[RFC PATCH 2/2] arch: x86: apl: Use devicetree for FSP configuration

[Simon Glass]

Hi Bernhard,

On Thu, 7 May 2020 at 01:53, Bernhard Messerklinger
<bernhard.messerklinger at br-automation.com> wrote:
>
> Hi Simon,
>
> >Hi Bernhard,
> >
> >On Thu, 30 Apr 2020 at 03:16, Bernhard Messerklinger
> ><bernhard.messerklinger at br-automation.com> wrote:
> >>
> >> A the moment the FSP configuration is a mix of hard coded values
> >and
> >> devicetree properties.
> >> This patch makes FSP-M and FSP-S full configurable from devicetree
> >by
> >> adding binding properties for all FSP parameters.
> >>
> >> Co-developed-by: Wolfgang Wallner
> ><wolfgang.wallner at br-automation.com>
> >> Signed-off-by: Wolfgang Wallner
> ><wolfgang.wallner at br-automation.com>
> >> Signed-off-by: Bernhard Messerklinger
> ><bernhard.messerklinger at br-automation.com>
> >>
> >> ---
> >>
> >>  arch/x86/cpu/apollolake/Makefile              |    1 +
> >>  arch/x86/cpu/apollolake/fsp_bindings.c        | 2096
> >+++++++++++++++++
> >>  arch/x86/cpu/apollolake/fsp_m.c               |  164 +-
> >>  arch/x86/cpu/apollolake/fsp_s.c               |  382 +--
> >>  arch/x86/dts/chromebook_coral.dts             |   72 +-
> >>  .../asm/arch-apollolake/fsp/fsp_m_upd.h       |  168 ++
> >>  .../asm/arch-apollolake/fsp/fsp_s_upd.h       |  202 ++
> >>  .../asm/arch-apollolake/fsp_bindings.h        |   74 +
> >>  .../fsp/fsp2/apollolake/fsp-m.txt             |  320 +++
> >>  .../fsp/fsp2/apollolake/fsp-s.txt             |  483 ++++
> >>  10 files changed, 3422 insertions(+), 540 deletions(-)
> >>  create mode 100644 arch/x86/cpu/apollolake/fsp_bindings.c
> >>  create mode 100644
> >arch/x86/include/asm/arch-apollolake/fsp_bindings.h
> >>  create mode 100644
> >doc/device-tree-bindings/fsp/fsp2/apollolake/fsp-m.txt
> >>  create mode 100644
> >doc/device-tree-bindings/fsp/fsp2/apollolake/fsp-s.txt
> >>
> >
> >Tested on coral:
> >Tested-by: Simon Glass <sjg at chromium.org>
> >
> >This looks good to me. I wonder if one day the binding table could be
> >created from the binding .txt file, or compared with it
> >programmatically?
> Yes that's true. But at the moment its just copy paste.
> I generated the binding table from the fsp_s and fsp_m config struct
> with a python script. But this script is also far from being finished.

OK thanks.

> > ...
> >> +#if defined(CONFIG_SPL_BUILD)
> >
> >Do you need these #ifs? I would hope the compiler would only include
> >them if needed.
> Without the #ifs the SPL size stays the same but the u-boot proper size
> increases by about 2 kb.

OK, well then we need the #ifs.

BTW next time you send this you could take off the RFC as I think we
should apply this.

Regards,
Simon