page *page,
1995                                         int src_nid, int dst_cpu);
1996 #else
1997 static inline void task_numa_fault(int last_node, int node, int pages,
1998                                    int flags)
1999 {
2000 }
2001 static inline pid_t task_numa_group_id(struct task_struct *p)
2002 {
2003         return 0;
2004 }
2005 static inline void set_numabalancing_state(bool enabled)
2006 {
2007 }
2008 static inline void task_numa_free(struct task_struct *p)
2009 {
2010 }
2011 static inline bool should_numa_migrate_memory(struct task_struct *p,
2012                                 struct page *page, int src_nid, int dst_cpu)
2013 {
2014         return true;
2015 }
2016 #endif
2017 
2018 static inline struct pid *task_pid(struct task_struct *task)
2019 {
2020         return task->pids[PIDTYPE_PID].pid;
2021 }
2022 
2023 static inline struct pid *task_tgid(struct task_struct *task)
2024 {
2025         return task->group_leader->pids[PIDTYPE_PID].pid;
2026 }
2027 
2028 /*
2029  * Without tasklist or rcu lock it is not safe to dereference
2030  * the result of task_pgrp/task_session even if task == current,
2031  * we can race with another thread doing sys_setsid/sys_setpgid.
2032  */
2033 static inline struct pid *task_pgrp(struct task_struct *task)
2034 {
2035         return task->group_leader->pids[PIDTYPE_PGID].pid;
2036 }
2037 
2038 static inline struct pid *task_session(struct task_struct *task)
2039 {
2040         return task->group_leader->pids[PIDTYPE_SID].pid;
2041 }
2042 
2043 struct pid_namespace;
2044 
2045 /*
2046  * the helpers to get the task's different pids as they are seen
2047  * from various namespaces
2048  *
2049  * task_xid_nr()     : global id, i.e. the id seen from the init namespace;
2050  * task_xid_vnr()    : virtual id, i.e. the id seen from the pid namespace of
2051  *                     current.
2052  * task_xid_nr_ns()  : id seen from the ns specified;
2053  *
2054  * set_task_vxid()   : assigns a virtual id to a task;
2055  *
2056  * see also pid_nr() etc in include/linux/pid.h
2057  */
2058 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type,
2059                         struct pid_namespace *ns);
2060 
2061 static inline pid_t task_pid_nr(struct task_struct *tsk)
2062 {
2063         return tsk->pid;
2064 }
2065 
2066 static inline pid_t task_pid_nr_ns(struct task_struct *tsk,
2067                                         struct pid_namespace *ns)
2068 {
2069         return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns);
2070 }
2071 
2072 static inline pid_t task_pid_vnr(struct task_struct *tsk)
2073 {
2074         return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL);
2075 }
2076 
2077 
2078 static inline pid_t task_tgid_nr(struct task_struct *tsk)
2079 {
2080         return tsk->tgid;
2081 }
2082 
2083 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns);
2084 
2085 static inline pid_t task_tgid_vnr(struct task_struct *tsk)
2086 {
2087         return pid_vnr(task_tgid(tsk));
2088 }
2089 
2090 
2091 static inline int pid_alive(const struct task_struct *p);
2092 static inline pid_t task_ppid_nr_ns(const struct task_struct *tsk, struct pid_namespace *ns)
2093 {
2094         pid_t pid = 0;
2095 
2096         rcu_read_lock();
2097         if (pid_alive(tsk))
2098                 pid = task_tgid_nr_ns(rcu_dereference(tsk->real_parent), ns);
2099         rcu_read_unlock();
2100 
2101         return pid;
2102 }
2103 
2104 static inline pid_t task_ppid_nr(const struct task_struct *tsk)
2105 {
2106         return task_ppid_nr_ns(tsk, &init_pid_ns);
2107 }
2108 
2109 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk,
2110                                         struct pid_namespace *ns)
2111 {
2112         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns);
2113 }
2114 
2115 static inline pid_t task_pgrp_vnr(struct task_struct *tsk)
2116 {
2117         return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL);
2118 }
2119 
2120 
2121 static inline pid_t task_session_nr_ns(struct task_struct *tsk,
2122                                         struct pid_namespace *ns)
2123 {
2124         return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns);
2125 }
2126 
2127 static inline pid_t task_session_vnr(struct task_struct *tsk)
2128 {
2129         return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL);
2130 }
2131 
2132 /* obsolete, do not use */
2133 static inline pid_t task_pgrp_nr(struct task_struct *tsk)
2134 {
2135         return task_pgrp_nr_ns(tsk, &init_pid_ns);
2136 }
2137 
2138 /**
2139  * pid_alive - check that a task structure is not stale
2140  * @p: Task structure to be checked.
2141  *
2142  * Test if a process is not yet dead (at most zombie state)
2143  * If pid_alive fails, then pointers within the task structure
2144  * can be stale and must not be dereferenced.
2145  *
2146  * Return: 1 if the process is alive. 0 otherwise.
2147  */
2148 static inline int pid_alive(const struct task_struct *p)
2149 {
2150         return p->pids[PIDTYPE_PID].pid != NULL;
2151 }
2152 
2153 /**
2154  * is_global_init - check if a task structure is init. Since init
2155  * is free to have sub-threads we need to check tgid.
2156  * @tsk: Task structure to be checked.
2157  *
2158  * Check if a task structure is the first user space task the kernel created.
2159  *
2160  * Return: 1 if the task structure is init. 0 otherwise.
2161  */
2162 static inline int is_global_init(struct task_struct *tsk)
2163 {
2164         return task_tgid_nr(tsk) == 1;
2165 }
2166 
2167 extern struct pid *cad_pid;
2168 
2169 extern void free_task(struct task_struct *tsk);
2170 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0)
2171 
2172 extern void __put_task_struct(struct task_struct *t);
2173 
2174 static inline void put_task_struct(struct task_struct *t)
2175 {
2176         if (atomic_dec_and_test(&t->usage))
2177                 __put_task_struct(t);
2178 }
2179 
2180 struct task_struct *task_rcu_dereference(struct task_struct **ptask);
2181 struct task_struct *try_get_task_struct(struct task_struct **ptask);
2182 
2183 #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN
2184 extern void task_cputime(struct task_struct *t,
2185                          cputime_t *utime, cputime_t *stime);
2186 extern void task_cputime_scaled(struct task_struct *t,
2187                                 cputime_t *utimescaled, cputime_t *stimescaled);
2188 extern cputime_t task_gtime(struct task_struct *t);
2189 #else
2190 static inline void task_cputime(struct task_struct *t,
2191                                 cputime_t *utime, cputime_t *stime)
2192 {
2193         if (utime)
2194                 *utime = t->utime;
2195         if (stime)
2196                 *stime = t->stime;
2197 }
2198 
2199 static inline void task_cputime_scaled(struct task_struct *t,
2200                                        cputime_t *utimescaled,
2201                                        cputime_t *stimescaled)
2202 {
2203         if (utimescaled)
2204                 *utimescaled = t->utimescaled;
2205         if (stimescaled)
2206                 *stimescaled = t->stimescaled;
2207 }
2208 
2209 static inline cputime_t task_gtime(struct task_struct *t)
2210 {
2211         return t->gtime;
2212 }
2213 #endif
2214 extern void task_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2215 extern void thread_group_cputime_adjusted(struct task_struct *p, cputime_t *ut, cputime_t *st);
2216 
2217 /*
2218  * Per process flags
2219  */
2220 #define PF_EXITING      0x00000004      /* getting shut down */
2221 #define PF_EXITPIDONE   0x00000008      /* pi exit done on shut down */
2222 #define PF_VCPU         0x00000010      /* I'm a virtual CPU */
2223 #define PF_WQ_WORKER    0x00000020      /* I'm a workqueue worker */
2224 #define PF_FORKNOEXEC   0x00000040      /* forked but didn't exec */
2225 #define PF_MCE_PROCESS  0x00000080      /* process policy on mce errors */
2226 #define PF_SUPERPRIV    0x00000100      /* used super-user privileges */
2227 #define PF_DUMPCORE     0x00000200      /* dumped core */
2228 #define PF_SIGNALED     0x00000400      /* killed by a signal */
2229 #define PF_MEMALLOC     0x00000800      /* Allocating memory */
2230 #define PF_NPROC_EXCEEDED 0x00001000    /* set_user noticed that RLIMIT_NPROC was exceeded */
2231 #define PF_USED_MATH    0x00002000      /* if unset the fpu must be initialized before use */
2232 #define PF_USED_ASYNC   0x00004000      /* used async_schedule*(), used by module init */
2233 #define PF_NOFREEZE     0x00008000      /* this thread should not be frozen */
2234 #define PF_FROZEN       0x00010000      /* frozen for system suspend */
2235 #define PF_FSTRANS      0x00020000      /* inside a filesystem transaction */
2236 #define PF_KSWAPD       0x00040000      /* I am kswapd */
2237 #define PF_MEMALLOC_NOIO 0x00080000     /* Allocating memory without IO involved */
2238 #define PF_LESS_THROTTLE 0x00100000     /* Throttle me less: I clean memory */
2239 #define PF_KTHREAD      0x00200000      /* I am a kernel thread */
2240 #define PF_RANDOMIZE    0x00400000      /* randomize virtual address space */
2241 #define PF_SWAPWRITE    0x00800000      /* Allowed to write to swap */
2242 #define PF_NO_SETAFFINITY 0x04000000    /* Userland is not allowed to meddle with cpus_allowed */
2243 #define PF_MCE_EARLY    0x08000000      /* Early kill for mce process policy */
2244 #define PF_MUTEX_TESTER 0x20000000      /* Thread belongs to the rt mutex tester */
2245 #define PF_FREEZER_SKIP 0x40000000      /* Freezer should not count it as freezable */
2246 #define PF_SUSPEND_TASK 0x80000000      /* this thread called freeze_processes and should not be frozen */
2247 
2248 /*
2249  * Only the _current_ task can read/write to tsk->flags, but other
2250  * tasks can access tsk->flags in readonly mode for example
2251  * with tsk_used_math (like during threaded core dumping).
2252  * There is however an exception to this rule during ptrace
2253  * or during fork: the ptracer task is allowed to write to the
2254  * child->flags of its traced child (same goes for fork, the parent
2255  * can write to the child->flags), because we're guaranteed the
2256  * child is not running and in turn not changing child->flags
2257  * at the same time the parent does it.
2258  */
2259 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0)
2260 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0)
2261 #define clear_used_math() clear_stopped_child_used_math(current)
2262 #define set_used_math() set_stopped_child_used_math(current)
2263 #define conditional_stopped_child_used_math(condition, child) \
2264         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0)
2265 #define conditional_used_math(condition) \
2266         conditional_stopped_child_used_math(condition, current)
2267 #define copy_to_stopped_child_used_math(child) \
2268         do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0)
2269 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */
2270 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH)
2271 #define used_math() tsk_used_math(current)
2272 
2273 /* __GFP_IO isn't allowed if PF_MEMALLOC_NOIO is set in current->flags
2274  * __GFP_FS is also cleared as it implies __GFP_IO.
2275  */
2276 static inline gfp_t memalloc_noio_flags(gfp_t flags)
2277 {
2278         if (unlikely(current->flags & PF_MEMALLOC_NOIO))
2279                 flags &= ~(__GFP_IO | __GFP_FS);
2280         return flags;
2281 }
2282 
2283 static inline unsigned int memalloc_noio_save(void)
2284 {
2285         unsigned int flags = current->flags & PF_MEMALLOC_NOIO;
2286         current->flags |= PF_MEMALLOC_NOIO;
2287         return flags;
2288 }
2289 
2290 static inline void memalloc_noio_restore(unsigned int flags)
2291 {
2292         current->flags = (current->flags & ~PF_MEMALLOC_NOIO) | flags;
2293 }
2294 
2295 /* Per-process atomic flags. */
2296 #define PFA_NO_NEW_PRIVS 0      /* May not gain new privileges. */
2297 #define PFA_SPREAD_PAGE  1      /* Spread page cache over cpuset */
2298 #define PFA_SPREAD_SLAB  2      /* Spread some slab caches over cpuset */
2299 #define PFA_LMK_WAITING  3      /* Lowmemorykiller is waiting */
2300 
2301 
2302 #define TASK_PFA_TEST(name, func)                                       \
2303         static inline bool task_##func(struct task_struct *p)           \
2304         { return test_bit(PFA_##name, &p->atomic_flags); }
2305 #define TASK_PFA_SET(name, func)                                        \
2306         static inline void task_set_##func(struct task_struct *p)       \
2307         { set_bit(PFA_##name, &p->atomic_flags); }
2308 #define TASK_PFA_CLEAR(name, func)                                      \
2309         static inline void task_clear_##func(struct task_struct *p)     \
2310         { clear_bit(PFA_##name, &p->atomic_flags); }
2311 
2312 TASK_PFA_TEST(NO_NEW_PRIVS, no_new_privs)
2313 TASK_PFA_SET(NO_NEW_PRIVS, no_new_privs)
2314 
2315 TASK_PFA_TEST(SPREAD_PAGE, spread_page)
2316 TASK_PFA_SET(SPREAD_PAGE, spread_page)
2317 TASK_PFA_CLEAR(SPREAD_PAGE, spread_page)
2318 
2319 TASK_PFA_TEST(SPREAD_SLAB, spread_slab)
2320 TASK_PFA_SET(SPREAD_SLAB, spread_slab)
2321 TASK_PFA_CLEAR(SPREAD_SLAB, spread_slab)
2322 
2323 TASK_PFA_TEST(LMK_WAITING, lmk_waiting)
2324 TASK_PFA_SET(LMK_WAITING, lmk_waiting)
2325 
2326 /*
2327  * task->jobctl flags
2328  */
2329 #define JOBCTL_STOP_SIGMASK     0xffff  /* signr of the last group stop */
2330 
2331 #define JOBCTL_STOP_DEQUEUED_BIT 16     /* stop signal dequeued */
2332 #define JOBCTL_STOP_PENDING_BIT 17      /* task should stop for group stop */
2333 #define JOBCTL_STOP_CONSUME_BIT 18      /* consume group stop count */
2334 #define JOBCTL_TRAP_STOP_BIT    19      /* trap for STOP */
2335 #define JOBCTL_TRAP_NOTIFY_BIT  20      /* trap for NOTIFY */
2336 #define JOBCTL_TRAPPING_BIT     21      /* switching to TRACED */
2337 #define JOBCTL_LISTENING_BIT    22      /* ptracer is listening for events */
2338 
2339 #define JOBCTL_STOP_DEQUEUED    (1UL << JOBCTL_STOP_DEQUEUED_BIT)
2340 #define JOBCTL_STOP_PENDING     (1UL << JOBCTL_STOP_PENDING_BIT)
2341 #define JOBCTL_STOP_CONSUME     (1UL << JOBCTL_STOP_CONSUME_BIT)
2342 #define JOBCTL_TRAP_STOP        (1UL << JOBCTL_TRAP_STOP_BIT)
2343 #define JOBCTL_TRAP_NOTIFY      (1UL << JOBCTL_TRAP_NOTIFY_BIT)
2344 #define JOBCTL_TRAPPING         (1UL << JOBCTL_TRAPPING_BIT)
2345 #define JOBCTL_LISTENING        (1UL << JOBCTL_LISTENING_BIT)
2346 
2347 #define JOBCTL_TRAP_MASK        (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY)
2348 #define JOBCTL_PENDING_MASK     (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK)
2349 
2350 extern bool task_set_jobctl_pending(struct task_struct *task,
2351                                     unsigned long mask);
2352 extern void task_clear_jobctl_trapping(struct task_struct *task);
2353 extern void task_clear_jobctl_pending(struct task_struct *task,
2354                                       unsigned long mask);
2355 
2356 static inline void rcu_copy_process(struct task_struct *p)
2357 {
2358 #ifdef CONFIG_PREEMPT_RCU
2359         p->rcu_read_lock_nesting = 0;
2360         p->rcu_read_unlock_special.s = 0;
2361         p->rcu_blocked_node = NULL;
2362         INIT_LIST_HEAD(&p->rcu_node_entry);
2363 #endif /* #ifdef CONFIG_PREEMPT_RCU */
2364 #ifdef CONFIG_TASKS_RCU
2365         p->rcu_tasks_holdout = false;
2366         INIT_LIST_HEAD(&p->rcu_tasks_holdout_list);
2367         p->rcu_tasks_idle_cpu = -1;
2368 #endif /* #ifdef CONFIG_TASKS_RCU */
2369 }
2370 
2371 static inline void tsk_restore_flags(struct task_struct *task,
2372                                 unsigned long orig_flags, unsigned long flags)
2373 {
2374         task->flags &= ~flags;
2375         task->flags |= orig_flags & flags;
2376 }
2377 
2378 extern int cpuset_cpumask_can_shrink(const struct cpumask *cur,
2379                                      const struct cpumask *trial);
2380 extern int task_can_attach(struct task_struct *p,
2381                            const struct cpumask *cs_cpus_allowed);
2382 #ifdef CONFIG_SMP
2383 extern void do_set_cpus_allowed(struct task_struct *p,
2384                                const struct cpumask *new_mask);
2385 
2386 extern int set_cpus_allowed_ptr(struct task_struct *p,
2387                                 const struct cpumask *new_mask);
2388 #else
2389 static inline void do_set_cpus_allowed(struct task_struct *p,
2390                                       const struct cpumask *new_mask)
2391 {
2392 }
2393 static inline int set_cpus_allowed_ptr(struct task_struct *p,
2394                                        const struct cpumask *new_mask)
2395 {
2396         if (!cpumask_test_cpu(0, new_mask))
2397                 return -EINVAL;
2398         return 0;
2399 }
2400 #endif
2401 
2402 #ifdef CONFIG_NO_HZ_COMMON
2403 void calc_load_enter_idle(void);
2404 void calc_load_exit_idle(void);
2405 #else
2406 static inline void calc_load_enter_idle(void) { }
2407 static inline void calc_load_exit_idle(void) { }
2408 #endif /* CONFIG_NO_HZ_COMMON */
2409 
2410 /*
2411  * Do not use outside of architecture code which knows its limitations.
2412  *
2413  * sched_clock() has no promise of monotonicity or bounded drift between
2414  * CPUs, use (which you should not) requires disabling IRQs.
2415  *
2416  * Please use one of the three interfaces below.
2417  */
2418 extern unsigned long long notrace sched_clock(void);
2419 /*
2420  * See the comment in kernel/sched/clock.c
2421  */
2422 extern u64 running_clock(void);
2423 extern u64 sched_clock_cpu(int cpu);
2424 
2425 
2426 extern void sched_clock_init(void);
2427 
2428 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
2429 static inline void sched_clock_tick(void)
2430 {
2431 }
2432 
2433 static inline void sched_clock_idle_sleep_event(void)
2434 {
2435 }
2436 
2437 static inline void sched_clock_idle_wakeup_event(u64 delta_ns)
2438 {
2439 }
2440 
2441 static inline u64 cpu_clock(int cpu)
2442 {
2443         return sched_clock();
2444 }
2445 
2446 static inline u64 local_clock(void)
2447 {
2448         return sched_clock();
2449 }
2450 #else
2451 /*
2452  * Architectures can set this to 1 if they have specified
2453  * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig,
2454  * but then during bootup it turns out that sched_clock()
2455  * is reliable after all:
2456  */
2457 extern int sched_clock_stable(void);
2458 extern void set_sched_clock_stable(void);
2459 extern void clear_sched_clock_stable(void);
2460 
2461 extern void sched_clock_tick(void);
2462 extern void sched_clock_idle_sleep_event(void);
2463 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2464 
2465 /*
2466  * As outlined in clock.c, provides a fast, high resolution, nanosecond
2467  * time source that is monotonic per cpu argument and has bounded drift
2468  * between cpus.
2469  *
2470  * ######################### BIG FAT WARNING ##########################
2471  * # when comparing cpu_clock(i) to cpu_clock(j) for i != j, time can #
2472  * # go backwards !!                                                  #
2473  * ####################################################################
2474  */
2475 static inline u64 cpu_clock(int cpu)
2476 {
2477         return sched_clock_cpu(cpu);
2478 }
2479 
2480 static inline u64 local_clock(void)
2481 {
2482         return sched_clock_cpu(raw_smp_processor_id());
2483 }
2484 #endif
2485 
2486 #ifdef CONFIG_IRQ_TIME_ACCOUNTING
2487 /*
2488  * An i/f to runtime opt-in for irq time accounting based off of sched_clock.
2489  * The reason for this explicit opt-in is not to have perf penalty with
2490  * slow sched_clocks.
2491  */
2492 extern void enable_sched_clock_irqtime(void);
2493 extern void disable_sched_clock_irqtime(void);
2494 #else
2495 static inline void enable_sched_clock_irqtime(void) {}
2496 static inline void disable_sched_clock_irqtime(void) {}
2497 #endif
2498 
2499 extern unsigned long long
2500 task_sched_runtime(struct task_struct *task);
2501 
2502 /* sched_exec is called by processes performing an exec */
2503 #ifdef CONFIG_SMP
2504 extern void sched_exec(void);
2505 #else
2506 #define sched_exec()   {}
2507 #endif
2508 
2509 extern void sched_clock_idle_sleep_event(void);
2510 extern void sched_clock_idle_wakeup_event(u64 delta_ns);
2511 
2512 #ifdef CONFIG_HOTPLUG_CPU
2513 extern void idle_task_exit(void);
2514 #else
2515 static inline void idle_task_exit(void) {}
2516 #endif
2517 
2518 #if defined(CONFIG_NO_HZ_COMMON) && defined(CONFIG_SMP)
2519 extern void wake_up_nohz_cpu(int cpu);
2520 #else
2521 static inline void wake_up_nohz_cpu(int cpu) { }
2522 #endif
2523 
2524 #ifdef CONFIG_NO_HZ_FULL
2525 extern u64 scheduler_tick_max_deferment(void);
2526 #endif
2527 
2528 #ifdef CONFIG_SCHED_AUTOGROUP
2529 extern void sched_autogroup_create_attach(struct task_struct *p);
2530 extern void sched_autogroup_detach(struct task_struct *p);
2531 extern void sched_autogroup_fork(struct signal_struct *sig);
2532 extern void sched_autogroup_exit(struct signal_struct *sig);
2533 #ifdef CONFIG_PROC_FS
2534 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m);
2535 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice);
2536 #endif
2537 #else
2538 static inline void sched_autogroup_create_attach(struct task_struct *p) { }
2539 static inline void sched_autogroup_detach(struct task_struct *p) { }
2540 static inline void sched_autogroup_fork(struct signal_struct *sig) { }
2541 static inline void sched_autogroup_exit(struct signal_struct *sig) { }
2542 #endif
2543 
2544 extern int yield_to(struct task_struct *p, bool preempt);
2545 extern void set_user_nice(struct task_struct *p, long nice);
2546 extern int task_prio(const struct task_struct *p);
2547 /**
2548  * task_nice - return the nice value of a given task.
2549  * @p: the task in question.
2550  *
2551  * Return: The nice value [ -20 ... 0 ... 19 ].
2552  */
2553 static inline int task_nice(const struct task_struct *p)
2554 {
2555         return PRIO_TO_NICE((p)->static_prio);
2556 }
2557 extern int can_nice(const struct task_struct *p, const int nice);
2558 extern int task_curr(const struct task_struct *p);
2559 extern int idle_cpu(int cpu);
2560 extern int sched_setscheduler(struct task_struct *, int,
2561                               const struct sched_param *);
2562 extern int sched_setscheduler_nocheck(struct task_struct *, int,
2563                                       const struct sched_param *);
2564 extern int sched_setattr(struct task_struct *,
2565                          const struct sched_attr *);
2566 extern struct task_struct *idle_task(int cpu);
2567 /**
2568  * is_idle_task - is the specified task an idle task?
2569  * @p: the task in question.
2570  *
2571  * Return: 1 if @p is an idle task. 0 otherwise.
2572  */
2573 static inline bool is_idle_task(const struct task_struct *p)
2574 {
2575         return p->pid == 0;
2576 }
2577 extern struct task_struct *curr_task(int cpu);
2578 extern void set_curr_task(int cpu, struct task_struct *p);
2579 
2580 void yield(void);
2581 
2582 union thread_union {
2583         struct thread_info thread_info;
2584         unsigned long stack[THREAD_SIZE/sizeof(long)];
2585 };
2586 
2587 #ifndef __HAVE_ARCH_KSTACK_END
2588 static inline int kstack_end(void *addr)
2589 {
2590         /* Reliable end of stack detection:
2591          * Some APM bios versions misalign the stack
2592          */
2593         return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*)));
2594 }
2595 #endif
2596 
2597 extern union thread_union init_thread_union;
2598 extern struct task_struct init_task;
2599 
2600 extern struct   mm_struct init_mm;
2601 
2602 extern struct pid_namespace init_pid_ns;
2603 
2604 /*
2605  * find a task by one of its numerical ids
2606  *
2607  * find_task_by_pid_ns():
2608  *      finds a task by its pid in the specified namespace
2609  * find_task_by_vpid():
2610  *      finds a task by its virtual pid
2611  *
2612  * see also find_vpid() etc in include/linux/pid.h
2613  */
2614 
2615 extern struct task_struct *find_task_by_vpid(pid_t nr);
2616 extern struct task_struct *find_task_by_pid_ns(pid_t nr,
2617                 struct pid_namespace *ns);
2618 
2619 /* per-UID process charging. */
2620 extern struct user_struct * alloc_uid(kuid_t);
2621 static inline struct user_struct *get_uid(struct user_struct *u)
2622 {
2623         atomic_inc(&u->__count);
2624         return u;
2625 }
2626 extern void free_uid(struct user_struct *);
2627 
2628 #include <asm/current.h>
2629 
2630 extern void xtime_update(unsigned long ticks);
2631 
2632 extern int wake_up_state(struct task_struct *tsk, unsigned int state);
2633 extern int wake_up_process(struct task_struct *tsk);
2634 extern void wake_up_new_task(struct task_struct *tsk);
2635 #ifdef CONFIG_SMP
2636  extern void kick_process(struct task_struct *tsk);
2637 #else
2638  static inline void kick_process(struct task_struct *tsk) { }
2639 #endif
2640 extern int sched_fork(unsigned long clone_flags, struct task_struct *p);
2641 extern void sched_dead(struct task_struct *p);
2642 
2643 extern void proc_caches_init(void);
2644 extern void flush_signals(struct task_struct *);
2645 extern void ignore_signals(struct task_struct *);
2646 extern void flush_signal_handlers(struct task_struct *, int force_default);
2647 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info);
2648 
2649 static inline int kernel_dequeue_signal(siginfo_t *info)
2650 {
2651         struct task_struct *tsk = current;
2652         siginfo_t __info;
2653         int ret;
2654 
2655         spin_lock_irq(&tsk->sighand->siglock);
2656         ret = dequeue_signal(tsk, &tsk->blocked, info ?: &__info);
2657         spin_unlock_irq(&tsk->sighand->siglock);
2658 
2659         return ret;
2660 }
2661 
2662 static inline void kernel_signal_stop(void)
2663 {
2664         spin_lock_irq(&current->sighand->siglock);
2665         if (current->jobctl & JOBCTL_STOP_DEQUEUED)
2666                 __set_current_state(TASK_STOPPED);
2667         spin_unlock_irq(&current->sighand->siglock);
2668 
2669         schedule();
2670 }
2671 
2672 extern void release_task(struct task_struct * p);
2673 extern int send_sig_info(int, struct siginfo *, struct task_struct *);
2674 extern int force_sigsegv(int, struct task_struct *);
2675 extern int force_sig_info(int, struct siginfo *, struct task_struct *);
2676 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp);
2677 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid);
2678 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *,
2679                                 const struct cred *, u32);
2680 extern int kill_pgrp(struct pid *pid, int sig, int priv);
2681 extern int kill_pid(struct pid *pid, int sig, int priv);
2682 extern int kill_proc_info(int, struct siginfo *, pid_t);
2683 extern __must_check bool do_notify_parent(struct task_struct *, int);
2684 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent);
2685 extern void force_sig(int, struct task_struct *);
2686 extern int send_sig(int, struct task_struct *, int);
2687 extern int zap_other_threads(struct task_struct *p);
2688 extern struct sigqueue *sigqueue_alloc(void);
2689 extern void sigqueue_free(struct sigqueue *);
2690 extern int send_sigqueue(struct sigqueue *,  struct task_struct *, int group);
2691 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *);
2692 
2693 #ifdef TIF_RESTORE_SIGMASK
2694 /*
2695  * Legacy restore_sigmask accessors.  These are inefficient on
2696  * SMP architectures because they require atomic operations.
2697  */
2698 
2699 /**
2700  * set_restore_sigmask() - make sure saved_sigmask processing gets done
2701  *
2702  * This sets TIF_RESTORE_SIGMASK and ensures that the arch signal code
2703  * will run before returning to user mode, to process the flag.  For
2704  * all callers, TIF_SIGPENDING is already set or it's no harm to set
2705  * it.  TIF_RESTORE_SIGMASK need not be in the set of bits that the
2706  * arch code will notice on return to user mode, in case those bits
2707  * are scarce.  We set TIF_SIGPENDING here to ensure that the arch
2708  * signal code always gets run when TIF_RESTORE_SIGMASK is set.
2709  */
2710 static inline void set_restore_sigmask(void)
2711 {
2712         set_thread_flag(TIF_RESTORE_SIGMASK);
2713         WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2714 }
2715 static inline void clear_restore_sigmask(void)
2716 {
2717         clear_thread_flag(TIF_RESTORE_SIGMASK);
2718 }
2719 static inline bool test_restore_sigmask(void)
2720 {
2721         return test_thread_flag(TIF_RESTORE_SIGMASK);
2722 }
2723 static inline bool test_and_clear_restore_sigmask(void)
2724 {
2725         return test_and_clear_thread_flag(TIF_RESTORE_SIGMASK);
2726 }
2727 
2728 #else   /* TIF_RESTORE_SIGMASK */
2729 
2730 /* Higher-quality implementation, used if TIF_RESTORE_SIGMASK doesn't exist. */
2731 static inline void set_restore_sigmask(void)
2732 {
2733         current->restore_sigmask = true;
2734         WARN_ON(!test_thread_flag(TIF_SIGPENDING));
2735 }
2736 static inline void clear_restore_sigmask(void)
2737 {
2738         current->restore_sigmask = false;
2739 }
2740 static inline bool test_restore_sigmask(void)
2741 {
2742         return current->restore_sigmask;
2743 }
2744 static inline bool test_and_clear_restore_sigmask(void)
2745 {
2746         if (!current->restore_sigmask)
2747                 return false;
2748         current->restore_sigmask = false;
2749         return true;
2750 }
2751 #endif
2752 
2753 static inline void restore_saved_sigmask(void)
2754 {
2755         if (test_and_clear_restore_sigmask())
2756                 __set_current_blocked(&current->saved_sigmask);
2757 }
2758 
2759 static inline sigset_t *sigmask_to_save(void)
2760 {
2761         sigset_t *res = &current->blocked;
2762         if (unlikely(test_restore_sigmask()))
2763                 res = &current->saved_sigmask;
2764         return res;
2765 }
2766 
2767 static inline int kill_cad_pid(int sig, int priv)
2768 {
2769         return kill_pid(cad_pid, sig, priv);
2770 }
2771 
2772 /* These can be the second arg to send_sig_info/send_group_sig_info.  */
2773 #define SEND_SIG_NOINFO ((struct siginfo *) 0)
2774 #define SEND_SIG_PRIV   ((struct siginfo *) 1)
2775 #define SEND_SIG_FORCED ((struct siginfo *) 2)
2776 
2777 /*
2778  * True if we are on the alternate signal stack.
2779  */
2780 static inline int on_sig_stack(unsigned long sp)
2781 {
2782         /*
2783          * If the signal stack is SS_AUTODISARM then, by construction, we
2784          * can't be on the signal stack unless user code deliberately set
2785          * SS_AUTODISARM when we were already on it.
2786          *
2787          * This improves reliability: if user state gets corrupted such that
2788          * the stack pointer points very close to the end of the signal stack,
2789          * then this check will enable the signal to be handled anyway.
2790          */
2791         if (current->sas_ss_flags & SS_AUTODISARM)
2792                 return 0;
2793 
2794 #ifdef CONFIG_STACK_GROWSUP
2795         return sp >= current->sas_ss_sp &&
2796                 sp - current->sas_ss_sp < current->sas_ss_size;
2797 #else
2798         return sp > current->sas_ss_sp &&
2799                 sp - current->sas_ss_sp <= current->sas_ss_size;
2800 #endif
2801 }
2802 
2803 static inline int sas_ss_flags(unsigned long sp)
2804 {
2805         if (!current->sas_ss_size)
2806                 return SS_DISABLE;
2807 
2808         return on_sig_stack(sp) ? SS_ONSTACK : 0;
2809 }
2810 
2811 static inline void sas_ss_reset(struct task_struct *p)
2812 {
2813         p->sas_ss_sp = 0;
2814         p->sas_ss_size = 0;
2815         p->sas_ss_flags = SS_DISABLE;
2816 }
2817 
2818 static inline unsigned long sigsp(unsigned long sp, struct ksignal *ksig)
2819 {
2820         if (unlikely((ksig->ka.sa.sa_flags & SA_ONSTACK)) && ! sas_ss_flags(sp))
2821 #ifdef CONFIG_STACK_GROWSUP
2822                 return current->sas_ss_sp;
2823 #else
2824                 return current->sas_ss_sp + current->sas_ss_size;
2825 #endif
2826         return sp;
2827 }
2828 
2829 /*
2830  * Routines for handling mm_structs
2831  */
2832 extern struct mm_struct * mm_alloc(void);
2833 
2834 /* mmdrop drops the mm and the page tables */
2835 extern void __mmdrop(struct mm_struct *);
2836 static inline void mmdrop(struct mm_struct *mm)
2837 {
2838         if (unlikely(atomic_dec_and_test(&mm->mm_count)))
2839                 __mmdrop(mm);
2840 }
2841 
2842 static inline bool mmget_not_zero(struct mm_struct *mm)
2843 {
2844         return atomic_inc_not_zero(&mm->mm_users);
2845 }
2846 
2847 /* mmput gets rid of the mappings and all user-space */
2848 extern void mmput(struct mm_struct *);
2849 #ifdef CONFIG_MMU
2850 /* same as above but performs the slow path from the async context. Can
2851  * be called from the atomic context as well
2852  */
2853 extern void mmput_async(struct mm_struct *);
2854 #endif
2855 
2856 /* Grab a reference to a task's mm, if it is not already going away */
2857 extern struct mm_struct *get_task_mm(struct task_struct *task);
2858 /*
2859  * Grab a reference to a task's mm, if it is not already going away
2860  * and ptrace_may_access with the mode parameter passed to it
2861  * succeeds.
2862  */
2863 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode);
2864 /* Remove the current tasks stale references to the old mm_struct */
2865 extern void mm_release(struct task_struct *, struct mm_struct *);
2866 
2867 #ifdef CONFIG_HAVE_COPY_THREAD_TLS
2868 extern int copy_thread_tls(unsigned long, unsigned long, unsigned long,
2869                         struct task_struct *, unsigned long);
2870 #else
2871 extern int copy_thread(unsigned long, unsigned long, unsigned long,
2872                         struct task_struct *);
2873 
2874 /* Architectures that haven't opted into copy_thread_tls get the tls argument
2875  * via pt_regs, so ignore the tls argument passed via C. */
2876 static inline int copy_thread_tls(
2877                 unsigned long clone_flags, unsigned long sp, unsigned long arg,
2878                 struct task_struct *p, unsigned long tls)
2879 {
2880         return copy_thread(clone_flags, sp, arg, p);
2881 }
2882 #endif
2883 extern void flush_thread(void);
2884 
2885 #ifdef CONFIG_HAVE_EXIT_THREAD
2886 extern void exit_thread(struct task_struct *tsk);
2887 #else
2888 static inline void exit_thread(struct task_struct *tsk)
2889 {
2890 }
2891 #endif
2892 
2893 extern void exit_files(struct task_struct *);
2894 extern void __cleanup_sighand(struct sighand_struct *);
2895 
2896 extern void exit_itimers(struct signal_struct *);
2897 extern void flush_itimer_signals(void);
2898 
2899 extern void do_group_exit(int);
2900 
2901 extern int do_execve(struct filename *,
2902                      const char __user * const __user *,
2903                      const char __user * const __user *);
2904 extern int do_execveat(int, struct filename *,
2905                        const char __user * const __user *,
2906                        const char __user * const __user *,
2907                        int);
2908 extern long _do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *, unsigned long);
2909 extern long do_fork(unsigned long, unsigned long, unsigned long, int __user *, int __user *);
2910 struct task_struct *fork_idle(int);
2911 extern pid_t kernel_thread(int (*fn)(void *), void *arg, unsigned long flags);
2912 
2913 extern void __set_task_comm(struct task_struct *tsk, const char *from, bool exec);
2914 static inline void set_task_comm(struct task_struct *tsk, const char *from)
2915 {
2916         __set_task_comm(tsk, from, false);
2917 }
2918 extern char *get_task_comm(char *to, struct task_struct *tsk);
2919 
2920 #ifdef CONFIG_SMP
2921 void scheduler_ipi(void);
2922 extern unsigned long wait_task_inactive(struct task_struct *, long match_state);
2923 #else
2924 static inline void scheduler_ipi(void) { }
2925 static inline unsigned long wait_task_inactive(struct task_struct *p,
2926                                                long match_state)
2927 {
2928         return 1;
2929 }
2930 #endif
2931 
2932 #define tasklist_empty() \
2933         list_empty(&init_task.tasks)
2934 
2935 #define next_task(p) \
2936         list_entry_rcu((p)->tasks.next, struct task_struct, tasks)
2937 
2938 #define for_each_process(p) \
2939         for (p = &init_task ; (p = next_task(p)) != &init_task ; )
2940 
2941 extern bool current_is_single_threaded(void);
2942 
2943 /*
2944  * Careful: do_each_thread/while_each_thread is a double loop so
2945  *          'break' will not work as expected - use goto instead.
2946  */
2947 #define do_each_thread(g, t) \
2948         for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do
2949 
2950 #define while_each_thread(g, t) \
2951         while ((t = next_thread(t)) != g)
2952 
2953 #define __for_each_thread(signal, t)    \
2954         list_for_each_entry_rcu(t, &(signal)->thread_head, thread_node)
2955 
2956 #define for_each_thread(p, t)           \
2957         __for_each_thread((p)->signal, t)
2958 
2959 /* Careful: this is a double loop, 'break' won't work as expected. */
2960 #define for_each_process_thread(p, t)   \
2961         for_each_process(p) for_each_thread(p, t)
2962 
2963 static inline int get_nr_threads(struct task_struct *tsk)
2964 {
2965         return tsk->signal->nr_threads;
2966 }
2967 
2968 static inline bool thread_group_leader(struct task_struct *p)
2969 {
2970         return p->exit_signal >= 0;
2971 }
2972 
2973 /* Do to the insanities of de_thread it is possible for a process
2974  * to have the pid of the thread group leader without actually being
2975  * the thread group leader.  For iteration through the pids in proc
2976  * all we care about is that we have a task with the appropriate
2977  * pid, we don't actually care if we have the right task.
2978  */
2979 static inline bool has_group_leader_pid(struct task_struct *p)
2980 {
2981         return task_pid(p) == p->signal->leader_pid;
2982 }
2983 
2984 static inline
2985 bool same_thread_group(struct task_struct *p1, struct task_struct *p2)
2986 {
2987         return p1->signal == p2->signal;
2988 }
2989 
2990 static inline struct task_struct *next_thread(const struct task_struct *p)
2991 {
2992         return list_entry_rcu(p->thread_group.next,
2993                               struct task_struct, thread_group);
2994 }
2995 
2996 static inline int thread_group_empty(struct task_struct *p)
2997 {
2998         return list_empty(&p->thread_group);
2999 }
3000 
3001 #define delay_group_leader(p) \
3002                 (thread_group_leader(p) && !thread_group_empty(p))
3003 
3004 /*
3005  * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring
3006  * subscriptions and synchronises with wait4().  Also used in procfs.  Also
3007  * pins the final release of task.io_context.  Also protects ->cpuset and
3008  * ->cgroup.subsys[]. And ->vfork_done.
3009  *
3010  * Nests both inside and outside of read_lock(&tasklist_lock).
3011  * It must not be nested with write_lock_irq(&tasklist_lock),
3012  * neither inside nor outside.
3013  */
3014 static inline void task_lock(struct task_struct *p)
3015 {
3016         spin_lock(&p->alloc_lock);
3017 }
3018 
3019 static inline void task_unlock(struct task_struct *p)
3020 {
3021         spin_unlock(&p->alloc_lock);
3022 }
3023 
3024 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk,
3025                                                         unsigned long *flags);
3026 
3027 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk,
3028                                                        unsigned long *flags)
3029 {
3030         struct sighand_struct *ret;
3031 
3032         ret = __lock_task_sighand(tsk, flags);
3033         (void)__cond_lock(&tsk->sighand->siglock, ret);
3034         return ret;
3035 }
3036 
3037 static inline void unlock_task_sighand(struct task_struct *tsk,
3038                                                 unsigned long *flags)
3039 {
3040         spin_unlock_irqrestore(&tsk->sighand->siglock, *flags);
3041 }
3042 
3043 /**
3044  * threadgroup_change_begin - mark the beginning of changes to a threadgroup
3045  * @tsk: task causing the changes
3046  *
3047  * All operations which modify a threadgroup - a new thread joining the
3048  * group, death of a member thread (the assertion of PF_EXITING) and
3049  * exec(2) dethreading the process and replacing the leader - are wrapped
3050  * by threadgroup_change_{begin|end}().  This is to provide a place which
3051  * subsystems needing threadgroup stability can hook into for
3052  * synchronization.
3053  */
3054 static inline void threadgroup_change_begin(struct task_struct *tsk)
3055 {
3056         might_sleep();
3057         cgroup_threadgroup_change_begin(tsk);
3058 }
3059 
3060 /**
3061  * threadgroup_change_end - mark the end of changes to a threadgroup
3062  * @tsk: task causing the changes
3063  *
3064  * See threadgroup_change_begin().
3065  */
3066 static inline void threadgroup_change_end(struct task_struct *tsk)
3067 {
3068         cgroup_threadgroup_change_end(tsk);
3069 }
3070 
3071 #ifndef __HAVE_THREAD_FUNCTIONS
3072 
3073 #define task_thread_info(task)  ((struct thread_info *)(task)->stack)
3074 #define task_stack_page(task)   ((task)->stack)
3075 
3076 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org)
3077 {
3078         *task_thread_info(p) = *task_thread_info(org);
3079         task_thread_info(p)->task = p;
3080 }
3081 
3082 /*
3083  * Return the address of the last usable long on the stack.
3084  *
3085  * When the stack grows down, this is just above the thread
3086  * info struct. Going any lower will corrupt the threadinfo.
3087  *
3088  * When the stack grows up, this is the highest address.
3089  * Beyond that position, we corrupt data on the next page.
3090  */
3091 static inline unsigned long *end_of_stack(struct task_struct *p)
3092 {
3093 #ifdef CONFIG_STACK_GROWSUP
3094         return (unsigned long *)((unsigned long)task_thread_info(p) + THREAD_SIZE) - 1;
3095 #else
3096         return (unsigned long *)(task_thread_info(p) + 1);
3097 #endif
3098 }
3099 
3100 #endif
3101 #define task_stack_end_corrupted(task) \
3102                 (*(end_of_stack(task)) != STACK_END_MAGIC)
3103 
3104 static inline int object_is_on_stack(void *obj)
3105 {
3106         void *stack = task_stack_page(current);
3107 
3108         return (obj >= stack) && (obj < (stack + THREAD_SIZE));
3109 }
3110 
3111 extern void thread_stack_cache_init(void);
3112 
3113 #ifdef CONFIG_DEBUG_STACK_USAGE
3114 static inline unsigned long stack_not_used(struct task_struct *p)
3115 {
3116         unsigned long *n = end_of_stack(p);
3117 
3118         do {    /* Skip over canary */
3119 # ifdef CONFIG_STACK_GROWSUP
3120                 n--;
3121 # else
3122                 n++;
3123 # endif
3124         } while (!*n);
3125 
3126 # ifdef CONFIG_STACK_GROWSUP
3127         return (unsigned long)end_of_stack(p) - (unsigned long)n;
3128 # else
3129         return (unsigned long)n - (unsigned long)end_of_stack(p);
3130 # endif
3131 }
3132 #endif
3133 extern void set_task_stack_end_magic(struct task_struct *tsk);
3134 
3135 /* set thread flags in other task's structures
3136  * - see asm/thread_info.h for TIF_xxxx flags available
3137  */
3138 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag)
3139 {
3140         set_ti_thread_flag(task_thread_info(tsk), flag);
3141 }
3142 
3143 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3144 {
3145         clear_ti_thread_flag(task_thread_info(tsk), flag);
3146 }
3147 
3148 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag)
3149 {
3150         return test_and_set_ti_thread_flag(task_thread_info(tsk), flag);
3151 }
3152 
3153 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag)
3154 {
3155         return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag);
3156 }
3157 
3158 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag)
3159 {
3160         return test_ti_thread_flag(task_thread_info(tsk), flag);
3161 }
3162 
3163 static inline void set_tsk_need_resched(struct task_struct *tsk)
3164 {
3165         set_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3166 }
3167 
3168 static inline void clear_tsk_need_resched(struct task_struct *tsk)
3169 {
3170         clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED);
3171 }
3172 
3173 static inline int test_tsk_need_resched(struct task_struct *tsk)
3174 {
3175         return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED));
3176 }
3177 
3178 static inline int restart_syscall(void)
3179 {
3180         set_tsk_thread_flag(current, TIF_SIGPENDING);
3181         return -ERESTARTNOINTR;
3182 }
3183 
3184 static inline int signal_pending(struct task_struct *p)
3185 {
3186         return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING));
3187 }
3188 
3189 static inline int __fatal_signal_pending(struct task_struct *p)
3190 {
3191         return unlikely(sigismember(&p->pending.signal, SIGKILL));
3192 }
3193 
3194 static inline int fatal_signal_pending(struct task_struct *p)
3195 {
3196         return signal_pending(p) && __fatal_signal_pending(p);
3197 }
3198 
3199 static inline int signal_pending_state(long state, struct task_struct *p)
3200 {
3201         if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL)))
3202                 return 0;
3203         if (!signal_pending(p))
3204                 return 0;
3205 
3206         return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p);
3207 }
3208 
3209 /*
3210  * cond_resched() and cond_resched_lock(): latency reduction via
3211  * explicit rescheduling in places that are safe. The return
3212  * value indicates whether a reschedule was done in fact.
3213  * cond_resched_lock() will drop the spinlock before scheduling,
3214  * cond_resched_softirq() will enable bhs before scheduling.
3215  */
3216 extern int _cond_resched(void);
3217 
3218 #define cond_resched() ({                       \
3219         ___might_sleep(__FILE__, __LINE__, 0);  \
3220         _cond_resched();                        \
3221 })
3222 
3223 extern int __cond_resched_lock(spinlock_t *lock);
3224 
3225 #define cond_resched_lock(lock) ({                              \
3226         ___might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET);\
3227         __cond_resched_lock(lock);                              \
3228 })
3229 
3230 extern int __cond_resched_softirq(void);
3231 
3232 #define cond_resched_softirq() ({                                       \
3233         ___might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET);     \
3234         __cond_resched_softirq();                                       \
3235 })
3236 
3237 static inline void cond_resched_rcu(void)
3238 {
3239 #if defined(CONFIG_DEBUG_ATOMIC_SLEEP) || !defined(CONFIG_PREEMPT_RCU)
3240         rcu_read_unlock();
3241         cond_resched();
3242         rcu_read_lock();
3243 #endif
3244 }
3245 
3246 /*
3247  * Does a critical section need to be broken due to another
3248  * task waiting?: (technically does not depend on CONFIG_PREEMPT,
3249  * but a general need for low latency)
3250  */
3251 static inline int spin_needbreak(spinlock_t *lock)
3252 {
3253 #ifdef CONFIG_PREEMPT
3254         return spin_is_contended(lock);
3255 #else
3256         return 0;
3257 #endif
3258 }
3259 
3260 /*
3261  * Idle thread specific functions to determine the need_resched
3262  * polling state.
3263  */
3264 #ifdef TIF_POLLING_NRFLAG
3265 static inline int tsk_is_polling(struct task_struct *p)
3266 {
3267         return test_tsk_thread_flag(p, TIF_POLLING_NRFLAG);
3268 }
3269 
3270 static inline void __current_set_polling(void)
3271 {
3272         set_thread_flag(TIF_POLLING_NRFLAG);
3273 }
3274 
3275 static inline bool __must_check current_set_polling_and_test(void)
3276 {
3277         __current_set_polling();
3278 
3279         /*
3280          * Polling state must be visible before we test NEED_RESCHED,
3281          * paired by resched_curr()
3282          */
3283         smp_mb__after_atomic();
3284 
3285         return unlikely(tif_need_resched());
3286 }
3287 
3288 static inline void __current_clr_polling(void)
3289 {
3290         clear_thread_flag(TIF_POLLING_NRFLAG);
3291 }
3292 
3293 static inline bool __must_check current_clr_polling_and_test(void)
3294 {
3295         __current_clr_polling();
3296 
3297         /*
3298          * Polling state must be visible before we test NEED_RESCHED,
3299          * paired by resched_curr()
3300          */
3301         smp_mb__after_atomic();
3302 
3303         return unlikely(tif_need_resched());
3304 }
3305 
3306 #else
3307 static inline int tsk_is_polling(struct task_struct *p) { return 0; }
3308 static inline void __current_set_polling(void) { }
3309 static inline void __current_clr_polling(void) { }
3310 
3311 static inline bool __must_check current_set_polling_and_test(void)
3312 {
3313         return unlikely(tif_need_resched());
3314 }
3315 static inline bool __must_check current_clr_polling_and_test(void)
3316 {
3317         return unlikely(tif_need_resched());
3318 }
3319 #endif
3320 
3321 static inline void current_clr_polling(void)
3322 {
3323         __current_clr_polling();
3324 
3325         /*
3326          * Ensure we check TIF_NEED_RESCHED after we clear the polling bit.
3327          * Once the bit is cleared, we'll get IPIs with every new
3328          * TIF_NEED_RESCHED and the IPI handler, scheduler_ipi(), will also
3329          * fold.
3330          */
3331         smp_mb(); /* paired with resched_curr() */
3332 
3333         preempt_fold_need_resched();
3334 }
3335 
3336 static __always_inline bool need_resched(void)
3337 {
3338         return unlikely(tif_need_resched());
3339 }
3340 
3341 /*
3342  * Thread group CPU time accounting.
3343  */
3344 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times);
3345 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times);
3346 
3347 /*
3348  * Reevaluate whether the task has signals pending delivery.
3349  * Wake the task if so.
3350  * This is required every time the blocked sigset_t changes.
3351  * callers must hold sighand->siglock.
3352  */
3353 extern void recalc_sigpending_and_wake(struct task_struct *t);
3354 extern void recalc_sigpending(void);
3355 
3356 extern void signal_wake_up_state(struct task_struct *t, unsigned int state);
3357 
3358 static inline void signal_wake_up(struct task_struct *t, bool resume)
3359 {
3360         signal_wake_up_state(t, resume ? TASK_WAKEKILL : 0);
3361 }
3362 static inline void ptrace_signal_wake_up(struct task_struct *t, bool resume)
3363 {
3364         signal_wake_up_state(t, resume ? __TASK_TRACED : 0);
3365 }
3366 
3367 /*
3368  * Wrappers for p->thread_info->cpu access. No-op on UP.
3369  */
3370 #ifdef CONFIG_SMP
3371 
3372 static inline unsigned int task_cpu(const struct task_struct *p)
3373 {
3374         return task_thread_info(p)->cpu;
3375 }
3376 
3377 static inline int task_node(const struct task_struct *p)
3378 {
3379         return cpu_to_node(task_cpu(p));
3380 }
3381 
3382 extern void set_task_cpu(struct task_struct *p, unsigned int cpu);
3383 
3384 #else
3385 
3386 static inline unsigned int task_cpu(const struct task_struct *p)
3387 {
3388         return 0;
3389 }
3390 
3391 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu)
3392 {
3393 }
3394 
3395 #endif /* CONFIG_SMP */
3396 
3397 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask);
3398 extern long sched_getaffinity(pid_t pid, struct cpumask *mask);
3399 
3400 #ifdef CONFIG_CGROUP_SCHED
3401 extern struct task_group root_task_group;
3402 #endif /* CONFIG_CGROUP_SCHED */
3403 
3404 extern int task_can_switch_user(struct user_struct *up,
3405                                         struct task_struct *tsk);
3406 
3407 #ifdef CONFIG_TASK_XACCT
3408 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3409 {
3410         tsk->ioac.rchar += amt;
3411 }
3412 
3413 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3414 {
3415         tsk->ioac.wchar += amt;
3416 }
3417 
3418 static inline void inc_syscr(struct task_struct *tsk)
3419 {
3420         tsk->ioac.syscr++;
3421 }
3422 
3423 static inline void inc_syscw(struct task_struct *tsk)
3424 {
3425         tsk->ioac.syscw++;
3426 }
3427 #else
3428 static inline void add_rchar(struct task_struct *tsk, ssize_t amt)
3429 {
3430 }
3431 
3432 static inline void add_wchar(struct task_struct *tsk, ssize_t amt)
3433 {
3434 }
3435 
3436 static inline void inc_syscr(struct task_struct *tsk)
3437 {
3438 }
3439 
3440 static inline void inc_syscw(struct task_struct *tsk)
3441 {
3442 }
3443 #endif
3444 
3445 #ifndef TASK_SIZE_OF
3446 #define TASK_SIZE_OF(tsk)       TASK_SIZE
3447 #endif
3448 
3449 #ifdef CONFIG_MEMCG
3450 extern void mm_update_next_owner(struct mm_struct *mm);
3451 #else
3452 static inline void mm_update_next_owner(struct mm_struct *mm)
3453 {
3454 }
3455 #endif /* CONFIG_MEMCG */
3456 
3457 static inline unsigned long task_rlimit(const struct task_struct *tsk,
3458                 unsigned int limit)
3459 {
3460         return READ_ONCE(tsk->signal->rlim[limit].rlim_cur);
3461 }
3462 
3463 static inline unsigned long task_rlimit_max(const struct task_struct *tsk,
3464                 unsigned int limit)
3465 {
3466         return READ_ONCE(tsk->signal->rlim[limit].rlim_max);
3467 }
3468 
3469 static inline unsigned long rlimit(unsigned int limit)
3470 {
3471         return task_rlimit(current, limit);
3472 }
3473 
3474 static inline unsigned long rlimit_max(unsigned int limit)
3475 {
3476         return task_rlimit_max(current, limit);
3477 }
3478 
3479 #ifdef CONFIG_CPU_FREQ
3480 struct update_util_data {
3481         void (*func)(struct update_util_data *data,
3482                      u64 time, unsigned long util, unsigned long max);
3483 };
3484 
3485 void cpufreq_add_update_util_hook(int cpu, struct update_util_data *data,
3486                         void (*func)(struct update_util_data *data, u64 time,
3487                                      unsigned long util, unsigned long max));
3488 void cpufreq_remove_update_util_hook(int cpu);
3489 #endif /* CONFIG_CPU_FREQ */
3490 
3491 #endif
3492 

~ [ source navigation ] ~ [ diff markup ] ~ [ identifier search ] ~

kernel.org | git.kernel.org | LWN.net | Project Home | Wiki (Japanese) | Wiki (English) | SVN repository | Mail admin

Linux® is a registered trademark of Linus Torvalds in the United States and other countries.
TOMOYO® is a registered trademark of NTT DATA CORPORATION.

osdn.jp