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

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