Version:  2.0.40 2.2.26 2.4.37 3.13 3.14 3.15 3.16 3.17 3.18 3.19 4.0 4.1 4.2 4.3 4.4 4.5 4.6 4.7 4.8 4.9 4.10

Linux/include/linux/sched.h

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

This page was automatically generated by LXR 0.3.1 (source).  •  Linux is a registered trademark of Linus Torvalds  •  Contact us