[paulmck-rcu:context-tracking.2022.05.19a 20/21] kernel/context_tracking.c:500:25: error: implicit declaration of function 'ct_kernel_exit'; did you mean 'is_kernel_text'?

From: kernel test robot
Date: Thu May 19 2022 - 17:15:44 EST

Next message: kernel test robot: "mm/page_alloc.c:984:9: warning: iteration 119304648 invokes undefined behavior"
Previous message: Matthew Wilcox: "Re: [PATCH v2 03/28] vsprintf: %pf(%p)"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

tree: https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git context-tracking.2022.05.19a
head: 6153dca9e37b244684bf472728f5c3688c5b0aca
commit: 02be5e537043004b7c133565ae473daf19a99e5d [20/21] rcu/context_tracking: Merge dynticks counter and context tracking states
config: arm-randconfig-r012-20220519 (https://download.01.org/0day-ci/archive/20220520/202205200530.nX4ZzOAH-lkp@xxxxxxxxx/config)
compiler: arm-linux-gnueabi-gcc (GCC) 11.3.0
reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross
chmod +x ~/bin/make.cross
# https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git/commit/?id=02be5e537043004b7c133565ae473daf19a99e5d
git remote add paulmck-rcu https://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu.git
git fetch --no-tags paulmck-rcu context-tracking.2022.05.19a
git checkout 02be5e537043004b7c133565ae473daf19a99e5d
# save the config file
mkdir build_dir && cp config build_dir/.config
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-11.3.0 make.cross W=1 O=build_dir ARCH=arm SHELL=/bin/bash

If you fix the issue, kindly add following tag as appropriate
Reported-by: kernel test robot <lkp@xxxxxxxxx>

All errors (new ones prefixed by >>):

kernel/context_tracking.c: In function '__ct_user_enter':
>> kernel/context_tracking.c:500:25: error: implicit declaration of function 'ct_kernel_exit'; did you mean 'is_kernel_text'? [-Werror=implicit-function-declaration]
500 | ct_kernel_exit(true, RCU_DYNTICKS_IDX + state);
| ^~~~~~~~~~~~~~
| is_kernel_text
kernel/context_tracking.c: In function '__ct_user_exit':
>> kernel/context_tracking.c:607:25: error: implicit declaration of function 'ct_kernel_enter'; did you mean 'ct_idle_enter'? [-Werror=implicit-function-declaration]
607 | ct_kernel_enter(true, RCU_DYNTICKS_IDX - state);
| ^~~~~~~~~~~~~~~
| ct_idle_enter
cc1: some warnings being treated as errors

vim +500 kernel/context_tracking.c

451
452 /**
453 * __ct_user_enter - Inform the context tracking that the CPU is going
454 * to enter user or guest space mode.
455 *
456 * This function must be called right before we switch from the kernel
457 * to user or guest space, when it's guaranteed the remaining kernel
458 * instructions to execute won't use any RCU read side critical section
459 * because this function sets RCU in extended quiescent state.
460 */
461 void noinstr __ct_user_enter(enum ctx_state state)
462 {
463 struct context_tracking *ct = this_cpu_ptr(&context_tracking);
464 lockdep_assert_irqs_disabled();
465
466 /* Kernel threads aren't supposed to go to userspace */
467 WARN_ON_ONCE(!current->mm);
468
469 if (!context_tracking_recursion_enter())
470 return;
471
472 if (__ct_state() != state) {
473 if (ct->active) {
474 /*
475 * At this stage, only low level arch entry code remains and
476 * then we'll run in userspace. We can assume there won't be
477 * any RCU read-side critical section until the next call to
478 * user_exit() or ct_irq_enter(). Let's remove RCU's dependency
479 * on the tick.
480 */
481 if (state == CONTEXT_USER) {
482 instrumentation_begin();
483 trace_user_enter(0);
484 vtime_user_enter(current);
485 instrumentation_end();
486 }
487 /*
488 * Other than generic entry implementation, we may be past the last
489 * rescheduling opportunity in the entry code. Trigger a self IPI
490 * that will fire and reschedule once we resume in user/guest mode.
491 */
492 rcu_irq_work_resched();
493
494 /*
495 * Enter RCU idle mode right before resuming userspace. No use of RCU
496 * is permitted between this call and rcu_eqs_exit(). This way the
497 * CPU doesn't need to maintain the tick for RCU maintenance purposes
498 * when the CPU runs in userspace.
499 */
> 500 ct_kernel_exit(true, RCU_DYNTICKS_IDX + state);
501
502 /*
503 * Special case if we only track user <-> kernel transitions for tickless
504 * cputime accounting but we don't support RCU extended quiescent state.
505 * In this we case we don't care about any concurrency/ordering.
506 */
507 if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE))
508 atomic_set(&ct->state, state);
509 } else {
510 /*
511 * Even if context tracking is disabled on this CPU, because it's outside
512 * the full dynticks mask for example, we still have to keep track of the
513 * context transitions and states to prevent inconsistency on those of
514 * other CPUs.
515 * If a task triggers an exception in userspace, sleep on the exception
516 * handler and then migrate to another CPU, that new CPU must know where
517 * the exception returns by the time we call exception_exit().
518 * This information can only be provided by the previous CPU when it called
519 * exception_enter().
520 * OTOH we can spare the calls to vtime and RCU when context_tracking.active
521 * is false because we know that CPU is not tickless.
522 */
523 if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) {
524 /* Tracking for vtime only, no concurrent RCU EQS accounting */
525 atomic_set(&ct->state, state);
526 } else {
527 /*
528 * Tracking for vtime and RCU EQS. Make sure we don't race
529 * with NMIs. OTOH we don't care about ordering here since
530 * RCU only requires RCU_DYNTICKS_IDX increments to be fully
531 * ordered.
532 */
533 atomic_add(state, &ct->state);
534 }
535 }
536 }
537 context_tracking_recursion_exit();
538 }
539 EXPORT_SYMBOL_GPL(__ct_user_enter);
540
541 /*
542 * FIXME: This function should be noinstr but the below local_irq_restore() is
543 * unsafe because it involves illegal RCU uses through tracing and lockdep. This
544 * must be fixed first.
545 */
546 void ct_user_enter(enum ctx_state state)
547 {
548 unsigned long flags;
549
550 /*
551 * Some contexts may involve an exception occuring in an irq,
552 * leading to that nesting:
553 * ct_irq_enter() rcu_eqs_exit(true) rcu_eqs_enter(true) ct_irq_exit()
554 * This would mess up the dyntick_nesting count though. And rcu_irq_*()
555 * helpers are enough to protect RCU uses inside the exception. So
556 * just return immediately if we detect we are in an IRQ.
557 */
558 if (in_interrupt())
559 return;
560
561 local_irq_save(flags);
562 __ct_user_enter(state);
563 local_irq_restore(flags);
564 }
565 NOKPROBE_SYMBOL(ct_user_enter);
566 EXPORT_SYMBOL_GPL(ct_user_enter);
567
568 /**
569 * user_enter_callable() - Unfortunate ASM callable version of user_enter() for
570 * archs that didn't manage to check the context tracking
571 * static key from low level code.
572 *
573 * FIXME: This function should be noinstr but it unsafely calls local_irq_restore(),
574 * involving illegal RCU uses through tracing and lockdep. This must be fixed first.
575 */
576 void user_enter_callable(void)
577 {
578 user_enter();
579 }
580 NOKPROBE_SYMBOL(user_enter_callable);
581
582 /**
583 * __ct_user_exit - Inform the context tracking that the CPU is
584 * exiting user or guest mode and entering the kernel.
585 *
586 * This function must be called after we entered the kernel from user or
587 * guest space before any use of RCU read side critical section. This
588 * potentially include any high level kernel code like syscalls, exceptions,
589 * signal handling, etc...
590 *
591 * This call supports re-entrancy. This way it can be called from any exception
592 * handler without needing to know if we came from userspace or not.
593 */
594 void noinstr __ct_user_exit(enum ctx_state state)
595 {
596 struct context_tracking *ct = this_cpu_ptr(&context_tracking);
597
598 if (!context_tracking_recursion_enter())
599 return;
600
601 if (__ct_state() == state) {
602 if (ct->active) {
603 /*
604 * Exit RCU idle mode while entering the kernel because it can
605 * run a RCU read side critical section anytime.
606 */
> 607 ct_kernel_enter(true, RCU_DYNTICKS_IDX - state);
608 if (state == CONTEXT_USER) {
609 instrumentation_begin();
610 vtime_user_exit(current);
611 trace_user_exit(0);
612 instrumentation_end();
613 }
614
615 /*
616 * Special case if we only track user <-> kernel transitions for tickless
617 * cputime accounting but we don't support RCU extended quiescent state.
618 * In this we case we don't care about any concurrency/ordering.
619 */
620 if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE))
621 atomic_set(&ct->state, CONTEXT_KERNEL);
622
623 } else {
624 if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) {
625 /* Tracking for vtime only, no concurrent RCU EQS accounting */
626 atomic_set(&ct->state, CONTEXT_KERNEL);
627 } else {
628 /*
629 * Tracking for vtime and RCU EQS. Make sure we don't race
630 * with NMIs. OTOH we don't care about ordering here since
631 * RCU only requires RCU_DYNTICKS_IDX increments to be fully
632 * ordered.
633 */
634 atomic_sub(state, &ct->state);
635 }
636 }
637 }
638 context_tracking_recursion_exit();
639 }
640 EXPORT_SYMBOL_GPL(__ct_user_exit);
641

--
0-DAY CI Kernel Test Service
https://01.org/lkp

Next message: kernel test robot: "mm/page_alloc.c:984:9: warning: iteration 119304648 invokes undefined behavior"
Previous message: Matthew Wilcox: "Re: [PATCH v2 03/28] vsprintf: %pf(%p)"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]