Re: git: 9cf5db63698b - main - acpi_ged: fix build, as module and non INTRNG case.

On Mon, Oct 24, 2022 at 10:16 AM Jessica Clarke <jrtc27@freebsd.org> wrote:
>
> On 24 Oct 2022, at 15:39, Takanori Watanabe <takawata@FreeBSD.org> wrote:
> >
> > The branch main has been updated by takawata:
> >
> > URL: https://cgit.FreeBSD.org/src/commit/?id=9cf5db63698b3c73edd632412bf68735d3c20d37
> >
> > commit 9cf5db63698b3c73edd632412bf68735d3c20d37
> > Author:     Takanori Watanabe <takawata@FreeBSD.org>
> > AuthorDate: 2022-10-24 14:19:12 +0000
> > Commit:     Takanori Watanabe <takawata@FreeBSD.org>
> > CommitDate: 2022-10-24 14:37:28 +0000
> >
> >    acpi_ged: fix build, as module and non INTRNG case.
> >
> >    Reviewed-by: cy
> >
> >    Differential Revision: https://reviews.freebsd.org/D37104
> > ---
> > sys/dev/acpica/acpi_ged.c          | 2 +-
> > sys/modules/acpi/acpi_ged/Makefile | 2 +-
> > 2 files changed, 2 insertions(+), 2 deletions(-)
> >
> > diff --git a/sys/dev/acpica/acpi_ged.c b/sys/dev/acpica/acpi_ged.c
> > index 9459ccc3525b..8ee56c8b0335 100644
> > --- a/sys/dev/acpica/acpi_ged.c
> > +++ b/sys/dev/acpica/acpi_ged.c
> > @@ -198,7 +198,7 @@ acpi_ged_attach(device_t dev)
> >                       }
> >               }
> > #else
> > -             rawirq = rman_get_start(sc->evt[i].r);
> > +             rawirq = rman_get_start(sc->evts[i].r);
> >               trig = INTR_TRIGGER_LEVEL;
> >               if (ACPI_SUCCESS(acpi_lookup_irq_resource
> >                               (dev, sc->evts[i].rid,
> > diff --git a/sys/modules/acpi/acpi_ged/Makefile b/sys/modules/acpi/acpi_ged/Makefile
> > index a937249357f4..87dd53b88b2b 100644
> > --- a/sys/modules/acpi/acpi_ged/Makefile
> > +++ b/sys/modules/acpi/acpi_ged/Makefile
> > @@ -1,7 +1,7 @@
> > #     $FreeBSD$
> >
> > .PATH:        ${SRCTOP}/sys/dev/acpica
> > -.if ${TARGET_ARCH} == aarch64
> > +.if ${MACHINE_CPUARCH} == "aarch64"
> > CFLAGS += -DINTRNG
> > .endif
> > KMOD= acpi_ged
>
> Why isn’t acpi_ged.c just #include’ing opt_global.h?
>

I suspect this is trying to cope with untied builds, though I agree
it's using the wrong approach. We should push this into
sys/conf/config.mk instead (+ arm, riscv), which should do the right
thing and actually #define it in opt_global.h (which is included via
CFLAGS).