Re: [PATCH v2 6/6] watchdog/hardlockup: Define HARDLOCKUP_DETECTOR_ARCH

[Doug Anderson]

Hi,

On Wed, Jun 21, 2023 at 6:08 AM Michael Ellerman <mpe@xxxxxxxxxxxxxx> wrote:
>
> Petr Mladek <pmladek@xxxxxxxx> writes:
> > The HAVE_ prefix means that the code could be enabled. Add another
> > variable for HAVE_HARDLOCKUP_DETECTOR_ARCH without this prefix.
> > It will be set when it should be built. It will make it compatible
> > with the other hardlockup detectors.
> >
> > The change allows to clean up dependencies of PPC_WATCHDOG
> > and HAVE_HARDLOCKUP_DETECTOR_PERF definitions for powerpc.
> >
> > As a result HAVE_HARDLOCKUP_DETECTOR_PERF has the same dependencies
> > on arm, x86, powerpc architectures.
> >
> > Signed-off-by: Petr Mladek <pmladek@xxxxxxxx>
> > Reviewed-by: Douglas Anderson <dianders@xxxxxxxxxxxx>
> > ---
> >  arch/powerpc/Kconfig | 5 ++---
> >  include/linux/nmi.h  | 2 +-
> >  lib/Kconfig.debug    | 9 +++++++++
> >  3 files changed, 12 insertions(+), 4 deletions(-)
>
> Something in this patch is breaking the powerpc g5_defconfig, I don't
> immediately see what though.
>
> ../arch/powerpc/kernel/stacktrace.c: In function ‘handle_backtrace_ipi’:
> ../arch/powerpc/kernel/stacktrace.c:171:9: error: implicit declaration of function ‘nmi_cpu_backtrace’ [-Werror=implicit-function-declaration]
>   171 |         nmi_cpu_backtrace(regs);
>       |         ^~~~~~~~~~~~~~~~~
> ../arch/powerpc/kernel/stacktrace.c: In function ‘arch_trigger_cpumask_backtrace’:
> ../arch/powerpc/kernel/stacktrace.c:226:9: error: implicit declaration of function ‘nmi_trigger_cpumask_backtrace’; did you mean ‘arch_trigger_cpumask_backtrace’? [-Werror=implicit-function-declaration]
>   226 |         nmi_trigger_cpumask_backtrace(mask, exclude_self, raise_backtrace_ipi);
>       |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>       |         arch_trigger_cpumask_backtrace
> cc1: all warnings being treated as errors

Yeah, I can reproduce that.

The problem is that before ${SUBJECT} patch "include/linux/nmi.h"
would include <asm/nmi.h>. Now it won't.

There are a ton of different ways to fix this, but I think the one
that makes sense is to be consistent with other architectures and move
the "arch_trigger_cpumask_backtrace" definitions to asm/irq.h.

https://lore.kernel.org/r/20230621164809.1.Ice67126857506712559078e7de26d32d26e64631@changeid

-Doug