[PATCH] Add Mitac 6513WU mainboard support

[Michael Gold]

This patch adds support for the Mitac 6513WU mainboard, a Compaq OEM
board using the i810 chipset.  Not all hardware has been tested, but my
test PC boots Linux (via FILO) without any problems.

The configuration leaves 32 kB for the video BIOS, but the onboard video
isn't currently working due to known problems with this chipset.

Signed-off-by: Michael Gold 
---
(At Uwe's suggestion, I've based this on r4393, but hopefully it will
still apply.)

[Uwe Hermann]

On Sat, Jul 04, 2009 at 05:36:20PM -0400, Michael Gold wrote:
> This patch adds support for the Mitac 6513WU mainboard, a Compaq OEM
> board using the i810 chipset.  Not all hardware has been tested, but my
> test PC boots Linux (via FILO) without any problems.
>
> The configuration leaves 32 kB for the video BIOS, but the onboard video
> isn't currently working due to known problems with this chipset.
>
> Signed-off-by: Michael Gold

Great stuff, thanks a lot! This is committed in r4401 with some minor
changes.

I dropped all "(C) Uwe Hermann", pretty much all these files are trivial
and/or generic, no need to carry over any (C) lines.
> device pci 1.0 on                   # Chipset Graphics Controller (CGC)
>  chip drivers/pci/onboard
>    device pci 1.0 on end
>    register "rom_address" = "0xfff80000" #512k image
>  end
> end

I changed this part a bit, this may be nested a bit too much, but maybe
I'm wrong, not sure. Either way, the version which is in svn now worked
for me on the MS-6178 board, please try a plain svn checkout (r4401) and
see if onboard VGA works (better) on your board.
> +romimage "normal"
> option CONFIG_USE_FALLBACK_IMAGE = 0
> option COREBOOT_EXTRA_VERSION = ".0Normal"
> payload ../../../../payloads/filo.elf

Changed this to the generic "../payload.elf" we try to use
everywhere now (but there may be more changes wrt to this
when we move to kconfig for v2).

Uwe.

[Michael Gold]

This patch enables onboard VGA support for the Mitac 6513WU mainboard by
defining CONFIG_VIDEO_MB.

Signed-off-by: Michael Gold 
---
(additional comments below)

On Sun, Jul 05, 2009 at 21:39:32 +0200, Uwe Hermann wrote:
> > device pci 1.0 on                   # Chipset Graphics Controller (CGC)
> >  chip drivers/pci/onboard
> >    device pci 1.0 on end
> >    register "rom_address" = "0xfff80000" #512k image
> >  end
> > end
>
> I changed this part a bit, this may be nested a bit too much, but maybe
> I'm wrong, not sure. Either way, the version which is in svn now worked
> for me on the MS-6178 board, please try a plain svn checkout (r4401) and
> see if onboard VGA works (better) on your board.

The nesting seemed strange to me as well, but the example on the "VGA
support" wiki page is written like that.

After defining CONFIG_VIDEO_MB, the VGA device appears in lspci.
I haven't tried hooking up a monitor.

I noticed that you moved the subtraction of 32 KB from CONFIG_ROM_SIZE
(for the video BIOS) from the mainboard's Options.lb to the target's
Config.lb, although the rom_address for this space in still listed in
the mainboard's Config.lb.  Why was it changed?  The division between
these files doesn't appear to be well documented, and is inconsistent in
existing targets.
> > +romimage "normal"
> > option CONFIG_USE_FALLBACK_IMAGE = 0
> > option COREBOOT_EXTRA_VERSION = ".0Normal"
> > payload ../../../../payloads/filo.elf
>
> Changed this to the generic "../payload.elf" we try to use
> everywhere now (but there may be more changes wrt to this
> when we move to kconfig for v2).

That's fine, but it would be good to document the recommended practice
since it's not done consistently in the existing code.

-- Michael

[ron minnich]

can you test with the committed r4401? I'd like to get an idea of
whether this board works with the new resource allocator.

ron

[Uwe Hermann]

On Sun, Jul 05, 2009 at 07:14:32PM -0400, Michael Gold wrote:
> This patch enables onboard VGA support for the Mitac 6513WU mainboard by
> defining CONFIG_VIDEO_MB.
>
> Signed-off-by: Michael Gold

Thanks, r4402.
> I noticed that you moved the subtraction of 32 KB from CONFIG_ROM_SIZE
> (for the video BIOS) from the mainboard's Options.lb to the target's
> Config.lb, although the rom_address for this space in still listed in
> the mainboard's Config.lb.  Why was it changed?  The division between
> these files doesn't appear to be well documented, and is inconsistent in
> existing targets.

Yeah, true. We're not really consistent with most of the config option
stuff. In general, targets/.../Config.lb overrides src/.../Options.lb,
but which options are set in which file is inconsistent. IMHO most stuff
that a user might want to change (ROM_SIZE, VGA or not, etc) should be
in targets/.../Config.lb.

This might all change soonish when we use kconfig for building v2, so
it's probably not worth messing with it too much right now.
> > Changed this to the generic "../payload.elf" we try to use
> > everywhere now (but there may be more changes wrt to this
> > when we move to kconfig for v2).
>
> That's fine, but it would be good to document the recommended practice
> since it's not done consistently in the existing code.

Yeah, also true. We could mass-patch all targets to use ../payload.elf
for consistency, but this will also be obsolete should we use kconfig
soon, so not sure if we want to do it right now.

I've set up a page for your board in the wiki:
http://www.coreboot.org/Mitac_6513WU
See http://www.coreboot.org/MSI_MS-6178 for an example page with status etc.

Feel free to update the status, either per email here and/or you request
a wiki account (email Stefan Reinauer ).
We can also upload one or two photos of the board if you can create some
and are willing to put them into the public domain (or use CC or GPL).

Uwe.

[Michael Gold]

On Mon, Jul 06, 2009 at 07:56:36 -0700, ron minnich wrote:
> can you test with the committed r4401? I'd like to get an idea of
> whether this board works with the new resource allocator.

I've already tested that and didn't see any problems.  The bootlog is
attached.

-- Michael