[PATCH 03/17] acpi: x86: Move MADT to common code

[Patrick Rudolph]

Hi Simon,
On Mon, Sep 2, 2024 at 11:14 AM Patrick Rudolph
<patrick.rudolph at 9elements.com> wrote:
>
> On Mon, Jul 29, 2024 at 5:28 PM Simon Glass <sjg at chromium.org> wrote:
> >
> > Hi Patrick,
> >
> > On Sat, 27 Jul 2024 at 01:20, Patrick Rudolph
> > <patrick.rudolph at 9elements.com> wrote:
> > >
> > > Write MADT in common code and let the SoC fill out the body by
> > > calling acpi_fill_madt() which must be implemented at SoC level.
> > >
> > > Signed-off-by: Patrick Rudolph <patrick.rudolph at 9elements.com>
> > > Cc: Simon Glass <sjg at chromium.org>
> > > Cc: Bin Meng <bmeng.cn at gmail.com>
> > > ---
> > >  arch/sandbox/lib/Makefile         |  9 +++---
> > >  arch/sandbox/lib/acpi_table.c     |  7 +++++
> > >  arch/x86/cpu/intel_common/acpi.c  | 20 +++++++++-----
> > >  arch/x86/cpu/tangier/acpi.c       |  6 +++-
> > >  arch/x86/include/asm/acpi_table.h |  3 +-
> > >  arch/x86/lib/acpi_table.c         | 46 ++++++-------------------------
> > >  include/acpi/acpi_table.h         | 11 ++++++++
> > >  lib/acpi/acpi_table.c             | 31 +++++++++++++++++++++
> > >  8 files changed, 81 insertions(+), 52 deletions(-)
> > >  create mode 100644 arch/sandbox/lib/acpi_table.c
> >
> > Same comment as previous patch.
> While it might be possible to write out GICC and LAPIC or X2APIC using
> a CPU driver,
> how would the other sub-tables integrate into the acpi_ops?
>
> On x86 you have ioapic and irq_overrides and on arm GICR and ITS sub-tables.
> Would I need to add "Interrupt controller" drivers for each platform,
> fill it with platform specific data
> and then fill out the MADT table using acpi_ops?
>
> While that seems nice to have, it would be quite a lot of work, since
> none of the interface
> or drivers exists at the moment.
> It even looks like not all platforms even use a CPU driver and even
> less use an interrupt controller
> driver...
>
I've implemented this in the latest patch series v3.
See commit "drivers/arm: Implement acpi_fill_madt" that implements the
driver model
fill_madt for arm and allows to use it on QEMU SBSA. Other platforms
are still using the
"old" approach until they have been ported to the driver model.

> >
> > Regards,
> > Simon