[PATCH] Board enable for MSI MS-7061 (KM4AM-V)

[Mattias Mattsson]

This patch is tested on a MSI MS-7061 (KM4AM-V).

There is a very similar board, KM4M-V (they share the same vendor
BIOS). I suspect this patch works for this board also, but I haven't
tested it.

Here are info on these boards from MSI:
KM4AM-V: http://www.msi.com/index.php?func=proddesc&prod_no=594
KM4M-V: http://www.msi.com/index.php?func=proddesc&prod_no=589

I chose to match on the PCI id:s of the RAID and VGA controller as the
other devices with subsystem id:s could be disabled in the BIOS setup
(ethernet and audio).

Thanks,
-mattias

[Michael Karcher]

Am Freitag, den 13.08.2010, 07:00 +0200 schrieb Mattias Mattsson:
> I chose to match on the PCI id:s of the RAID and VGA controller as the
> other devices with subsystem id:s could be disabled in the BIOS setup
> (ethernet and audio).

This board is a typical problem board, as there seems to be no sane
device. OTOH, MSI is quite good at their subsystem IDs, in that the
subsystem ID is the four-digit board number 7061 in this case, so
matching one subsystem ID on MSI boards is good enough.

An ID that can't be turned off is the DMI string, but we prefer to have
flashrom working without DMI, too. I am really at loss what to do here.
Probably pick just one of the MSI-branded devices (to reduce the
likelyhood of hitting a disabled one) and the south bridge (with
meaningless subsystem IDs, as you correctly found out).

We have:
 - VGA - probably goes away if an AGP card is put in.
 - Audio - could be disabled in server use (onboard VGA boards might be
used in servers quite likely)
 - Network - Just 100 MBit. Might be replaced by an gigabit card in
server use
 - SATA RAID - This chip thankfully seems to have just one PCI ID
(unlike Intel where the same chip has up to three IDs for
AHCI/RAID/compatibility).

So for desktop use of that board, the VGA device is probably unreliable,
for server use audio and network. So I suggest to go with SATA/RAID and
south bridge.

For either your original PCI ID set, or my suggestion (after testing
that the match still applies), the patch is

Acked-by: Michael Karcher

Regards,
  Michael Karcher

[Mattias Mattsson]

On Fri, Aug 13, 2010 at 11:52, Michael Karcher
 wrote:
> This board is a typical problem board, as there seems to be no sane
> device. OTOH, MSI is quite good at their subsystem IDs, in that the
> subsystem ID is the four-digit board number 7061 in this case, so
> matching one subsystem ID on MSI boards is good enough.
>
> An ID that can't be turned off is the DMI string, but we prefer to have
> flashrom working without DMI, too. I am really at loss what to do here.
> Probably pick just one of the MSI-branded devices (to reduce the
> likelyhood of hitting a disabled one) and the south bridge (with
> meaningless subsystem IDs, as you correctly found out).
>
> We have:
>  - VGA - probably goes away if an AGP card is put in.
>  - Audio - could be disabled in server use (onboard VGA boards might be
> used in servers quite likely)
>  - Network - Just 100 MBit. Might be replaced by an gigabit card in
> server use
>  - SATA RAID - This chip thankfully seems to have just one PCI ID
> (unlike Intel where the same chip has up to three IDs for
> AHCI/RAID/compatibility).
>
> So for desktop use of that board, the VGA device is probably unreliable,
> for server use audio and network. So I suggest to go with SATA/RAID and
> south bridge.
>
> For either your original PCI ID set, or my suggestion (after testing
> that the match still applies), the patch is
>
> Acked-by: Michael Karcher

Thanks for your comments!

Here is the patch again. Matching on SATA/RAID (with subsystem IDs)
and south bridge (without subsystem IDs) as suggested. Tested and
works.

Signed-off-by: Mattias Mattsson

-mattias

[Michael Karcher]

Am Sonntag, den 15.08.2010, 07:57 +0200 schrieb Mattias Mattsson:
> > Acked-by: Michael Karcher
>
> Thanks for your comments!
>
> Here is the patch again. Matching on SATA/RAID (with subsystem IDs)
> and south bridge (without subsystem IDs) as suggested. Tested and
> works.
>
> Signed-off-by: Mattias Mattsson

URL added (didn't want another iteration for that) and committed in
r1139.

Thanks for your patch and testing,
  Michael Karcher