1 #ifndef _LINUX_SCHED_H 2 #define _LINUX_SCHED_H 3 4 /* 5 * cloning flags: 6 */ 7 #define CSIGNAL 0x000000ff /* signal mask to be sent at exit */ 8 #define CLONE_VM 0x00000100 /* set if VM shared between processes */ 9 #define CLONE_FS 0x00000200 /* set if fs info shared between processes */ 10 #define CLONE_FILES 0x00000400 /* set if open files shared between processes */ 11 #define CLONE_SIGHAND 0x00000800 /* set if signal handlers and blocked signals shared */ 12 #define CLONE_PTRACE 0x00002000 /* set if we want to let tracing continue on the child too */ 13 #define CLONE_VFORK 0x00004000 /* set if the parent wants the child to wake it up on mm_release */ 14 #define CLONE_PARENT 0x00008000 /* set if we want to have the same parent as the cloner */ 15 #define CLONE_THREAD 0x00010000 /* Same thread group? */ 16 #define CLONE_NEWNS 0x00020000 /* New namespace group? */ 17 #define CLONE_SYSVSEM 0x00040000 /* share system V SEM_UNDO semantics */ 18 #define CLONE_SETTLS 0x00080000 /* create a new TLS for the child */ 19 #define CLONE_PARENT_SETTID 0x00100000 /* set the TID in the parent */ 20 #define CLONE_CHILD_CLEARTID 0x00200000 /* clear the TID in the child */ 21 #define CLONE_DETACHED 0x00400000 /* Unused, ignored */ 22 #define CLONE_UNTRACED 0x00800000 /* set if the tracing process can't force CLONE_PTRACE on this clone */ 23 #define CLONE_CHILD_SETTID 0x01000000 /* set the TID in the child */ 24 /* 0x02000000 was previously the unused CLONE_STOPPED (Start in stopped state) 25 and is now available for re-use. */ 26 #define CLONE_NEWUTS 0x04000000 /* New utsname group? */ 27 #define CLONE_NEWIPC 0x08000000 /* New ipcs */ 28 #define CLONE_NEWUSER 0x10000000 /* New user namespace */ 29 #define CLONE_NEWPID 0x20000000 /* New pid namespace */ 30 #define CLONE_NEWNET 0x40000000 /* New network namespace */ 31 #define CLONE_IO 0x80000000 /* Clone io context */ 32 33 /* 34 * Scheduling policies 35 */ 36 #define SCHED_NORMAL 0 37 #define SCHED_FIFO 1 38 #define SCHED_RR 2 39 #define SCHED_BATCH 3 40 /* SCHED_ISO: reserved but not implemented yet */ 41 #define SCHED_IDLE 5 42 /* Can be ORed in to make sure the process is reverted back to SCHED_NORMAL on fork */ 43 #define SCHED_RESET_ON_FORK 0x40000000 44 45 #ifdef __KERNEL__ 46 47 struct sched_param { 48 int sched_priority; 49 }; 50 51 #include <asm/param.h> /* for HZ */ 52 53 #include <linux/capability.h> 54 #include <linux/threads.h> 55 #include <linux/kernel.h> 56 #include <linux/types.h> 57 #include <linux/timex.h> 58 #include <linux/jiffies.h> 59 #include <linux/rbtree.h> 60 #include <linux/thread_info.h> 61 #include <linux/cpumask.h> 62 #include <linux/errno.h> 63 #include <linux/nodemask.h> 64 #include <linux/mm_types.h> 65 66 #include <asm/page.h> 67 #include <asm/ptrace.h> 68 #include <asm/cputime.h> 69 70 #include <linux/smp.h> 71 #include <linux/sem.h> 72 #include <linux/signal.h> 73 #include <linux/compiler.h> 74 #include <linux/completion.h> 75 #include <linux/pid.h> 76 #include <linux/percpu.h> 77 #include <linux/topology.h> 78 #include <linux/proportions.h> 79 #include <linux/seccomp.h> 80 #include <linux/rcupdate.h> 81 #include <linux/rculist.h> 82 #include <linux/rtmutex.h> 83 84 #include <linux/time.h> 85 #include <linux/param.h> 86 #include <linux/resource.h> 87 #include <linux/timer.h> 88 #include <linux/hrtimer.h> 89 #include <linux/task_io_accounting.h> 90 #include <linux/latencytop.h> 91 #include <linux/cred.h> 92 #include <linux/llist.h> 93 94 #include <asm/processor.h> 95 96 struct exec_domain; 97 struct futex_pi_state; 98 struct robust_list_head; 99 struct bio_list; 100 struct fs_struct; 101 struct perf_event_context; 102 struct blk_plug; 103 104 /* 105 * List of flags we want to share for kernel threads, 106 * if only because they are not used by them anyway. 107 */ 108 #define CLONE_KERNEL (CLONE_FS | CLONE_FILES | CLONE_SIGHAND) 109 110 /* 111 * These are the constant used to fake the fixed-point load-average 112 * counting. Some notes: 113 * - 11 bit fractions expand to 22 bits by the multiplies: this gives 114 * a load-average precision of 10 bits integer + 11 bits fractional 115 * - if you want to count load-averages more often, you need more 116 * precision, or rounding will get you. With 2-second counting freq, 117 * the EXP_n values would be 1981, 2034 and 2043 if still using only 118 * 11 bit fractions. 119 */ 120 extern unsigned long avenrun[]; /* Load averages */ 121 extern void get_avenrun(unsigned long *loads, unsigned long offset, int shift); 122 123 #define FSHIFT 11 /* nr of bits of precision */ 124 #define FIXED_1 (1<<FSHIFT) /* 1.0 as fixed-point */ 125 #define LOAD_FREQ (5*HZ+1) /* 5 sec intervals */ 126 #define EXP_1 1884 /* 1/exp(5sec/1min) as fixed-point */ 127 #define EXP_5 2014 /* 1/exp(5sec/5min) */ 128 #define EXP_15 2037 /* 1/exp(5sec/15min) */ 129 130 #define CALC_LOAD(load,exp,n) \ 131 load *= exp; \ 132 load += n*(FIXED_1-exp); \ 133 load >>= FSHIFT; 134 135 extern unsigned long total_forks; 136 extern int nr_threads; 137 DECLARE_PER_CPU(unsigned long, process_counts); 138 extern int nr_processes(void); 139 extern unsigned long nr_running(void); 140 extern unsigned long nr_uninterruptible(void); 141 extern unsigned long nr_iowait(void); 142 extern unsigned long nr_iowait_cpu(int cpu); 143 extern unsigned long this_cpu_load(void); 144 145 146 extern void calc_global_load(unsigned long ticks); 147 148 extern unsigned long get_parent_ip(unsigned long addr); 149 150 struct seq_file; 151 struct cfs_rq; 152 struct task_group; 153 #ifdef CONFIG_SCHED_DEBUG 154 extern void proc_sched_show_task(struct task_struct *p, struct seq_file *m); 155 extern void proc_sched_set_task(struct task_struct *p); 156 extern void 157 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq); 158 #else 159 static inline void 160 proc_sched_show_task(struct task_struct *p, struct seq_file *m) 161 { 162 } 163 static inline void proc_sched_set_task(struct task_struct *p) 164 { 165 } 166 static inline void 167 print_cfs_rq(struct seq_file *m, int cpu, struct cfs_rq *cfs_rq) 168 { 169 } 170 #endif 171 172 /* 173 * Task state bitmask. NOTE! These bits are also 174 * encoded in fs/proc/array.c: get_task_state(). 175 * 176 * We have two separate sets of flags: task->state 177 * is about runnability, while task->exit_state are 178 * about the task exiting. Confusing, but this way 179 * modifying one set can't modify the other one by 180 * mistake. 181 */ 182 #define TASK_RUNNING 0 183 #define TASK_INTERRUPTIBLE 1 184 #define TASK_UNINTERRUPTIBLE 2 185 #define __TASK_STOPPED 4 186 #define __TASK_TRACED 8 187 /* in tsk->exit_state */ 188 #define EXIT_ZOMBIE 16 189 #define EXIT_DEAD 32 190 /* in tsk->state again */ 191 #define TASK_DEAD 64 192 #define TASK_WAKEKILL 128 193 #define TASK_WAKING 256 194 #define TASK_STATE_MAX 512 195 196 #define TASK_STATE_TO_CHAR_STR "RSDTtZXxKW" 197 198 extern char ___assert_task_state[1 - 2*!!( 199 sizeof(TASK_STATE_TO_CHAR_STR)-1 != ilog2(TASK_STATE_MAX)+1)]; 200 201 /* Convenience macros for the sake of set_task_state */ 202 #define TASK_KILLABLE (TASK_WAKEKILL | TASK_UNINTERRUPTIBLE) 203 #define TASK_STOPPED (TASK_WAKEKILL | __TASK_STOPPED) 204 #define TASK_TRACED (TASK_WAKEKILL | __TASK_TRACED) 205 206 /* Convenience macros for the sake of wake_up */ 207 #define TASK_NORMAL (TASK_INTERRUPTIBLE | TASK_UNINTERRUPTIBLE) 208 #define TASK_ALL (TASK_NORMAL | __TASK_STOPPED | __TASK_TRACED) 209 210 /* get_task_state() */ 211 #define TASK_REPORT (TASK_RUNNING | TASK_INTERRUPTIBLE | \ 212 TASK_UNINTERRUPTIBLE | __TASK_STOPPED | \ 213 __TASK_TRACED) 214 215 #define task_is_traced(task) ((task->state & __TASK_TRACED) != 0) 216 #define task_is_stopped(task) ((task->state & __TASK_STOPPED) != 0) 217 #define task_is_dead(task) ((task)->exit_state != 0) 218 #define task_is_stopped_or_traced(task) \ 219 ((task->state & (__TASK_STOPPED | __TASK_TRACED)) != 0) 220 #define task_contributes_to_load(task) \ 221 ((task->state & TASK_UNINTERRUPTIBLE) != 0 && \ 222 (task->flags & PF_FROZEN) == 0) 223 224 #define __set_task_state(tsk, state_value) \ 225 do { (tsk)->state = (state_value); } while (0) 226 #define set_task_state(tsk, state_value) \ 227 set_mb((tsk)->state, (state_value)) 228 229 /* 230 * set_current_state() includes a barrier so that the write of current->state 231 * is correctly serialised wrt the caller's subsequent test of whether to 232 * actually sleep: 233 * 234 * set_current_state(TASK_UNINTERRUPTIBLE); 235 * if (do_i_need_to_sleep()) 236 * schedule(); 237 * 238 * If the caller does not need such serialisation then use __set_current_state() 239 */ 240 #define __set_current_state(state_value) \ 241 do { current->state = (state_value); } while (0) 242 #define set_current_state(state_value) \ 243 set_mb(current->state, (state_value)) 244 245 /* Task command name length */ 246 #define TASK_COMM_LEN 16 247 248 #include <linux/spinlock.h> 249 250 /* 251 * This serializes "schedule()" and also protects 252 * the run-queue from deletions/modifications (but 253 * _adding_ to the beginning of the run-queue has 254 * a separate lock). 255 */ 256 extern rwlock_t tasklist_lock; 257 extern spinlock_t mmlist_lock; 258 259 struct task_struct; 260 261 #ifdef CONFIG_PROVE_RCU 262 extern int lockdep_tasklist_lock_is_held(void); 263 #endif /* #ifdef CONFIG_PROVE_RCU */ 264 265 extern void sched_init(void); 266 extern void sched_init_smp(void); 267 extern asmlinkage void schedule_tail(struct task_struct *prev); 268 extern void init_idle(struct task_struct *idle, int cpu); 269 extern void init_idle_bootup_task(struct task_struct *idle); 270 271 extern int runqueue_is_locked(int cpu); 272 273 #if defined(CONFIG_SMP) && defined(CONFIG_NO_HZ) 274 extern void select_nohz_load_balancer(int stop_tick); 275 extern void set_cpu_sd_state_idle(void); 276 extern int get_nohz_timer_target(void); 277 #else 278 static inline void select_nohz_load_balancer(int stop_tick) { } 279 static inline void set_cpu_sd_state_idle(void) { } 280 #endif 281 282 /* 283 * Only dump TASK_* tasks. (0 for all tasks) 284 */ 285 extern void show_state_filter(unsigned long state_filter); 286 287 static inline void show_state(void) 288 { 289 show_state_filter(0); 290 } 291 292 extern void show_regs(struct pt_regs *); 293 294 /* 295 * TASK is a pointer to the task whose backtrace we want to see (or NULL for current 296 * task), SP is the stack pointer of the first frame that should be shown in the back 297 * trace (or NULL if the entire call-chain of the task should be shown). 298 */ 299 extern void show_stack(struct task_struct *task, unsigned long *sp); 300 301 void io_schedule(void); 302 long io_schedule_timeout(long timeout); 303 304 extern void cpu_init (void); 305 extern void trap_init(void); 306 extern void update_process_times(int user); 307 extern void scheduler_tick(void); 308 309 extern void sched_show_task(struct task_struct *p); 310 311 #ifdef CONFIG_LOCKUP_DETECTOR 312 extern void touch_softlockup_watchdog(void); 313 extern void touch_softlockup_watchdog_sync(void); 314 extern void touch_all_softlockup_watchdogs(void); 315 extern int proc_dowatchdog_thresh(struct ctl_table *table, int write, 316 void __user *buffer, 317 size_t *lenp, loff_t *ppos); 318 extern unsigned int softlockup_panic; 319 void lockup_detector_init(void); 320 #else 321 static inline void touch_softlockup_watchdog(void) 322 { 323 } 324 static inline void touch_softlockup_watchdog_sync(void) 325 { 326 } 327 static inline void touch_all_softlockup_watchdogs(void) 328 { 329 } 330 static inline void lockup_detector_init(void) 331 { 332 } 333 #endif 334 335 #ifdef CONFIG_DETECT_HUNG_TASK 336 extern unsigned int sysctl_hung_task_panic; 337 extern unsigned long sysctl_hung_task_check_count; 338 extern unsigned long sysctl_hung_task_timeout_secs; 339 extern unsigned long sysctl_hung_task_warnings; 340 extern int proc_dohung_task_timeout_secs(struct ctl_table *table, int write, 341 void __user *buffer, 342 size_t *lenp, loff_t *ppos); 343 #else 344 /* Avoid need for ifdefs elsewhere in the code */ 345 enum { sysctl_hung_task_timeout_secs = 0 }; 346 #endif 347 348 /* Attach to any functions which should be ignored in wchan output. */ 349 #define __sched __attribute__((__section__(".sched.text"))) 350 351 /* Linker adds these: start and end of __sched functions */ 352 extern char __sched_text_start[], __sched_text_end[]; 353 354 /* Is this address in the __sched functions? */ 355 extern int in_sched_functions(unsigned long addr); 356 357 #define MAX_SCHEDULE_TIMEOUT LONG_MAX 358 extern signed long schedule_timeout(signed long timeout); 359 extern signed long schedule_timeout_interruptible(signed long timeout); 360 extern signed long schedule_timeout_killable(signed long timeout); 361 extern signed long schedule_timeout_uninterruptible(signed long timeout); 362 asmlinkage void schedule(void); 363 extern void schedule_preempt_disabled(void); 364 extern int mutex_spin_on_owner(struct mutex *lock, struct task_struct *owner); 365 366 struct nsproxy; 367 struct user_namespace; 368 369 /* 370 * Default maximum number of active map areas, this limits the number of vmas 371 * per mm struct. Users can overwrite this number by sysctl but there is a 372 * problem. 373 * 374 * When a program's coredump is generated as ELF format, a section is created 375 * per a vma. In ELF, the number of sections is represented in unsigned short. 376 * This means the number of sections should be smaller than 65535 at coredump. 377 * Because the kernel adds some informative sections to a image of program at 378 * generating coredump, we need some margin. The number of extra sections is 379 * 1-3 now and depends on arch. We use "5" as safe margin, here. 380 */ 381 #define MAPCOUNT_ELF_CORE_MARGIN (5) 382 #define DEFAULT_MAX_MAP_COUNT (USHRT_MAX - MAPCOUNT_ELF_CORE_MARGIN) 383 384 extern int sysctl_max_map_count; 385 386 #include <linux/aio.h> 387 388 #ifdef CONFIG_MMU 389 extern void arch_pick_mmap_layout(struct mm_struct *mm); 390 extern unsigned long 391 arch_get_unmapped_area(struct file *, unsigned long, unsigned long, 392 unsigned long, unsigned long); 393 extern unsigned long 394 arch_get_unmapped_area_topdown(struct file *filp, unsigned long addr, 395 unsigned long len, unsigned long pgoff, 396 unsigned long flags); 397 extern void arch_unmap_area(struct mm_struct *, unsigned long); 398 extern void arch_unmap_area_topdown(struct mm_struct *, unsigned long); 399 #else 400 static inline void arch_pick_mmap_layout(struct mm_struct *mm) {} 401 #endif 402 403 404 extern void set_dumpable(struct mm_struct *mm, int value); 405 extern int get_dumpable(struct mm_struct *mm); 406 407 /* mm flags */ 408 /* dumpable bits */ 409 #define MMF_DUMPABLE 0 /* core dump is permitted */ 410 #define MMF_DUMP_SECURELY 1 /* core file is readable only by root */ 411 412 #define MMF_DUMPABLE_BITS 2 413 #define MMF_DUMPABLE_MASK ((1 << MMF_DUMPABLE_BITS) - 1) 414 415 /* coredump filter bits */ 416 #define MMF_DUMP_ANON_PRIVATE 2 417 #define MMF_DUMP_ANON_SHARED 3 418 #define MMF_DUMP_MAPPED_PRIVATE 4 419 #define MMF_DUMP_MAPPED_SHARED 5 420 #define MMF_DUMP_ELF_HEADERS 6 421 #define MMF_DUMP_HUGETLB_PRIVATE 7 422 #define MMF_DUMP_HUGETLB_SHARED 8 423 424 #define MMF_DUMP_FILTER_SHIFT MMF_DUMPABLE_BITS 425 #define MMF_DUMP_FILTER_BITS 7 426 #define MMF_DUMP_FILTER_MASK \ 427 (((1 << MMF_DUMP_FILTER_BITS) - 1) << MMF_DUMP_FILTER_SHIFT) 428 #define MMF_DUMP_FILTER_DEFAULT \ 429 ((1 << MMF_DUMP_ANON_PRIVATE) | (1 << MMF_DUMP_ANON_SHARED) |\ 430 (1 << MMF_DUMP_HUGETLB_PRIVATE) | MMF_DUMP_MASK_DEFAULT_ELF) 431 432 #ifdef CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS 433 # define MMF_DUMP_MASK_DEFAULT_ELF (1 << MMF_DUMP_ELF_HEADERS) 434 #else 435 # define MMF_DUMP_MASK_DEFAULT_ELF 0 436 #endif 437 /* leave room for more dump flags */ 438 #define MMF_VM_MERGEABLE 16 /* KSM may merge identical pages */ 439 #define MMF_VM_HUGEPAGE 17 /* set when VM_HUGEPAGE is set on vma */ 440 441 #define MMF_INIT_MASK (MMF_DUMPABLE_MASK | MMF_DUMP_FILTER_MASK) 442 443 struct sighand_struct { 444 atomic_t count; 445 struct k_sigaction action[_NSIG]; 446 spinlock_t siglock; 447 wait_queue_head_t signalfd_wqh; 448 }; 449 450 struct pacct_struct { 451 int ac_flag; 452 long ac_exitcode; 453 unsigned long ac_mem; 454 cputime_t ac_utime, ac_stime; 455 unsigned long ac_minflt, ac_majflt; 456 }; 457 458 struct cpu_itimer { 459 cputime_t expires; 460 cputime_t incr; 461 u32 error; 462 u32 incr_error; 463 }; 464 465 /** 466 * struct task_cputime - collected CPU time counts 467 * @utime: time spent in user mode, in &cputime_t units 468 * @stime: time spent in kernel mode, in &cputime_t units 469 * @sum_exec_runtime: total time spent on the CPU, in nanoseconds 470 * 471 * This structure groups together three kinds of CPU time that are 472 * tracked for threads and thread groups. Most things considering 473 * CPU time want to group these counts together and treat all three 474 * of them in parallel. 475 */ 476 struct task_cputime { 477 cputime_t utime; 478 cputime_t stime; 479 unsigned long long sum_exec_runtime; 480 }; 481 /* Alternate field names when used to cache expirations. */ 482 #define prof_exp stime 483 #define virt_exp utime 484 #define sched_exp sum_exec_runtime 485 486 #define INIT_CPUTIME \ 487 (struct task_cputime) { \ 488 .utime = 0, \ 489 .stime = 0, \ 490 .sum_exec_runtime = 0, \ 491 } 492 493 /* 494 * Disable preemption until the scheduler is running. 495 * Reset by start_kernel()->sched_init()->init_idle(). 496 * 497 * We include PREEMPT_ACTIVE to avoid cond_resched() from working 498 * before the scheduler is active -- see should_resched(). 499 */ 500 #define INIT_PREEMPT_COUNT (1 + PREEMPT_ACTIVE) 501 502 /** 503 * struct thread_group_cputimer - thread group interval timer counts 504 * @cputime: thread group interval timers. 505 * @running: non-zero when there are timers running and 506 * @cputime receives updates. 507 * @lock: lock for fields in this struct. 508 * 509 * This structure contains the version of task_cputime, above, that is 510 * used for thread group CPU timer calculations. 511 */ 512 struct thread_group_cputimer { 513 struct task_cputime cputime; 514 int running; 515 raw_spinlock_t lock; 516 }; 517 518 #include <linux/rwsem.h> 519 struct autogroup; 520 521 /* 522 * NOTE! "signal_struct" does not have its own 523 * locking, because a shared signal_struct always 524 * implies a shared sighand_struct, so locking 525 * sighand_struct is always a proper superset of 526 * the locking of signal_struct. 527 */ 528 struct signal_struct { 529 atomic_t sigcnt; 530 atomic_t live; 531 int nr_threads; 532 533 wait_queue_head_t wait_chldexit; /* for wait4() */ 534 535 /* current thread group signal load-balancing target: */ 536 struct task_struct *curr_target; 537 538 /* shared signal handling: */ 539 struct sigpending shared_pending; 540 541 /* thread group exit support */ 542 int group_exit_code; 543 /* overloaded: 544 * - notify group_exit_task when ->count is equal to notify_count 545 * - everyone except group_exit_task is stopped during signal delivery 546 * of fatal signals, group_exit_task processes the signal. 547 */ 548 int notify_count; 549 struct task_struct *group_exit_task; 550 551 /* thread group stop support, overloads group_exit_code too */ 552 int group_stop_count; 553 unsigned int flags; /* see SIGNAL_* flags below */ 554 555 /* 556 * PR_SET_CHILD_SUBREAPER marks a process, like a service 557 * manager, to re-parent orphan (double-forking) child processes 558 * to this process instead of 'init'. The service manager is 559 * able to receive SIGCHLD signals and is able to investigate 560 * the process until it calls wait(). All children of this 561 * process will inherit a flag if they should look for a 562 * child_subreaper process at exit. 563 */ 564 unsigned int is_child_subreaper:1; 565 unsigned int has_child_subreaper:1; 566 567 /* POSIX.1b Interval Timers */ 568 struct list_head posix_timers; 569 570 /* ITIMER_REAL timer for the process */ 571 struct hrtimer real_timer; 572 struct pid *leader_pid; 573 ktime_t it_real_incr; 574 575 /* 576 * ITIMER_PROF and ITIMER_VIRTUAL timers for the process, we use 577 * CPUCLOCK_PROF and CPUCLOCK_VIRT for indexing array as these 578 * values are defined to 0 and 1 respectively 579 */ 580 struct cpu_itimer it[2]; 581 582 /* 583 * Thread group totals for process CPU timers. 584 * See thread_group_cputimer(), et al, for details. 585 */ 586 struct thread_group_cputimer cputimer; 587 588 /* Earliest-expiration cache. */ 589 struct task_cputime cputime_expires; 590 591 struct list_head cpu_timers[3]; 592 593 struct pid *tty_old_pgrp; 594 595 /* boolean value for session group leader */ 596 int leader; 597 598 struct tty_struct *tty; /* NULL if no tty */ 599 600 #ifdef CONFIG_SCHED_AUTOGROUP 601 struct autogroup *autogroup; 602 #endif 603 /* 604 * Cumulative resource counters for dead threads in the group, 605 * and for reaped dead child processes forked by this group. 606 * Live threads maintain their own counters and add to these 607 * in __exit_signal, except for the group leader. 608 */ 609 cputime_t utime, stime, cutime, cstime; 610 cputime_t gtime; 611 cputime_t cgtime; 612 #ifndef CONFIG_VIRT_CPU_ACCOUNTING 613 cputime_t prev_utime, prev_stime; 614 #endif 615 unsigned long nvcsw, nivcsw, cnvcsw, cnivcsw; 616 unsigned long min_flt, maj_flt, cmin_flt, cmaj_flt; 617 unsigned long inblock, oublock, cinblock, coublock; 618 unsigned long maxrss, cmaxrss; 619 struct task_io_accounting ioac; 620 621 /* 622 * Cumulative ns of schedule CPU time fo dead threads in the 623 * group, not including a zombie group leader, (This only differs 624 * from jiffies_to_ns(utime + stime) if sched_clock uses something 625 * other than jiffies.) 626 */ 627 unsigned long long sum_sched_runtime; 628 629 /* 630 * We don't bother to synchronize most readers of this at all, 631 * because there is no reader checking a limit that actually needs 632 * to get both rlim_cur and rlim_max atomically, and either one 633 * alone is a single word that can safely be read normally. 634 * getrlimit/setrlimit use task_lock(current->group_leader) to 635 * protect this instead of the siglock, because they really 636 * have no need to disable irqs. 637 */ 638 struct rlimit rlim[RLIM_NLIMITS]; 639 640 #ifdef CONFIG_BSD_PROCESS_ACCT 641 struct pacct_struct pacct; /* per-process accounting information */ 642 #endif 643 #ifdef CONFIG_TASKSTATS 644 struct taskstats *stats; 645 #endif 646 #ifdef CONFIG_AUDIT 647 unsigned audit_tty; 648 struct tty_audit_buf *tty_audit_buf; 649 #endif 650 #ifdef CONFIG_CGROUPS 651 /* 652 * group_rwsem prevents new tasks from entering the threadgroup and 653 * member tasks from exiting,a more specifically, setting of 654 * PF_EXITING. fork and exit paths are protected with this rwsem 655 * using threadgroup_change_begin/end(). Users which require 656 * threadgroup to remain stable should use threadgroup_[un]lock() 657 * which also takes care of exec path. Currently, cgroup is the 658 * only user. 659 */ 660 struct rw_semaphore group_rwsem; 661 #endif 662 663 int oom_adj; /* OOM kill score adjustment (bit shift) */ 664 int oom_score_adj; /* OOM kill score adjustment */ 665 int oom_score_adj_min; /* OOM kill score adjustment minimum value. 666 * Only settable by CAP_SYS_RESOURCE. */ 667 668 struct mutex cred_guard_mutex; /* guard against foreign influences on 669 * credential calculations 670 * (notably. ptrace) */ 671 }; 672 673 /* Context switch must be unlocked if interrupts are to be enabled */ 674 #ifdef __ARCH_WANT_INTERRUPTS_ON_CTXSW 675 # define __ARCH_WANT_UNLOCKED_CTXSW 676 #endif 677 678 /* 679 * Bits in flags field of signal_struct. 680 */ 681 #define SIGNAL_STOP_STOPPED 0x00000001 /* job control stop in effect */ 682 #define SIGNAL_STOP_CONTINUED 0x00000002 /* SIGCONT since WCONTINUED reap */ 683 #define SIGNAL_GROUP_EXIT 0x00000004 /* group exit in progress */ 684 /* 685 * Pending notifications to parent. 686 */ 687 #define SIGNAL_CLD_STOPPED 0x00000010 688 #define SIGNAL_CLD_CONTINUED 0x00000020 689 #define SIGNAL_CLD_MASK (SIGNAL_CLD_STOPPED|SIGNAL_CLD_CONTINUED) 690 691 #define SIGNAL_UNKILLABLE 0x00000040 /* for init: ignore fatal signals */ 692 693 /* If true, all threads except ->group_exit_task have pending SIGKILL */ 694 static inline int signal_group_exit(const struct signal_struct *sig) 695 { 696 return (sig->flags & SIGNAL_GROUP_EXIT) || 697 (sig->group_exit_task != NULL); 698 } 699 700 /* 701 * Some day this will be a full-fledged user tracking system.. 702 */ 703 struct user_struct { 704 atomic_t __count; /* reference count */ 705 atomic_t processes; /* How many processes does this user have? */ 706 atomic_t files; /* How many open files does this user have? */ 707 atomic_t sigpending; /* How many pending signals does this user have? */ 708 #ifdef CONFIG_INOTIFY_USER 709 atomic_t inotify_watches; /* How many inotify watches does this user have? */ 710 atomic_t inotify_devs; /* How many inotify devs does this user have opened? */ 711 #endif 712 #ifdef CONFIG_FANOTIFY 713 atomic_t fanotify_listeners; 714 #endif 715 #ifdef CONFIG_EPOLL 716 atomic_long_t epoll_watches; /* The number of file descriptors currently watched */ 717 #endif 718 #ifdef CONFIG_POSIX_MQUEUE 719 /* protected by mq_lock */ 720 unsigned long mq_bytes; /* How many bytes can be allocated to mqueue? */ 721 #endif 722 unsigned long locked_shm; /* How many pages of mlocked shm ? */ 723 724 #ifdef CONFIG_KEYS 725 struct key *uid_keyring; /* UID specific keyring */ 726 struct key *session_keyring; /* UID's default session keyring */ 727 #endif 728 729 /* Hash table maintenance information */ 730 struct hlist_node uidhash_node; 731 uid_t uid; 732 struct user_namespace *user_ns; 733 734 #ifdef CONFIG_PERF_EVENTS 735 atomic_long_t locked_vm; 736 #endif 737 }; 738 739 extern int uids_sysfs_init(void); 740 741 extern struct user_struct *find_user(uid_t); 742 743 extern struct user_struct root_user; 744 #define INIT_USER (&root_user) 745 746 747 struct backing_dev_info; 748 struct reclaim_state; 749 750 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 751 struct sched_info { 752 /* cumulative counters */ 753 unsigned long pcount; /* # of times run on this cpu */ 754 unsigned long long run_delay; /* time spent waiting on a runqueue */ 755 756 /* timestamps */ 757 unsigned long long last_arrival,/* when we last ran on a cpu */ 758 last_queued; /* when we were last queued to run */ 759 }; 760 #endif /* defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) */ 761 762 #ifdef CONFIG_TASK_DELAY_ACCT 763 struct task_delay_info { 764 spinlock_t lock; 765 unsigned int flags; /* Private per-task flags */ 766 767 /* For each stat XXX, add following, aligned appropriately 768 * 769 * struct timespec XXX_start, XXX_end; 770 * u64 XXX_delay; 771 * u32 XXX_count; 772 * 773 * Atomicity of updates to XXX_delay, XXX_count protected by 774 * single lock above (split into XXX_lock if contention is an issue). 775 */ 776 777 /* 778 * XXX_count is incremented on every XXX operation, the delay 779 * associated with the operation is added to XXX_delay. 780 * XXX_delay contains the accumulated delay time in nanoseconds. 781 */ 782 struct timespec blkio_start, blkio_end; /* Shared by blkio, swapin */ 783 u64 blkio_delay; /* wait for sync block io completion */ 784 u64 swapin_delay; /* wait for swapin block io completion */ 785 u32 blkio_count; /* total count of the number of sync block */ 786 /* io operations performed */ 787 u32 swapin_count; /* total count of the number of swapin block */ 788 /* io operations performed */ 789 790 struct timespec freepages_start, freepages_end; 791 u64 freepages_delay; /* wait for memory reclaim */ 792 u32 freepages_count; /* total count of memory reclaim */ 793 }; 794 #endif /* CONFIG_TASK_DELAY_ACCT */ 795 796 static inline int sched_info_on(void) 797 { 798 #ifdef CONFIG_SCHEDSTATS 799 return 1; 800 #elif defined(CONFIG_TASK_DELAY_ACCT) 801 extern int delayacct_on; 802 return delayacct_on; 803 #else 804 return 0; 805 #endif 806 } 807 808 enum cpu_idle_type { 809 CPU_IDLE, 810 CPU_NOT_IDLE, 811 CPU_NEWLY_IDLE, 812 CPU_MAX_IDLE_TYPES 813 }; 814 815 /* 816 * Increase resolution of nice-level calculations for 64-bit architectures. 817 * The extra resolution improves shares distribution and load balancing of 818 * low-weight task groups (eg. nice +19 on an autogroup), deeper taskgroup 819 * hierarchies, especially on larger systems. This is not a user-visible change 820 * and does not change the user-interface for setting shares/weights. 821 * 822 * We increase resolution only if we have enough bits to allow this increased 823 * resolution (i.e. BITS_PER_LONG > 32). The costs for increasing resolution 824 * when BITS_PER_LONG <= 32 are pretty high and the returns do not justify the 825 * increased costs. 826 */ 827 #if 0 /* BITS_PER_LONG > 32 -- currently broken: it increases power usage under light load */ 828 # define SCHED_LOAD_RESOLUTION 10 829 # define scale_load(w) ((w) << SCHED_LOAD_RESOLUTION) 830 # define scale_load_down(w) ((w) >> SCHED_LOAD_RESOLUTION) 831 #else 832 # define SCHED_LOAD_RESOLUTION 0 833 # define scale_load(w) (w) 834 # define scale_load_down(w) (w) 835 #endif 836 837 #define SCHED_LOAD_SHIFT (10 + SCHED_LOAD_RESOLUTION) 838 #define SCHED_LOAD_SCALE (1L << SCHED_LOAD_SHIFT) 839 840 /* 841 * Increase resolution of cpu_power calculations 842 */ 843 #define SCHED_POWER_SHIFT 10 844 #define SCHED_POWER_SCALE (1L << SCHED_POWER_SHIFT) 845 846 /* 847 * sched-domains (multiprocessor balancing) declarations: 848 */ 849 #ifdef CONFIG_SMP 850 #define SD_LOAD_BALANCE 0x0001 /* Do load balancing on this domain. */ 851 #define SD_BALANCE_NEWIDLE 0x0002 /* Balance when about to become idle */ 852 #define SD_BALANCE_EXEC 0x0004 /* Balance on exec */ 853 #define SD_BALANCE_FORK 0x0008 /* Balance on fork, clone */ 854 #define SD_BALANCE_WAKE 0x0010 /* Balance on wakeup */ 855 #define SD_WAKE_AFFINE 0x0020 /* Wake task to waking CPU */ 856 #define SD_PREFER_LOCAL 0x0040 /* Prefer to keep tasks local to this domain */ 857 #define SD_SHARE_CPUPOWER 0x0080 /* Domain members share cpu power */ 858 #define SD_POWERSAVINGS_BALANCE 0x0100 /* Balance for power savings */ 859 #define SD_SHARE_PKG_RESOURCES 0x0200 /* Domain members share cpu pkg resources */ 860 #define SD_SERIALIZE 0x0400 /* Only a single load balancing instance */ 861 #define SD_ASYM_PACKING 0x0800 /* Place busy groups earlier in the domain */ 862 #define SD_PREFER_SIBLING 0x1000 /* Prefer to place tasks in a sibling domain */ 863 #define SD_OVERLAP 0x2000 /* sched_domains of this level overlap */ 864 865 enum powersavings_balance_level { 866 POWERSAVINGS_BALANCE_NONE = 0, /* No power saving load balance */ 867 POWERSAVINGS_BALANCE_BASIC, /* Fill one thread/core/package 868 * first for long running threads 869 */ 870 POWERSAVINGS_BALANCE_WAKEUP, /* Also bias task wakeups to semi-idle 871 * cpu package for power savings 872 */ 873 MAX_POWERSAVINGS_BALANCE_LEVELS 874 }; 875 876 extern int sched_mc_power_savings, sched_smt_power_savings; 877 878 static inline int sd_balance_for_mc_power(void) 879 { 880 if (sched_smt_power_savings) 881 return SD_POWERSAVINGS_BALANCE; 882 883 if (!sched_mc_power_savings) 884 return SD_PREFER_SIBLING; 885 886 return 0; 887 } 888 889 static inline int sd_balance_for_package_power(void) 890 { 891 if (sched_mc_power_savings | sched_smt_power_savings) 892 return SD_POWERSAVINGS_BALANCE; 893 894 return SD_PREFER_SIBLING; 895 } 896 897 extern int __weak arch_sd_sibiling_asym_packing(void); 898 899 /* 900 * Optimise SD flags for power savings: 901 * SD_BALANCE_NEWIDLE helps aggressive task consolidation and power savings. 902 * Keep default SD flags if sched_{smt,mc}_power_saving=0 903 */ 904 905 static inline int sd_power_saving_flags(void) 906 { 907 if (sched_mc_power_savings | sched_smt_power_savings) 908 return SD_BALANCE_NEWIDLE; 909 910 return 0; 911 } 912 913 struct sched_group_power { 914 atomic_t ref; 915 /* 916 * CPU power of this group, SCHED_LOAD_SCALE being max power for a 917 * single CPU. 918 */ 919 unsigned int power, power_orig; 920 unsigned long next_update; 921 /* 922 * Number of busy cpus in this group. 923 */ 924 atomic_t nr_busy_cpus; 925 }; 926 927 struct sched_group { 928 struct sched_group *next; /* Must be a circular list */ 929 atomic_t ref; 930 931 unsigned int group_weight; 932 struct sched_group_power *sgp; 933 934 /* 935 * The CPUs this group covers. 936 * 937 * NOTE: this field is variable length. (Allocated dynamically 938 * by attaching extra space to the end of the structure, 939 * depending on how many CPUs the kernel has booted up with) 940 */ 941 unsigned long cpumask[0]; 942 }; 943 944 static inline struct cpumask *sched_group_cpus(struct sched_group *sg) 945 { 946 return to_cpumask(sg->cpumask); 947 } 948 949 /** 950 * group_first_cpu - Returns the first cpu in the cpumask of a sched_group. 951 * @group: The group whose first cpu is to be returned. 952 */ 953 static inline unsigned int group_first_cpu(struct sched_group *group) 954 { 955 return cpumask_first(sched_group_cpus(group)); 956 } 957 958 struct sched_domain_attr { 959 int relax_domain_level; 960 }; 961 962 #define SD_ATTR_INIT (struct sched_domain_attr) { \ 963 .relax_domain_level = -1, \ 964 } 965 966 extern int sched_domain_level_max; 967 968 struct sched_domain { 969 /* These fields must be setup */ 970 struct sched_domain *parent; /* top domain must be null terminated */ 971 struct sched_domain *child; /* bottom domain must be null terminated */ 972 struct sched_group *groups; /* the balancing groups of the domain */ 973 unsigned long min_interval; /* Minimum balance interval ms */ 974 unsigned long max_interval; /* Maximum balance interval ms */ 975 unsigned int busy_factor; /* less balancing by factor if busy */ 976 unsigned int imbalance_pct; /* No balance until over watermark */ 977 unsigned int cache_nice_tries; /* Leave cache hot tasks for # tries */ 978 unsigned int busy_idx; 979 unsigned int idle_idx; 980 unsigned int newidle_idx; 981 unsigned int wake_idx; 982 unsigned int forkexec_idx; 983 unsigned int smt_gain; 984 int flags; /* See SD_* */ 985 int level; 986 987 /* Runtime fields. */ 988 unsigned long last_balance; /* init to jiffies. units in jiffies */ 989 unsigned int balance_interval; /* initialise to 1. units in ms. */ 990 unsigned int nr_balance_failed; /* initialise to 0 */ 991 992 u64 last_update; 993 994 #ifdef CONFIG_SCHEDSTATS 995 /* load_balance() stats */ 996 unsigned int lb_count[CPU_MAX_IDLE_TYPES]; 997 unsigned int lb_failed[CPU_MAX_IDLE_TYPES]; 998 unsigned int lb_balanced[CPU_MAX_IDLE_TYPES]; 999 unsigned int lb_imbalance[CPU_MAX_IDLE_TYPES]; 1000 unsigned int lb_gained[CPU_MAX_IDLE_TYPES]; 1001 unsigned int lb_hot_gained[CPU_MAX_IDLE_TYPES]; 1002 unsigned int lb_nobusyg[CPU_MAX_IDLE_TYPES]; 1003 unsigned int lb_nobusyq[CPU_MAX_IDLE_TYPES]; 1004 1005 /* Active load balancing */ 1006 unsigned int alb_count; 1007 unsigned int alb_failed; 1008 unsigned int alb_pushed; 1009 1010 /* SD_BALANCE_EXEC stats */ 1011 unsigned int sbe_count; 1012 unsigned int sbe_balanced; 1013 unsigned int sbe_pushed; 1014 1015 /* SD_BALANCE_FORK stats */ 1016 unsigned int sbf_count; 1017 unsigned int sbf_balanced; 1018 unsigned int sbf_pushed; 1019 1020 /* try_to_wake_up() stats */ 1021 unsigned int ttwu_wake_remote; 1022 unsigned int ttwu_move_affine; 1023 unsigned int ttwu_move_balance; 1024 #endif 1025 #ifdef CONFIG_SCHED_DEBUG 1026 char *name; 1027 #endif 1028 union { 1029 void *private; /* used during construction */ 1030 struct rcu_head rcu; /* used during destruction */ 1031 }; 1032 1033 unsigned int span_weight; 1034 /* 1035 * Span of all CPUs in this domain. 1036 * 1037 * NOTE: this field is variable length. (Allocated dynamically 1038 * by attaching extra space to the end of the structure, 1039 * depending on how many CPUs the kernel has booted up with) 1040 */ 1041 unsigned long span[0]; 1042 }; 1043 1044 static inline struct cpumask *sched_domain_span(struct sched_domain *sd) 1045 { 1046 return to_cpumask(sd->span); 1047 } 1048 1049 extern void partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 1050 struct sched_domain_attr *dattr_new); 1051 1052 /* Allocate an array of sched domains, for partition_sched_domains(). */ 1053 cpumask_var_t *alloc_sched_domains(unsigned int ndoms); 1054 void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms); 1055 1056 /* Test a flag in parent sched domain */ 1057 static inline int test_sd_parent(struct sched_domain *sd, int flag) 1058 { 1059 if (sd->parent && (sd->parent->flags & flag)) 1060 return 1; 1061 1062 return 0; 1063 } 1064 1065 unsigned long default_scale_freq_power(struct sched_domain *sd, int cpu); 1066 unsigned long default_scale_smt_power(struct sched_domain *sd, int cpu); 1067 1068 bool cpus_share_cache(int this_cpu, int that_cpu); 1069 1070 #else /* CONFIG_SMP */ 1071 1072 struct sched_domain_attr; 1073 1074 static inline void 1075 partition_sched_domains(int ndoms_new, cpumask_var_t doms_new[], 1076 struct sched_domain_attr *dattr_new) 1077 { 1078 } 1079 1080 static inline bool cpus_share_cache(int this_cpu, int that_cpu) 1081 { 1082 return true; 1083 } 1084 1085 #endif /* !CONFIG_SMP */ 1086 1087 1088 struct io_context; /* See blkdev.h */ 1089 1090 1091 #ifdef ARCH_HAS_PREFETCH_SWITCH_STACK 1092 extern void prefetch_stack(struct task_struct *t); 1093 #else 1094 static inline void prefetch_stack(struct task_struct *t) { } 1095 #endif 1096 1097 struct audit_context; /* See audit.c */ 1098 struct mempolicy; 1099 struct pipe_inode_info; 1100 struct uts_namespace; 1101 1102 struct rq; 1103 struct sched_domain; 1104 1105 /* 1106 * wake flags 1107 */ 1108 #define WF_SYNC 0x01 /* waker goes to sleep after wakup */ 1109 #define WF_FORK 0x02 /* child wakeup after fork */ 1110 #define WF_MIGRATED 0x04 /* internal use, task got migrated */ 1111 1112 #define ENQUEUE_WAKEUP 1 1113 #define ENQUEUE_HEAD 2 1114 #ifdef CONFIG_SMP 1115 #define ENQUEUE_WAKING 4 /* sched_class::task_waking was called */ 1116 #else 1117 #define ENQUEUE_WAKING 0 1118 #endif 1119 1120 #define DEQUEUE_SLEEP 1 1121 1122 struct sched_class { 1123 const struct sched_class *next; 1124 1125 void (*enqueue_task) (struct rq *rq, struct task_struct *p, int flags); 1126 void (*dequeue_task) (struct rq *rq, struct task_struct *p, int flags); 1127 void (*yield_task) (struct rq *rq); 1128 bool (*yield_to_task) (struct rq *rq, struct task_struct *p, bool preempt); 1129 1130 void (*check_preempt_curr) (struct rq *rq, struct task_struct *p, int flags); 1131 1132 struct task_struct * (*pick_next_task) (struct rq *rq); 1133 void (*put_prev_task) (struct rq *rq, struct task_struct *p); 1134 1135 #ifdef CONFIG_SMP 1136 int (*select_task_rq)(struct task_struct *p, int sd_flag, int flags); 1137 1138 void (*pre_schedule) (struct rq *this_rq, struct task_struct *task); 1139 void (*post_schedule) (struct rq *this_rq); 1140 void (*task_waking) (struct task_struct *task); 1141 void (*task_woken) (struct rq *this_rq, struct task_struct *task); 1142 1143 void (*set_cpus_allowed)(struct task_struct *p, 1144 const struct cpumask *newmask); 1145 1146 void (*rq_online)(struct rq *rq); 1147 void (*rq_offline)(struct rq *rq); 1148 #endif 1149 1150 void (*set_curr_task) (struct rq *rq); 1151 void (*task_tick) (struct rq *rq, struct task_struct *p, int queued); 1152 void (*task_fork) (struct task_struct *p); 1153 1154 void (*switched_from) (struct rq *this_rq, struct task_struct *task); 1155 void (*switched_to) (struct rq *this_rq, struct task_struct *task); 1156 void (*prio_changed) (struct rq *this_rq, struct task_struct *task, 1157 int oldprio); 1158 1159 unsigned int (*get_rr_interval) (struct rq *rq, 1160 struct task_struct *task); 1161 1162 #ifdef CONFIG_FAIR_GROUP_SCHED 1163 void (*task_move_group) (struct task_struct *p, int on_rq); 1164 #endif 1165 }; 1166 1167 struct load_weight { 1168 unsigned long weight, inv_weight; 1169 }; 1170 1171 #ifdef CONFIG_SCHEDSTATS 1172 struct sched_statistics { 1173 u64 wait_start; 1174 u64 wait_max; 1175 u64 wait_count; 1176 u64 wait_sum; 1177 u64 iowait_count; 1178 u64 iowait_sum; 1179 1180 u64 sleep_start; 1181 u64 sleep_max; 1182 s64 sum_sleep_runtime; 1183 1184 u64 block_start; 1185 u64 block_max; 1186 u64 exec_max; 1187 u64 slice_max; 1188 1189 u64 nr_migrations_cold; 1190 u64 nr_failed_migrations_affine; 1191 u64 nr_failed_migrations_running; 1192 u64 nr_failed_migrations_hot; 1193 u64 nr_forced_migrations; 1194 1195 u64 nr_wakeups; 1196 u64 nr_wakeups_sync; 1197 u64 nr_wakeups_migrate; 1198 u64 nr_wakeups_local; 1199 u64 nr_wakeups_remote; 1200 u64 nr_wakeups_affine; 1201 u64 nr_wakeups_affine_attempts; 1202 u64 nr_wakeups_passive; 1203 u64 nr_wakeups_idle; 1204 }; 1205 #endif 1206 1207 struct sched_entity { 1208 struct load_weight load; /* for load-balancing */ 1209 struct rb_node run_node; 1210 struct list_head group_node; 1211 unsigned int on_rq; 1212 1213 u64 exec_start; 1214 u64 sum_exec_runtime; 1215 u64 vruntime; 1216 u64 prev_sum_exec_runtime; 1217 1218 u64 nr_migrations; 1219 1220 #ifdef CONFIG_SCHEDSTATS 1221 struct sched_statistics statistics; 1222 #endif 1223 1224 #ifdef CONFIG_FAIR_GROUP_SCHED 1225 struct sched_entity *parent; 1226 /* rq on which this entity is (to be) queued: */ 1227 struct cfs_rq *cfs_rq; 1228 /* rq "owned" by this entity/group: */ 1229 struct cfs_rq *my_q; 1230 #endif 1231 }; 1232 1233 struct sched_rt_entity { 1234 struct list_head run_list; 1235 unsigned long timeout; 1236 unsigned int time_slice; 1237 int nr_cpus_allowed; 1238 1239 struct sched_rt_entity *back; 1240 #ifdef CONFIG_RT_GROUP_SCHED 1241 struct sched_rt_entity *parent; 1242 /* rq on which this entity is (to be) queued: */ 1243 struct rt_rq *rt_rq; 1244 /* rq "owned" by this entity/group: */ 1245 struct rt_rq *my_q; 1246 #endif 1247 }; 1248 1249 /* 1250 * default timeslice is 100 msecs (used only for SCHED_RR tasks). 1251 * Timeslices get refilled after they expire. 1252 */ 1253 #define RR_TIMESLICE (100 * HZ / 1000) 1254 1255 struct rcu_node; 1256 1257 enum perf_event_task_context { 1258 perf_invalid_context = -1, 1259 perf_hw_context = 0, 1260 perf_sw_context, 1261 perf_nr_task_contexts, 1262 }; 1263 1264 struct task_struct { 1265 volatile long state; /* -1 unrunnable, 0 runnable, >0 stopped */ 1266 void *stack; 1267 atomic_t usage; 1268 unsigned int flags; /* per process flags, defined below */ 1269 unsigned int ptrace; 1270 1271 #ifdef CONFIG_SMP 1272 struct llist_node wake_entry; 1273 int on_cpu; 1274 #endif 1275 int on_rq; 1276 1277 int prio, static_prio, normal_prio; 1278 unsigned int rt_priority; 1279 const struct sched_class *sched_class; 1280 struct sched_entity se; 1281 struct sched_rt_entity rt; 1282 1283 #ifdef CONFIG_PREEMPT_NOTIFIERS 1284 /* list of struct preempt_notifier: */ 1285 struct hlist_head preempt_notifiers; 1286 #endif 1287 1288 /* 1289 * fpu_counter contains the number of consecutive context switches 1290 * that the FPU is used. If this is over a threshold, the lazy fpu 1291 * saving becomes unlazy to save the trap. This is an unsigned char 1292 * so that after 256 times the counter wraps and the behavior turns 1293 * lazy again; this to deal with bursty apps that only use FPU for 1294 * a short time 1295 */ 1296 unsigned char fpu_counter; 1297 #ifdef CONFIG_BLK_DEV_IO_TRACE 1298 unsigned int btrace_seq; 1299 #endif 1300 1301 unsigned int policy; 1302 cpumask_t cpus_allowed; 1303 1304 #ifdef CONFIG_PREEMPT_RCU 1305 int rcu_read_lock_nesting; 1306 char rcu_read_unlock_special; 1307 struct list_head rcu_node_entry; 1308 #endif /* #ifdef CONFIG_PREEMPT_RCU */ 1309 #ifdef CONFIG_TREE_PREEMPT_RCU 1310 struct rcu_node *rcu_blocked_node; 1311 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ 1312 #ifdef CONFIG_RCU_BOOST 1313 struct rt_mutex *rcu_boost_mutex; 1314 #endif /* #ifdef CONFIG_RCU_BOOST */ 1315 1316 #if defined(CONFIG_SCHEDSTATS) || defined(CONFIG_TASK_DELAY_ACCT) 1317 struct sched_info sched_info; 1318 #endif 1319 1320 struct list_head tasks; 1321 #ifdef CONFIG_SMP 1322 struct plist_node pushable_tasks; 1323 #endif 1324 1325 struct mm_struct *mm, *active_mm; 1326 #ifdef CONFIG_COMPAT_BRK 1327 unsigned brk_randomized:1; 1328 #endif 1329 #if defined(SPLIT_RSS_COUNTING) 1330 struct task_rss_stat rss_stat; 1331 #endif 1332 /* task state */ 1333 int exit_state; 1334 int exit_code, exit_signal; 1335 int pdeath_signal; /* The signal sent when the parent dies */ 1336 unsigned int jobctl; /* JOBCTL_*, siglock protected */ 1337 /* ??? */ 1338 unsigned int personality; 1339 unsigned did_exec:1; 1340 unsigned in_execve:1; /* Tell the LSMs that the process is doing an 1341 * execve */ 1342 unsigned in_iowait:1; 1343 1344 1345 /* Revert to default priority/policy when forking */ 1346 unsigned sched_reset_on_fork:1; 1347 unsigned sched_contributes_to_load:1; 1348 1349 #ifdef CONFIG_GENERIC_HARDIRQS 1350 /* IRQ handler threads */ 1351 unsigned irq_thread:1; 1352 #endif 1353 1354 pid_t pid; 1355 pid_t tgid; 1356 1357 #ifdef CONFIG_CC_STACKPROTECTOR 1358 /* Canary value for the -fstack-protector gcc feature */ 1359 unsigned long stack_canary; 1360 #endif 1361 1362 /* 1363 * pointers to (original) parent process, youngest child, younger sibling, 1364 * older sibling, respectively. (p->father can be replaced with 1365 * p->real_parent->pid) 1366 */ 1367 struct task_struct __rcu *real_parent; /* real parent process */ 1368 struct task_struct __rcu *parent; /* recipient of SIGCHLD, wait4() reports */ 1369 /* 1370 * children/sibling forms the list of my natural children 1371 */ 1372 struct list_head children; /* list of my children */ 1373 struct list_head sibling; /* linkage in my parent's children list */ 1374 struct task_struct *group_leader; /* threadgroup leader */ 1375 1376 /* 1377 * ptraced is the list of tasks this task is using ptrace on. 1378 * This includes both natural children and PTRACE_ATTACH targets. 1379 * p->ptrace_entry is p's link on the p->parent->ptraced list. 1380 */ 1381 struct list_head ptraced; 1382 struct list_head ptrace_entry; 1383 1384 /* PID/PID hash table linkage. */ 1385 struct pid_link pids[PIDTYPE_MAX]; 1386 struct list_head thread_group; 1387 1388 struct completion *vfork_done; /* for vfork() */ 1389 int __user *set_child_tid; /* CLONE_CHILD_SETTID */ 1390 int __user *clear_child_tid; /* CLONE_CHILD_CLEARTID */ 1391 1392 cputime_t utime, stime, utimescaled, stimescaled; 1393 cputime_t gtime; 1394 #ifndef CONFIG_VIRT_CPU_ACCOUNTING 1395 cputime_t prev_utime, prev_stime; 1396 #endif 1397 unsigned long nvcsw, nivcsw; /* context switch counts */ 1398 struct timespec start_time; /* monotonic time */ 1399 struct timespec real_start_time; /* boot based time */ 1400 /* mm fault and swap info: this can arguably be seen as either mm-specific or thread-specific */ 1401 unsigned long min_flt, maj_flt; 1402 1403 struct task_cputime cputime_expires; 1404 struct list_head cpu_timers[3]; 1405 1406 /* process credentials */ 1407 const struct cred __rcu *real_cred; /* objective and real subjective task 1408 * credentials (COW) */ 1409 const struct cred __rcu *cred; /* effective (overridable) subjective task 1410 * credentials (COW) */ 1411 struct cred *replacement_session_keyring; /* for KEYCTL_SESSION_TO_PARENT */ 1412 1413 char comm[TASK_COMM_LEN]; /* executable name excluding path 1414 - access with [gs]et_task_comm (which lock 1415 it with task_lock()) 1416 - initialized normally by setup_new_exec */ 1417 /* file system info */ 1418 int link_count, total_link_count; 1419 #ifdef CONFIG_SYSVIPC 1420 /* ipc stuff */ 1421 struct sysv_sem sysvsem; 1422 #endif 1423 #ifdef CONFIG_DETECT_HUNG_TASK 1424 /* hung task detection */ 1425 unsigned long last_switch_count; 1426 #endif 1427 /* CPU-specific state of this task */ 1428 struct thread_struct thread; 1429 /* filesystem information */ 1430 struct fs_struct *fs; 1431 /* open file information */ 1432 struct files_struct *files; 1433 /* namespaces */ 1434 struct nsproxy *nsproxy; 1435 /* signal handlers */ 1436 struct signal_struct *signal; 1437 struct sighand_struct *sighand; 1438 1439 sigset_t blocked, real_blocked; 1440 sigset_t saved_sigmask; /* restored if set_restore_sigmask() was used */ 1441 struct sigpending pending; 1442 1443 unsigned long sas_ss_sp; 1444 size_t sas_ss_size; 1445 int (*notifier)(void *priv); 1446 void *notifier_data; 1447 sigset_t *notifier_mask; 1448 struct audit_context *audit_context; 1449 #ifdef CONFIG_AUDITSYSCALL 1450 uid_t loginuid; 1451 unsigned int sessionid; 1452 #endif 1453 seccomp_t seccomp; 1454 1455 /* Thread group tracking */ 1456 u32 parent_exec_id; 1457 u32 self_exec_id; 1458 /* Protection of (de-)allocation: mm, files, fs, tty, keyrings, mems_allowed, 1459 * mempolicy */ 1460 spinlock_t alloc_lock; 1461 1462 /* Protection of the PI data structures: */ 1463 raw_spinlock_t pi_lock; 1464 1465 #ifdef CONFIG_RT_MUTEXES 1466 /* PI waiters blocked on a rt_mutex held by this task */ 1467 struct plist_head pi_waiters; 1468 /* Deadlock detection and priority inheritance handling */ 1469 struct rt_mutex_waiter *pi_blocked_on; 1470 #endif 1471 1472 #ifdef CONFIG_DEBUG_MUTEXES 1473 /* mutex deadlock detection */ 1474 struct mutex_waiter *blocked_on; 1475 #endif 1476 #ifdef CONFIG_TRACE_IRQFLAGS 1477 unsigned int irq_events; 1478 unsigned long hardirq_enable_ip; 1479 unsigned long hardirq_disable_ip; 1480 unsigned int hardirq_enable_event; 1481 unsigned int hardirq_disable_event; 1482 int hardirqs_enabled; 1483 int hardirq_context; 1484 unsigned long softirq_disable_ip; 1485 unsigned long softirq_enable_ip; 1486 unsigned int softirq_disable_event; 1487 unsigned int softirq_enable_event; 1488 int softirqs_enabled; 1489 int softirq_context; 1490 #endif 1491 #ifdef CONFIG_LOCKDEP 1492 # define MAX_LOCK_DEPTH 48UL 1493 u64 curr_chain_key; 1494 int lockdep_depth; 1495 unsigned int lockdep_recursion; 1496 struct held_lock held_locks[MAX_LOCK_DEPTH]; 1497 gfp_t lockdep_reclaim_gfp; 1498 #endif 1499 1500 /* journalling filesystem info */ 1501 void *journal_info; 1502 1503 /* stacked block device info */ 1504 struct bio_list *bio_list; 1505 1506 #ifdef CONFIG_BLOCK 1507 /* stack plugging */ 1508 struct blk_plug *plug; 1509 #endif 1510 1511 /* VM state */ 1512 struct reclaim_state *reclaim_state; 1513 1514 struct backing_dev_info *backing_dev_info; 1515 1516 struct io_context *io_context; 1517 1518 unsigned long ptrace_message; 1519 siginfo_t *last_siginfo; /* For ptrace use. */ 1520 struct task_io_accounting ioac; 1521 #if defined(CONFIG_TASK_XACCT) 1522 u64 acct_rss_mem1; /* accumulated rss usage */ 1523 u64 acct_vm_mem1; /* accumulated virtual memory usage */ 1524 cputime_t acct_timexpd; /* stime + utime since last update */ 1525 #endif 1526 #ifdef CONFIG_CPUSETS 1527 nodemask_t mems_allowed; /* Protected by alloc_lock */ 1528 seqcount_t mems_allowed_seq; /* Seqence no to catch updates */ 1529 int cpuset_mem_spread_rotor; 1530 int cpuset_slab_spread_rotor; 1531 #endif 1532 #ifdef CONFIG_CGROUPS 1533 /* Control Group info protected by css_set_lock */ 1534 struct css_set __rcu *cgroups; 1535 /* cg_list protected by css_set_lock and tsk->alloc_lock */ 1536 struct list_head cg_list; 1537 #endif 1538 #ifdef CONFIG_FUTEX 1539 struct robust_list_head __user *robust_list; 1540 #ifdef CONFIG_COMPAT 1541 struct compat_robust_list_head __user *compat_robust_list; 1542 #endif 1543 struct list_head pi_state_list; 1544 struct futex_pi_state *pi_state_cache; 1545 #endif 1546 #ifdef CONFIG_PERF_EVENTS 1547 struct perf_event_context *perf_event_ctxp[perf_nr_task_contexts]; 1548 struct mutex perf_event_mutex; 1549 struct list_head perf_event_list; 1550 #endif 1551 #ifdef CONFIG_NUMA 1552 struct mempolicy *mempolicy; /* Protected by alloc_lock */ 1553 short il_next; 1554 short pref_node_fork; 1555 #endif 1556 struct rcu_head rcu; 1557 1558 /* 1559 * cache last used pipe for splice 1560 */ 1561 struct pipe_inode_info *splice_pipe; 1562 #ifdef CONFIG_TASK_DELAY_ACCT 1563 struct task_delay_info *delays; 1564 #endif 1565 #ifdef CONFIG_FAULT_INJECTION 1566 int make_it_fail; 1567 #endif 1568 /* 1569 * when (nr_dirtied >= nr_dirtied_pause), it's time to call 1570 * balance_dirty_pages() for some dirty throttling pause 1571 */ 1572 int nr_dirtied; 1573 int nr_dirtied_pause; 1574 unsigned long dirty_paused_when; /* start of a write-and-pause period */ 1575 1576 #ifdef CONFIG_LATENCYTOP 1577 int latency_record_count; 1578 struct latency_record latency_record[LT_SAVECOUNT]; 1579 #endif 1580 /* 1581 * time slack values; these are used to round up poll() and 1582 * select() etc timeout values. These are in nanoseconds. 1583 */ 1584 unsigned long timer_slack_ns; 1585 unsigned long default_timer_slack_ns; 1586 1587 struct list_head *scm_work_list; 1588 #ifdef CONFIG_FUNCTION_GRAPH_TRACER 1589 /* Index of current stored address in ret_stack */ 1590 int curr_ret_stack; 1591 /* Stack of return addresses for return function tracing */ 1592 struct ftrace_ret_stack *ret_stack; 1593 /* time stamp for last schedule */ 1594 unsigned long long ftrace_timestamp; 1595 /* 1596 * Number of functions that haven't been traced 1597 * because of depth overrun. 1598 */ 1599 atomic_t trace_overrun; 1600 /* Pause for the tracing */ 1601 atomic_t tracing_graph_pause; 1602 #endif 1603 #ifdef CONFIG_TRACING 1604 /* state flags for use by tracers */ 1605 unsigned long trace; 1606 /* bitmask and counter of trace recursion */ 1607 unsigned long trace_recursion; 1608 #endif /* CONFIG_TRACING */ 1609 #ifdef CONFIG_CGROUP_MEM_RES_CTLR /* memcg uses this to do batch job */ 1610 struct memcg_batch_info { 1611 int do_batch; /* incremented when batch uncharge started */ 1612 struct mem_cgroup *memcg; /* target memcg of uncharge */ 1613 unsigned long nr_pages; /* uncharged usage */ 1614 unsigned long memsw_nr_pages; /* uncharged mem+swap usage */ 1615 } memcg_batch; 1616 #endif 1617 #ifdef CONFIG_HAVE_HW_BREAKPOINT 1618 atomic_t ptrace_bp_refcnt; 1619 #endif 1620 }; 1621 1622 /* Future-safe accessor for struct task_struct's cpus_allowed. */ 1623 #define tsk_cpus_allowed(tsk) (&(tsk)->cpus_allowed) 1624 1625 /* 1626 * Priority of a process goes from 0..MAX_PRIO-1, valid RT 1627 * priority is 0..MAX_RT_PRIO-1, and SCHED_NORMAL/SCHED_BATCH 1628 * tasks are in the range MAX_RT_PRIO..MAX_PRIO-1. Priority 1629 * values are inverted: lower p->prio value means higher priority. 1630 * 1631 * The MAX_USER_RT_PRIO value allows the actual maximum 1632 * RT priority to be separate from the value exported to 1633 * user-space. This allows kernel threads to set their 1634 * priority to a value higher than any user task. Note: 1635 * MAX_RT_PRIO must not be smaller than MAX_USER_RT_PRIO. 1636 */ 1637 1638 #define MAX_USER_RT_PRIO 100 1639 #define MAX_RT_PRIO MAX_USER_RT_PRIO 1640 1641 #define MAX_PRIO (MAX_RT_PRIO + 40) 1642 #define DEFAULT_PRIO (MAX_RT_PRIO + 20) 1643 1644 static inline int rt_prio(int prio) 1645 { 1646 if (unlikely(prio < MAX_RT_PRIO)) 1647 return 1; 1648 return 0; 1649 } 1650 1651 static inline int rt_task(struct task_struct *p) 1652 { 1653 return rt_prio(p->prio); 1654 } 1655 1656 static inline struct pid *task_pid(struct task_struct *task) 1657 { 1658 return task->pids[PIDTYPE_PID].pid; 1659 } 1660 1661 static inline struct pid *task_tgid(struct task_struct *task) 1662 { 1663 return task->group_leader->pids[PIDTYPE_PID].pid; 1664 } 1665 1666 /* 1667 * Without tasklist or rcu lock it is not safe to dereference 1668 * the result of task_pgrp/task_session even if task == current, 1669 * we can race with another thread doing sys_setsid/sys_setpgid. 1670 */ 1671 static inline struct pid *task_pgrp(struct task_struct *task) 1672 { 1673 return task->group_leader->pids[PIDTYPE_PGID].pid; 1674 } 1675 1676 static inline struct pid *task_session(struct task_struct *task) 1677 { 1678 return task->group_leader->pids[PIDTYPE_SID].pid; 1679 } 1680 1681 struct pid_namespace; 1682 1683 /* 1684 * the helpers to get the task's different pids as they are seen 1685 * from various namespaces 1686 * 1687 * task_xid_nr() : global id, i.e. the id seen from the init namespace; 1688 * task_xid_vnr() : virtual id, i.e. the id seen from the pid namespace of 1689 * current. 1690 * task_xid_nr_ns() : id seen from the ns specified; 1691 * 1692 * set_task_vxid() : assigns a virtual id to a task; 1693 * 1694 * see also pid_nr() etc in include/linux/pid.h 1695 */ 1696 pid_t __task_pid_nr_ns(struct task_struct *task, enum pid_type type, 1697 struct pid_namespace *ns); 1698 1699 static inline pid_t task_pid_nr(struct task_struct *tsk) 1700 { 1701 return tsk->pid; 1702 } 1703 1704 static inline pid_t task_pid_nr_ns(struct task_struct *tsk, 1705 struct pid_namespace *ns) 1706 { 1707 return __task_pid_nr_ns(tsk, PIDTYPE_PID, ns); 1708 } 1709 1710 static inline pid_t task_pid_vnr(struct task_struct *tsk) 1711 { 1712 return __task_pid_nr_ns(tsk, PIDTYPE_PID, NULL); 1713 } 1714 1715 1716 static inline pid_t task_tgid_nr(struct task_struct *tsk) 1717 { 1718 return tsk->tgid; 1719 } 1720 1721 pid_t task_tgid_nr_ns(struct task_struct *tsk, struct pid_namespace *ns); 1722 1723 static inline pid_t task_tgid_vnr(struct task_struct *tsk) 1724 { 1725 return pid_vnr(task_tgid(tsk)); 1726 } 1727 1728 1729 static inline pid_t task_pgrp_nr_ns(struct task_struct *tsk, 1730 struct pid_namespace *ns) 1731 { 1732 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, ns); 1733 } 1734 1735 static inline pid_t task_pgrp_vnr(struct task_struct *tsk) 1736 { 1737 return __task_pid_nr_ns(tsk, PIDTYPE_PGID, NULL); 1738 } 1739 1740 1741 static inline pid_t task_session_nr_ns(struct task_struct *tsk, 1742 struct pid_namespace *ns) 1743 { 1744 return __task_pid_nr_ns(tsk, PIDTYPE_SID, ns); 1745 } 1746 1747 static inline pid_t task_session_vnr(struct task_struct *tsk) 1748 { 1749 return __task_pid_nr_ns(tsk, PIDTYPE_SID, NULL); 1750 } 1751 1752 /* obsolete, do not use */ 1753 static inline pid_t task_pgrp_nr(struct task_struct *tsk) 1754 { 1755 return task_pgrp_nr_ns(tsk, &init_pid_ns); 1756 } 1757 1758 /** 1759 * pid_alive - check that a task structure is not stale 1760 * @p: Task structure to be checked. 1761 * 1762 * Test if a process is not yet dead (at most zombie state) 1763 * If pid_alive fails, then pointers within the task structure 1764 * can be stale and must not be dereferenced. 1765 */ 1766 static inline int pid_alive(struct task_struct *p) 1767 { 1768 return p->pids[PIDTYPE_PID].pid != NULL; 1769 } 1770 1771 /** 1772 * is_global_init - check if a task structure is init 1773 * @tsk: Task structure to be checked. 1774 * 1775 * Check if a task structure is the first user space task the kernel created. 1776 */ 1777 static inline int is_global_init(struct task_struct *tsk) 1778 { 1779 return tsk->pid == 1; 1780 } 1781 1782 /* 1783 * is_container_init: 1784 * check whether in the task is init in its own pid namespace. 1785 */ 1786 extern int is_container_init(struct task_struct *tsk); 1787 1788 extern struct pid *cad_pid; 1789 1790 extern void free_task(struct task_struct *tsk); 1791 #define get_task_struct(tsk) do { atomic_inc(&(tsk)->usage); } while(0) 1792 1793 extern void __put_task_struct(struct task_struct *t); 1794 1795 static inline void put_task_struct(struct task_struct *t) 1796 { 1797 if (atomic_dec_and_test(&t->usage)) 1798 __put_task_struct(t); 1799 } 1800 1801 extern void task_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1802 extern void thread_group_times(struct task_struct *p, cputime_t *ut, cputime_t *st); 1803 1804 /* 1805 * Per process flags 1806 */ 1807 #define PF_EXITING 0x00000004 /* getting shut down */ 1808 #define PF_EXITPIDONE 0x00000008 /* pi exit done on shut down */ 1809 #define PF_VCPU 0x00000010 /* I'm a virtual CPU */ 1810 #define PF_WQ_WORKER 0x00000020 /* I'm a workqueue worker */ 1811 #define PF_FORKNOEXEC 0x00000040 /* forked but didn't exec */ 1812 #define PF_MCE_PROCESS 0x00000080 /* process policy on mce errors */ 1813 #define PF_SUPERPRIV 0x00000100 /* used super-user privileges */ 1814 #define PF_DUMPCORE 0x00000200 /* dumped core */ 1815 #define PF_SIGNALED 0x00000400 /* killed by a signal */ 1816 #define PF_MEMALLOC 0x00000800 /* Allocating memory */ 1817 #define PF_NPROC_EXCEEDED 0x00001000 /* set_user noticed that RLIMIT_NPROC was exceeded */ 1818 #define PF_USED_MATH 0x00002000 /* if unset the fpu must be initialized before use */ 1819 #define PF_NOFREEZE 0x00008000 /* this thread should not be frozen */ 1820 #define PF_FROZEN 0x00010000 /* frozen for system suspend */ 1821 #define PF_FSTRANS 0x00020000 /* inside a filesystem transaction */ 1822 #define PF_KSWAPD 0x00040000 /* I am kswapd */ 1823 #define PF_LESS_THROTTLE 0x00100000 /* Throttle me less: I clean memory */ 1824 #define PF_KTHREAD 0x00200000 /* I am a kernel thread */ 1825 #define PF_RANDOMIZE 0x00400000 /* randomize virtual address space */ 1826 #define PF_SWAPWRITE 0x00800000 /* Allowed to write to swap */ 1827 #define PF_SPREAD_PAGE 0x01000000 /* Spread page cache over cpuset */ 1828 #define PF_SPREAD_SLAB 0x02000000 /* Spread some slab caches over cpuset */ 1829 #define PF_THREAD_BOUND 0x04000000 /* Thread bound to specific cpu */ 1830 #define PF_MCE_EARLY 0x08000000 /* Early kill for mce process policy */ 1831 #define PF_MEMPOLICY 0x10000000 /* Non-default NUMA mempolicy */ 1832 #define PF_MUTEX_TESTER 0x20000000 /* Thread belongs to the rt mutex tester */ 1833 #define PF_FREEZER_SKIP 0x40000000 /* Freezer should not count it as freezable */ 1834 1835 /* 1836 * Only the _current_ task can read/write to tsk->flags, but other 1837 * tasks can access tsk->flags in readonly mode for example 1838 * with tsk_used_math (like during threaded core dumping). 1839 * There is however an exception to this rule during ptrace 1840 * or during fork: the ptracer task is allowed to write to the 1841 * child->flags of its traced child (same goes for fork, the parent 1842 * can write to the child->flags), because we're guaranteed the 1843 * child is not running and in turn not changing child->flags 1844 * at the same time the parent does it. 1845 */ 1846 #define clear_stopped_child_used_math(child) do { (child)->flags &= ~PF_USED_MATH; } while (0) 1847 #define set_stopped_child_used_math(child) do { (child)->flags |= PF_USED_MATH; } while (0) 1848 #define clear_used_math() clear_stopped_child_used_math(current) 1849 #define set_used_math() set_stopped_child_used_math(current) 1850 #define conditional_stopped_child_used_math(condition, child) \ 1851 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= (condition) ? PF_USED_MATH : 0; } while (0) 1852 #define conditional_used_math(condition) \ 1853 conditional_stopped_child_used_math(condition, current) 1854 #define copy_to_stopped_child_used_math(child) \ 1855 do { (child)->flags &= ~PF_USED_MATH, (child)->flags |= current->flags & PF_USED_MATH; } while (0) 1856 /* NOTE: this will return 0 or PF_USED_MATH, it will never return 1 */ 1857 #define tsk_used_math(p) ((p)->flags & PF_USED_MATH) 1858 #define used_math() tsk_used_math(current) 1859 1860 /* 1861 * task->jobctl flags 1862 */ 1863 #define JOBCTL_STOP_SIGMASK 0xffff /* signr of the last group stop */ 1864 1865 #define JOBCTL_STOP_DEQUEUED_BIT 16 /* stop signal dequeued */ 1866 #define JOBCTL_STOP_PENDING_BIT 17 /* task should stop for group stop */ 1867 #define JOBCTL_STOP_CONSUME_BIT 18 /* consume group stop count */ 1868 #define JOBCTL_TRAP_STOP_BIT 19 /* trap for STOP */ 1869 #define JOBCTL_TRAP_NOTIFY_BIT 20 /* trap for NOTIFY */ 1870 #define JOBCTL_TRAPPING_BIT 21 /* switching to TRACED */ 1871 #define JOBCTL_LISTENING_BIT 22 /* ptracer is listening for events */ 1872 1873 #define JOBCTL_STOP_DEQUEUED (1 << JOBCTL_STOP_DEQUEUED_BIT) 1874 #define JOBCTL_STOP_PENDING (1 << JOBCTL_STOP_PENDING_BIT) 1875 #define JOBCTL_STOP_CONSUME (1 << JOBCTL_STOP_CONSUME_BIT) 1876 #define JOBCTL_TRAP_STOP (1 << JOBCTL_TRAP_STOP_BIT) 1877 #define JOBCTL_TRAP_NOTIFY (1 << JOBCTL_TRAP_NOTIFY_BIT) 1878 #define JOBCTL_TRAPPING (1 << JOBCTL_TRAPPING_BIT) 1879 #define JOBCTL_LISTENING (1 << JOBCTL_LISTENING_BIT) 1880 1881 #define JOBCTL_TRAP_MASK (JOBCTL_TRAP_STOP | JOBCTL_TRAP_NOTIFY) 1882 #define JOBCTL_PENDING_MASK (JOBCTL_STOP_PENDING | JOBCTL_TRAP_MASK) 1883 1884 extern bool task_set_jobctl_pending(struct task_struct *task, 1885 unsigned int mask); 1886 extern void task_clear_jobctl_trapping(struct task_struct *task); 1887 extern void task_clear_jobctl_pending(struct task_struct *task, 1888 unsigned int mask); 1889 1890 #ifdef CONFIG_PREEMPT_RCU 1891 1892 #define RCU_READ_UNLOCK_BLOCKED (1 << 0) /* blocked while in RCU read-side. */ 1893 #define RCU_READ_UNLOCK_NEED_QS (1 << 1) /* RCU core needs CPU response. */ 1894 1895 static inline void rcu_copy_process(struct task_struct *p) 1896 { 1897 p->rcu_read_lock_nesting = 0; 1898 p->rcu_read_unlock_special = 0; 1899 #ifdef CONFIG_TREE_PREEMPT_RCU 1900 p->rcu_blocked_node = NULL; 1901 #endif /* #ifdef CONFIG_TREE_PREEMPT_RCU */ 1902 #ifdef CONFIG_RCU_BOOST 1903 p->rcu_boost_mutex = NULL; 1904 #endif /* #ifdef CONFIG_RCU_BOOST */ 1905 INIT_LIST_HEAD(&p->rcu_node_entry); 1906 } 1907 1908 #else 1909 1910 static inline void rcu_copy_process(struct task_struct *p) 1911 { 1912 } 1913 1914 #endif 1915 1916 #ifdef CONFIG_SMP 1917 extern void do_set_cpus_allowed(struct task_struct *p, 1918 const struct cpumask *new_mask); 1919 1920 extern int set_cpus_allowed_ptr(struct task_struct *p, 1921 const struct cpumask *new_mask); 1922 #else 1923 static inline void do_set_cpus_allowed(struct task_struct *p, 1924 const struct cpumask *new_mask) 1925 { 1926 } 1927 static inline int set_cpus_allowed_ptr(struct task_struct *p, 1928 const struct cpumask *new_mask) 1929 { 1930 if (!cpumask_test_cpu(0, new_mask)) 1931 return -EINVAL; 1932 return 0; 1933 } 1934 #endif 1935 1936 #ifndef CONFIG_CPUMASK_OFFSTACK 1937 static inline int set_cpus_allowed(struct task_struct *p, cpumask_t new_mask) 1938 { 1939 return set_cpus_allowed_ptr(p, &new_mask); 1940 } 1941 #endif 1942 1943 /* 1944 * Do not use outside of architecture code which knows its limitations. 1945 * 1946 * sched_clock() has no promise of monotonicity or bounded drift between 1947 * CPUs, use (which you should not) requires disabling IRQs. 1948 * 1949 * Please use one of the three interfaces below. 1950 */ 1951 extern unsigned long long notrace sched_clock(void); 1952 /* 1953 * See the comment in kernel/sched_clock.c 1954 */ 1955 extern u64 cpu_clock(int cpu); 1956 extern u64 local_clock(void); 1957 extern u64 sched_clock_cpu(int cpu); 1958 1959 1960 extern void sched_clock_init(void); 1961 1962 #ifndef CONFIG_HAVE_UNSTABLE_SCHED_CLOCK 1963 static inline void sched_clock_tick(void) 1964 { 1965 } 1966 1967 static inline void sched_clock_idle_sleep_event(void) 1968 { 1969 } 1970 1971 static inline void sched_clock_idle_wakeup_event(u64 delta_ns) 1972 { 1973 } 1974 #else 1975 /* 1976 * Architectures can set this to 1 if they have specified 1977 * CONFIG_HAVE_UNSTABLE_SCHED_CLOCK in their arch Kconfig, 1978 * but then during bootup it turns out that sched_clock() 1979 * is reliable after all: 1980 */ 1981 extern int sched_clock_stable; 1982 1983 extern void sched_clock_tick(void); 1984 extern void sched_clock_idle_sleep_event(void); 1985 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 1986 #endif 1987 1988 #ifdef CONFIG_IRQ_TIME_ACCOUNTING 1989 /* 1990 * An i/f to runtime opt-in for irq time accounting based off of sched_clock. 1991 * The reason for this explicit opt-in is not to have perf penalty with 1992 * slow sched_clocks. 1993 */ 1994 extern void enable_sched_clock_irqtime(void); 1995 extern void disable_sched_clock_irqtime(void); 1996 #else 1997 static inline void enable_sched_clock_irqtime(void) {} 1998 static inline void disable_sched_clock_irqtime(void) {} 1999 #endif 2000 2001 extern unsigned long long 2002 task_sched_runtime(struct task_struct *task); 2003 2004 /* sched_exec is called by processes performing an exec */ 2005 #ifdef CONFIG_SMP 2006 extern void sched_exec(void); 2007 #else 2008 #define sched_exec() {} 2009 #endif 2010 2011 extern void sched_clock_idle_sleep_event(void); 2012 extern void sched_clock_idle_wakeup_event(u64 delta_ns); 2013 2014 #ifdef CONFIG_HOTPLUG_CPU 2015 extern void idle_task_exit(void); 2016 #else 2017 static inline void idle_task_exit(void) {} 2018 #endif 2019 2020 #if defined(CONFIG_NO_HZ) && defined(CONFIG_SMP) 2021 extern void wake_up_idle_cpu(int cpu); 2022 #else 2023 static inline void wake_up_idle_cpu(int cpu) { } 2024 #endif 2025 2026 extern unsigned int sysctl_sched_latency; 2027 extern unsigned int sysctl_sched_min_granularity; 2028 extern unsigned int sysctl_sched_wakeup_granularity; 2029 extern unsigned int sysctl_sched_child_runs_first; 2030 2031 enum sched_tunable_scaling { 2032 SCHED_TUNABLESCALING_NONE, 2033 SCHED_TUNABLESCALING_LOG, 2034 SCHED_TUNABLESCALING_LINEAR, 2035 SCHED_TUNABLESCALING_END, 2036 }; 2037 extern enum sched_tunable_scaling sysctl_sched_tunable_scaling; 2038 2039 #ifdef CONFIG_SCHED_DEBUG 2040 extern unsigned int sysctl_sched_migration_cost; 2041 extern unsigned int sysctl_sched_nr_migrate; 2042 extern unsigned int sysctl_sched_time_avg; 2043 extern unsigned int sysctl_timer_migration; 2044 extern unsigned int sysctl_sched_shares_window; 2045 2046 int sched_proc_update_handler(struct ctl_table *table, int write, 2047 void __user *buffer, size_t *length, 2048 loff_t *ppos); 2049 #endif 2050 #ifdef CONFIG_SCHED_DEBUG 2051 static inline unsigned int get_sysctl_timer_migration(void) 2052 { 2053 return sysctl_timer_migration; 2054 } 2055 #else 2056 static inline unsigned int get_sysctl_timer_migration(void) 2057 { 2058 return 1; 2059 } 2060 #endif 2061 extern unsigned int sysctl_sched_rt_period; 2062 extern int sysctl_sched_rt_runtime; 2063 2064 int sched_rt_handler(struct ctl_table *table, int write, 2065 void __user *buffer, size_t *lenp, 2066 loff_t *ppos); 2067 2068 #ifdef CONFIG_SCHED_AUTOGROUP 2069 extern unsigned int sysctl_sched_autogroup_enabled; 2070 2071 extern void sched_autogroup_create_attach(struct task_struct *p); 2072 extern void sched_autogroup_detach(struct task_struct *p); 2073 extern void sched_autogroup_fork(struct signal_struct *sig); 2074 extern void sched_autogroup_exit(struct signal_struct *sig); 2075 #ifdef CONFIG_PROC_FS 2076 extern void proc_sched_autogroup_show_task(struct task_struct *p, struct seq_file *m); 2077 extern int proc_sched_autogroup_set_nice(struct task_struct *p, int nice); 2078 #endif 2079 #else 2080 static inline void sched_autogroup_create_attach(struct task_struct *p) { } 2081 static inline void sched_autogroup_detach(struct task_struct *p) { } 2082 static inline void sched_autogroup_fork(struct signal_struct *sig) { } 2083 static inline void sched_autogroup_exit(struct signal_struct *sig) { } 2084 #endif 2085 2086 #ifdef CONFIG_CFS_BANDWIDTH 2087 extern unsigned int sysctl_sched_cfs_bandwidth_slice; 2088 #endif 2089 2090 #ifdef CONFIG_RT_MUTEXES 2091 extern int rt_mutex_getprio(struct task_struct *p); 2092 extern void rt_mutex_setprio(struct task_struct *p, int prio); 2093 extern void rt_mutex_adjust_pi(struct task_struct *p); 2094 static inline bool tsk_is_pi_blocked(struct task_struct *tsk) 2095 { 2096 return tsk->pi_blocked_on != NULL; 2097 } 2098 #else 2099 static inline int rt_mutex_getprio(struct task_struct *p) 2100 { 2101 return p->normal_prio; 2102 } 2103 # define rt_mutex_adjust_pi(p) do { } while (0) 2104 static inline bool tsk_is_pi_blocked(struct task_struct *tsk) 2105 { 2106 return false; 2107 } 2108 #endif 2109 2110 extern bool yield_to(struct task_struct *p, bool preempt); 2111 extern void set_user_nice(struct task_struct *p, long nice); 2112 extern int task_prio(const struct task_struct *p); 2113 extern int task_nice(const struct task_struct *p); 2114 extern int can_nice(const struct task_struct *p, const int nice); 2115 extern int task_curr(const struct task_struct *p); 2116 extern int idle_cpu(int cpu); 2117 extern int sched_setscheduler(struct task_struct *, int, 2118 const struct sched_param *); 2119 extern int sched_setscheduler_nocheck(struct task_struct *, int, 2120 const struct sched_param *); 2121 extern struct task_struct *idle_task(int cpu); 2122 /** 2123 * is_idle_task - is the specified task an idle task? 2124 * @p: the task in question. 2125 */ 2126 static inline bool is_idle_task(const struct task_struct *p) 2127 { 2128 return p->pid == 0; 2129 } 2130 extern struct task_struct *curr_task(int cpu); 2131 extern void set_curr_task(int cpu, struct task_struct *p); 2132 2133 void yield(void); 2134 2135 /* 2136 * The default (Linux) execution domain. 2137 */ 2138 extern struct exec_domain default_exec_domain; 2139 2140 union thread_union { 2141 struct thread_info thread_info; 2142 unsigned long stack[THREAD_SIZE/sizeof(long)]; 2143 }; 2144 2145 #ifndef __HAVE_ARCH_KSTACK_END 2146 static inline int kstack_end(void *addr) 2147 { 2148 /* Reliable end of stack detection: 2149 * Some APM bios versions misalign the stack 2150 */ 2151 return !(((unsigned long)addr+sizeof(void*)-1) & (THREAD_SIZE-sizeof(void*))); 2152 } 2153 #endif 2154 2155 extern union thread_union init_thread_union; 2156 extern struct task_struct init_task; 2157 2158 extern struct mm_struct init_mm; 2159 2160 extern struct pid_namespace init_pid_ns; 2161 2162 /* 2163 * find a task by one of its numerical ids 2164 * 2165 * find_task_by_pid_ns(): 2166 * finds a task by its pid in the specified namespace 2167 * find_task_by_vpid(): 2168 * finds a task by its virtual pid 2169 * 2170 * see also find_vpid() etc in include/linux/pid.h 2171 */ 2172 2173 extern struct task_struct *find_task_by_vpid(pid_t nr); 2174 extern struct task_struct *find_task_by_pid_ns(pid_t nr, 2175 struct pid_namespace *ns); 2176 2177 extern void __set_special_pids(struct pid *pid); 2178 2179 /* per-UID process charging. */ 2180 extern struct user_struct * alloc_uid(struct user_namespace *, uid_t); 2181 static inline struct user_struct *get_uid(struct user_struct *u) 2182 { 2183 atomic_inc(&u->__count); 2184 return u; 2185 } 2186 extern void free_uid(struct user_struct *); 2187 extern void release_uids(struct user_namespace *ns); 2188 2189 #include <asm/current.h> 2190 2191 extern void xtime_update(unsigned long ticks); 2192 2193 extern int wake_up_state(struct task_struct *tsk, unsigned int state); 2194 extern int wake_up_process(struct task_struct *tsk); 2195 extern void wake_up_new_task(struct task_struct *tsk); 2196 #ifdef CONFIG_SMP 2197 extern void kick_process(struct task_struct *tsk); 2198 #else 2199 static inline void kick_process(struct task_struct *tsk) { } 2200 #endif 2201 extern void sched_fork(struct task_struct *p); 2202 extern void sched_dead(struct task_struct *p); 2203 2204 extern void proc_caches_init(void); 2205 extern void flush_signals(struct task_struct *); 2206 extern void __flush_signals(struct task_struct *); 2207 extern void ignore_signals(struct task_struct *); 2208 extern void flush_signal_handlers(struct task_struct *, int force_default); 2209 extern int dequeue_signal(struct task_struct *tsk, sigset_t *mask, siginfo_t *info); 2210 2211 static inline int dequeue_signal_lock(struct task_struct *tsk, sigset_t *mask, siginfo_t *info) 2212 { 2213 unsigned long flags; 2214 int ret; 2215 2216 spin_lock_irqsave(&tsk->sighand->siglock, flags); 2217 ret = dequeue_signal(tsk, mask, info); 2218 spin_unlock_irqrestore(&tsk->sighand->siglock, flags); 2219 2220 return ret; 2221 } 2222 2223 extern void block_all_signals(int (*notifier)(void *priv), void *priv, 2224 sigset_t *mask); 2225 extern void unblock_all_signals(void); 2226 extern void release_task(struct task_struct * p); 2227 extern int send_sig_info(int, struct siginfo *, struct task_struct *); 2228 extern int force_sigsegv(int, struct task_struct *); 2229 extern int force_sig_info(int, struct siginfo *, struct task_struct *); 2230 extern int __kill_pgrp_info(int sig, struct siginfo *info, struct pid *pgrp); 2231 extern int kill_pid_info(int sig, struct siginfo *info, struct pid *pid); 2232 extern int kill_pid_info_as_cred(int, struct siginfo *, struct pid *, 2233 const struct cred *, u32); 2234 extern int kill_pgrp(struct pid *pid, int sig, int priv); 2235 extern int kill_pid(struct pid *pid, int sig, int priv); 2236 extern int kill_proc_info(int, struct siginfo *, pid_t); 2237 extern __must_check bool do_notify_parent(struct task_struct *, int); 2238 extern void __wake_up_parent(struct task_struct *p, struct task_struct *parent); 2239 extern void force_sig(int, struct task_struct *); 2240 extern int send_sig(int, struct task_struct *, int); 2241 extern int zap_other_threads(struct task_struct *p); 2242 extern struct sigqueue *sigqueue_alloc(void); 2243 extern void sigqueue_free(struct sigqueue *); 2244 extern int send_sigqueue(struct sigqueue *, struct task_struct *, int group); 2245 extern int do_sigaction(int, struct k_sigaction *, struct k_sigaction *); 2246 extern int do_sigaltstack(const stack_t __user *, stack_t __user *, unsigned long); 2247 2248 static inline int kill_cad_pid(int sig, int priv) 2249 { 2250 return kill_pid(cad_pid, sig, priv); 2251 } 2252 2253 /* These can be the second arg to send_sig_info/send_group_sig_info. */ 2254 #define SEND_SIG_NOINFO ((struct siginfo *) 0) 2255 #define SEND_SIG_PRIV ((struct siginfo *) 1) 2256 #define SEND_SIG_FORCED ((struct siginfo *) 2) 2257 2258 /* 2259 * True if we are on the alternate signal stack. 2260 */ 2261 static inline int on_sig_stack(unsigned long sp) 2262 { 2263 #ifdef CONFIG_STACK_GROWSUP 2264 return sp >= current->sas_ss_sp && 2265 sp - current->sas_ss_sp < current->sas_ss_size; 2266 #else 2267 return sp > current->sas_ss_sp && 2268 sp - current->sas_ss_sp <= current->sas_ss_size; 2269 #endif 2270 } 2271 2272 static inline int sas_ss_flags(unsigned long sp) 2273 { 2274 return (current->sas_ss_size == 0 ? SS_DISABLE 2275 : on_sig_stack(sp) ? SS_ONSTACK : 0); 2276 } 2277 2278 /* 2279 * Routines for handling mm_structs 2280 */ 2281 extern struct mm_struct * mm_alloc(void); 2282 2283 /* mmdrop drops the mm and the page tables */ 2284 extern void __mmdrop(struct mm_struct *); 2285 static inline void mmdrop(struct mm_struct * mm) 2286 { 2287 if (unlikely(atomic_dec_and_test(&mm->mm_count))) 2288 __mmdrop(mm); 2289 } 2290 2291 /* mmput gets rid of the mappings and all user-space */ 2292 extern void mmput(struct mm_struct *); 2293 /* Grab a reference to a task's mm, if it is not already going away */ 2294 extern struct mm_struct *get_task_mm(struct task_struct *task); 2295 /* 2296 * Grab a reference to a task's mm, if it is not already going away 2297 * and ptrace_may_access with the mode parameter passed to it 2298 * succeeds. 2299 */ 2300 extern struct mm_struct *mm_access(struct task_struct *task, unsigned int mode); 2301 /* Remove the current tasks stale references to the old mm_struct */ 2302 extern void mm_release(struct task_struct *, struct mm_struct *); 2303 /* Allocate a new mm structure and copy contents from tsk->mm */ 2304 extern struct mm_struct *dup_mm(struct task_struct *tsk); 2305 2306 extern int copy_thread(unsigned long, unsigned long, unsigned long, 2307 struct task_struct *, struct pt_regs *); 2308 extern void flush_thread(void); 2309 extern void exit_thread(void); 2310 2311 extern void exit_files(struct task_struct *); 2312 extern void __cleanup_sighand(struct sighand_struct *); 2313 2314 extern void exit_itimers(struct signal_struct *); 2315 extern void flush_itimer_signals(void); 2316 2317 extern void do_group_exit(int); 2318 2319 extern void daemonize(const char *, ...); 2320 extern int allow_signal(int); 2321 extern int disallow_signal(int); 2322 2323 extern int do_execve(const char *, 2324 const char __user * const __user *, 2325 const char __user * const __user *, struct pt_regs *); 2326 extern long do_fork(unsigned long, unsigned long, struct pt_regs *, unsigned long, int __user *, int __user *); 2327 struct task_struct *fork_idle(int); 2328 2329 extern void set_task_comm(struct task_struct *tsk, char *from); 2330 extern char *get_task_comm(char *to, struct task_struct *tsk); 2331 2332 #ifdef CONFIG_SMP 2333 void scheduler_ipi(void); 2334 extern unsigned long wait_task_inactive(struct task_struct *, long match_state); 2335 #else 2336 static inline void scheduler_ipi(void) { } 2337 static inline unsigned long wait_task_inactive(struct task_struct *p, 2338 long match_state) 2339 { 2340 return 1; 2341 } 2342 #endif 2343 2344 #define next_task(p) \ 2345 list_entry_rcu((p)->tasks.next, struct task_struct, tasks) 2346 2347 #define for_each_process(p) \ 2348 for (p = &init_task ; (p = next_task(p)) != &init_task ; ) 2349 2350 extern bool current_is_single_threaded(void); 2351 2352 /* 2353 * Careful: do_each_thread/while_each_thread is a double loop so 2354 * 'break' will not work as expected - use goto instead. 2355 */ 2356 #define do_each_thread(g, t) \ 2357 for (g = t = &init_task ; (g = t = next_task(g)) != &init_task ; ) do 2358 2359 #define while_each_thread(g, t) \ 2360 while ((t = next_thread(t)) != g) 2361 2362 static inline int get_nr_threads(struct task_struct *tsk) 2363 { 2364 return tsk->signal->nr_threads; 2365 } 2366 2367 static inline bool thread_group_leader(struct task_struct *p) 2368 { 2369 return p->exit_signal >= 0; 2370 } 2371 2372 /* Do to the insanities of de_thread it is possible for a process 2373 * to have the pid of the thread group leader without actually being 2374 * the thread group leader. For iteration through the pids in proc 2375 * all we care about is that we have a task with the appropriate 2376 * pid, we don't actually care if we have the right task. 2377 */ 2378 static inline int has_group_leader_pid(struct task_struct *p) 2379 { 2380 return p->pid == p->tgid; 2381 } 2382 2383 static inline 2384 int same_thread_group(struct task_struct *p1, struct task_struct *p2) 2385 { 2386 return p1->tgid == p2->tgid; 2387 } 2388 2389 static inline struct task_struct *next_thread(const struct task_struct *p) 2390 { 2391 return list_entry_rcu(p->thread_group.next, 2392 struct task_struct, thread_group); 2393 } 2394 2395 static inline int thread_group_empty(struct task_struct *p) 2396 { 2397 return list_empty(&p->thread_group); 2398 } 2399 2400 #define delay_group_leader(p) \ 2401 (thread_group_leader(p) && !thread_group_empty(p)) 2402 2403 /* 2404 * Protects ->fs, ->files, ->mm, ->group_info, ->comm, keyring 2405 * subscriptions and synchronises with wait4(). Also used in procfs. Also 2406 * pins the final release of task.io_context. Also protects ->cpuset and 2407 * ->cgroup.subsys[]. And ->vfork_done. 2408 * 2409 * Nests both inside and outside of read_lock(&tasklist_lock). 2410 * It must not be nested with write_lock_irq(&tasklist_lock), 2411 * neither inside nor outside. 2412 */ 2413 static inline void task_lock(struct task_struct *p) 2414 { 2415 spin_lock(&p->alloc_lock); 2416 } 2417 2418 static inline void task_unlock(struct task_struct *p) 2419 { 2420 spin_unlock(&p->alloc_lock); 2421 } 2422 2423 extern struct sighand_struct *__lock_task_sighand(struct task_struct *tsk, 2424 unsigned long *flags); 2425 2426 static inline struct sighand_struct *lock_task_sighand(struct task_struct *tsk, 2427 unsigned long *flags) 2428 { 2429 struct sighand_struct *ret; 2430 2431 ret = __lock_task_sighand(tsk, flags); 2432 (void)__cond_lock(&tsk->sighand->siglock, ret); 2433 return ret; 2434 } 2435 2436 static inline void unlock_task_sighand(struct task_struct *tsk, 2437 unsigned long *flags) 2438 { 2439 spin_unlock_irqrestore(&tsk->sighand->siglock, *flags); 2440 } 2441 2442 #ifdef CONFIG_CGROUPS 2443 static inline void threadgroup_change_begin(struct task_struct *tsk) 2444 { 2445 down_read(&tsk->signal->group_rwsem); 2446 } 2447 static inline void threadgroup_change_end(struct task_struct *tsk) 2448 { 2449 up_read(&tsk->signal->group_rwsem); 2450 } 2451 2452 /** 2453 * threadgroup_lock - lock threadgroup 2454 * @tsk: member task of the threadgroup to lock 2455 * 2456 * Lock the threadgroup @tsk belongs to. No new task is allowed to enter 2457 * and member tasks aren't allowed to exit (as indicated by PF_EXITING) or 2458 * perform exec. This is useful for cases where the threadgroup needs to 2459 * stay stable across blockable operations. 2460 * 2461 * fork and exit paths explicitly call threadgroup_change_{begin|end}() for 2462 * synchronization. While held, no new task will be added to threadgroup 2463 * and no existing live task will have its PF_EXITING set. 2464 * 2465 * During exec, a task goes and puts its thread group through unusual 2466 * changes. After de-threading, exclusive access is assumed to resources 2467 * which are usually shared by tasks in the same group - e.g. sighand may 2468 * be replaced with a new one. Also, the exec'ing task takes over group 2469 * leader role including its pid. Exclude these changes while locked by 2470 * grabbing cred_guard_mutex which is used to synchronize exec path. 2471 */ 2472 static inline void threadgroup_lock(struct task_struct *tsk) 2473 { 2474 /* 2475 * exec uses exit for de-threading nesting group_rwsem inside 2476 * cred_guard_mutex. Grab cred_guard_mutex first. 2477 */ 2478 mutex_lock(&tsk->signal->cred_guard_mutex); 2479 down_write(&tsk->signal->group_rwsem); 2480 } 2481 2482 /** 2483 * threadgroup_unlock - unlock threadgroup 2484 * @tsk: member task of the threadgroup to unlock 2485 * 2486 * Reverse threadgroup_lock(). 2487 */ 2488 static inline void threadgroup_unlock(struct task_struct *tsk) 2489 { 2490 up_write(&tsk->signal->group_rwsem); 2491 mutex_unlock(&tsk->signal->cred_guard_mutex); 2492 } 2493 #else 2494 static inline void threadgroup_change_begin(struct task_struct *tsk) {} 2495 static inline void threadgroup_change_end(struct task_struct *tsk) {} 2496 static inline void threadgroup_lock(struct task_struct *tsk) {} 2497 static inline void threadgroup_unlock(struct task_struct *tsk) {} 2498 #endif 2499 2500 #ifndef __HAVE_THREAD_FUNCTIONS 2501 2502 #define task_thread_info(task) ((struct thread_info *)(task)->stack) 2503 #define task_stack_page(task) ((task)->stack) 2504 2505 static inline void setup_thread_stack(struct task_struct *p, struct task_struct *org) 2506 { 2507 *task_thread_info(p) = *task_thread_info(org); 2508 task_thread_info(p)->task = p; 2509 } 2510 2511 static inline unsigned long *end_of_stack(struct task_struct *p) 2512 { 2513 return (unsigned long *)(task_thread_info(p) + 1); 2514 } 2515 2516 #endif 2517 2518 static inline int object_is_on_stack(void *obj) 2519 { 2520 void *stack = task_stack_page(current); 2521 2522 return (obj >= stack) && (obj < (stack + THREAD_SIZE)); 2523 } 2524 2525 extern void thread_info_cache_init(void); 2526 2527 #ifdef CONFIG_DEBUG_STACK_USAGE 2528 static inline unsigned long stack_not_used(struct task_struct *p) 2529 { 2530 unsigned long *n = end_of_stack(p); 2531 2532 do { /* Skip over canary */ 2533 n++; 2534 } while (!*n); 2535 2536 return (unsigned long)n - (unsigned long)end_of_stack(p); 2537 } 2538 #endif 2539 2540 /* set thread flags in other task's structures 2541 * - see asm/thread_info.h for TIF_xxxx flags available 2542 */ 2543 static inline void set_tsk_thread_flag(struct task_struct *tsk, int flag) 2544 { 2545 set_ti_thread_flag(task_thread_info(tsk), flag); 2546 } 2547 2548 static inline void clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2549 { 2550 clear_ti_thread_flag(task_thread_info(tsk), flag); 2551 } 2552 2553 static inline int test_and_set_tsk_thread_flag(struct task_struct *tsk, int flag) 2554 { 2555 return test_and_set_ti_thread_flag(task_thread_info(tsk), flag); 2556 } 2557 2558 static inline int test_and_clear_tsk_thread_flag(struct task_struct *tsk, int flag) 2559 { 2560 return test_and_clear_ti_thread_flag(task_thread_info(tsk), flag); 2561 } 2562 2563 static inline int test_tsk_thread_flag(struct task_struct *tsk, int flag) 2564 { 2565 return test_ti_thread_flag(task_thread_info(tsk), flag); 2566 } 2567 2568 static inline void set_tsk_need_resched(struct task_struct *tsk) 2569 { 2570 set_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2571 } 2572 2573 static inline void clear_tsk_need_resched(struct task_struct *tsk) 2574 { 2575 clear_tsk_thread_flag(tsk,TIF_NEED_RESCHED); 2576 } 2577 2578 static inline int test_tsk_need_resched(struct task_struct *tsk) 2579 { 2580 return unlikely(test_tsk_thread_flag(tsk,TIF_NEED_RESCHED)); 2581 } 2582 2583 static inline int restart_syscall(void) 2584 { 2585 set_tsk_thread_flag(current, TIF_SIGPENDING); 2586 return -ERESTARTNOINTR; 2587 } 2588 2589 static inline int signal_pending(struct task_struct *p) 2590 { 2591 return unlikely(test_tsk_thread_flag(p,TIF_SIGPENDING)); 2592 } 2593 2594 static inline int __fatal_signal_pending(struct task_struct *p) 2595 { 2596 return unlikely(sigismember(&p->pending.signal, SIGKILL)); 2597 } 2598 2599 static inline int fatal_signal_pending(struct task_struct *p) 2600 { 2601 return signal_pending(p) && __fatal_signal_pending(p); 2602 } 2603 2604 static inline int signal_pending_state(long state, struct task_struct *p) 2605 { 2606 if (!(state & (TASK_INTERRUPTIBLE | TASK_WAKEKILL))) 2607 return 0; 2608 if (!signal_pending(p)) 2609 return 0; 2610 2611 return (state & TASK_INTERRUPTIBLE) || __fatal_signal_pending(p); 2612 } 2613 2614 static inline int need_resched(void) 2615 { 2616 return unlikely(test_thread_flag(TIF_NEED_RESCHED)); 2617 } 2618 2619 /* 2620 * cond_resched() and cond_resched_lock(): latency reduction via 2621 * explicit rescheduling in places that are safe. The return 2622 * value indicates whether a reschedule was done in fact. 2623 * cond_resched_lock() will drop the spinlock before scheduling, 2624 * cond_resched_softirq() will enable bhs before scheduling. 2625 */ 2626 extern int _cond_resched(void); 2627 2628 #define cond_resched() ({ \ 2629 __might_sleep(__FILE__, __LINE__, 0); \ 2630 _cond_resched(); \ 2631 }) 2632 2633 extern int __cond_resched_lock(spinlock_t *lock); 2634 2635 #ifdef CONFIG_PREEMPT_COUNT 2636 #define PREEMPT_LOCK_OFFSET PREEMPT_OFFSET 2637 #else 2638 #define PREEMPT_LOCK_OFFSET 0 2639 #endif 2640 2641 #define cond_resched_lock(lock) ({ \ 2642 __might_sleep(__FILE__, __LINE__, PREEMPT_LOCK_OFFSET); \ 2643 __cond_resched_lock(lock); \ 2644 }) 2645 2646 extern int __cond_resched_softirq(void); 2647 2648 #define cond_resched_softirq() ({ \ 2649 __might_sleep(__FILE__, __LINE__, SOFTIRQ_DISABLE_OFFSET); \ 2650 __cond_resched_softirq(); \ 2651 }) 2652 2653 /* 2654 * Does a critical section need to be broken due to another 2655 * task waiting?: (technically does not depend on CONFIG_PREEMPT, 2656 * but a general need for low latency) 2657 */ 2658 static inline int spin_needbreak(spinlock_t *lock) 2659 { 2660 #ifdef CONFIG_PREEMPT 2661 return spin_is_contended(lock); 2662 #else 2663 return 0; 2664 #endif 2665 } 2666 2667 /* 2668 * Thread group CPU time accounting. 2669 */ 2670 void thread_group_cputime(struct task_struct *tsk, struct task_cputime *times); 2671 void thread_group_cputimer(struct task_struct *tsk, struct task_cputime *times); 2672 2673 static inline void thread_group_cputime_init(struct signal_struct *sig) 2674 { 2675 raw_spin_lock_init(&sig->cputimer.lock); 2676 } 2677 2678 /* 2679 * Reevaluate whether the task has signals pending delivery. 2680 * Wake the task if so. 2681 * This is required every time the blocked sigset_t changes. 2682 * callers must hold sighand->siglock. 2683 */ 2684 extern void recalc_sigpending_and_wake(struct task_struct *t); 2685 extern void recalc_sigpending(void); 2686 2687 extern void signal_wake_up(struct task_struct *t, int resume_stopped); 2688 2689 /* 2690 * Wrappers for p->thread_info->cpu access. No-op on UP. 2691 */ 2692 #ifdef CONFIG_SMP 2693 2694 static inline unsigned int task_cpu(const struct task_struct *p) 2695 { 2696 return task_thread_info(p)->cpu; 2697 } 2698 2699 extern void set_task_cpu(struct task_struct *p, unsigned int cpu); 2700 2701 #else 2702 2703 static inline unsigned int task_cpu(const struct task_struct *p) 2704 { 2705 return 0; 2706 } 2707 2708 static inline void set_task_cpu(struct task_struct *p, unsigned int cpu) 2709 { 2710 } 2711 2712 #endif /* CONFIG_SMP */ 2713 2714 extern long sched_setaffinity(pid_t pid, const struct cpumask *new_mask); 2715 extern long sched_getaffinity(pid_t pid, struct cpumask *mask); 2716 2717 extern void normalize_rt_tasks(void); 2718 2719 #ifdef CONFIG_CGROUP_SCHED 2720 2721 extern struct task_group root_task_group; 2722 2723 extern struct task_group *sched_create_group(struct task_group *parent); 2724 extern void sched_destroy_group(struct task_group *tg); 2725 extern void sched_move_task(struct task_struct *tsk); 2726 #ifdef CONFIG_FAIR_GROUP_SCHED 2727 extern int sched_group_set_shares(struct task_group *tg, unsigned long shares); 2728 extern unsigned long sched_group_shares(struct task_group *tg); 2729 #endif 2730 #ifdef CONFIG_RT_GROUP_SCHED 2731 extern int sched_group_set_rt_runtime(struct task_group *tg, 2732 long rt_runtime_us); 2733 extern long sched_group_rt_runtime(struct task_group *tg); 2734 extern int sched_group_set_rt_period(struct task_group *tg, 2735 long rt_period_us); 2736 extern long sched_group_rt_period(struct task_group *tg); 2737 extern int sched_rt_can_attach(struct task_group *tg, struct task_struct *tsk); 2738 #endif 2739 #endif 2740 2741 extern int task_can_switch_user(struct user_struct *up, 2742 struct task_struct *tsk); 2743 2744 #ifdef CONFIG_TASK_XACCT 2745 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2746 { 2747 tsk->ioac.rchar += amt; 2748 } 2749 2750 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2751 { 2752 tsk->ioac.wchar += amt; 2753 } 2754 2755 static inline void inc_syscr(struct task_struct *tsk) 2756 { 2757 tsk->ioac.syscr++; 2758 } 2759 2760 static inline void inc_syscw(struct task_struct *tsk) 2761 { 2762 tsk->ioac.syscw++; 2763 } 2764 #else 2765 static inline void add_rchar(struct task_struct *tsk, ssize_t amt) 2766 { 2767 } 2768 2769 static inline void add_wchar(struct task_struct *tsk, ssize_t amt) 2770 { 2771 } 2772 2773 static inline void inc_syscr(struct task_struct *tsk) 2774 { 2775 } 2776 2777 static inline void inc_syscw(struct task_struct *tsk) 2778 { 2779 } 2780 #endif 2781 2782 #ifndef TASK_SIZE_OF 2783 #define TASK_SIZE_OF(tsk) TASK_SIZE 2784 #endif 2785 2786 #ifdef CONFIG_MM_OWNER 2787 extern void mm_update_next_owner(struct mm_struct *mm); 2788 extern void mm_init_owner(struct mm_struct *mm, struct task_struct *p); 2789 #else 2790 static inline void mm_update_next_owner(struct mm_struct *mm) 2791 { 2792 } 2793 2794 static inline void mm_init_owner(struct mm_struct *mm, struct task_struct *p) 2795 { 2796 } 2797 #endif /* CONFIG_MM_OWNER */ 2798 2799 static inline unsigned long task_rlimit(const struct task_struct *tsk, 2800 unsigned int limit) 2801 { 2802 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_cur); 2803 } 2804 2805 static inline unsigned long task_rlimit_max(const struct task_struct *tsk, 2806 unsigned int limit) 2807 { 2808 return ACCESS_ONCE(tsk->signal->rlim[limit].rlim_max); 2809 } 2810 2811 static inline unsigned long rlimit(unsigned int limit) 2812 { 2813 return task_rlimit(current, limit); 2814 } 2815 2816 static inline unsigned long rlimit_max(unsigned int limit) 2817 { 2818 return task_rlimit_max(current, limit); 2819 } 2820 2821 #endif /* __KERNEL__ */ 2822 2823 #endif 2824