Re: [SeaBIOS] Proper support for PCI-based option rom loading (was Re: [Qemu-devel] Re: qdev property bug?)

[Kevin O'Connor]

On Tue, Dec 15, 2009 at 06:41:33PM -0600, Anthony Liguori wrote:
> Kevin OConnor wrote:
>> SeaBIOS is currently using over 64K of space.  So, it is extending
>> into the e-segment and thus needs to have the e-segment copied
>> properly (I missed that in my earlier email).
>
> Okay.  So I assume this is something that SeaBIOS needs to do?  I've  
> been trying to understand the origin of:
>
> // Need to copy optionroms to work around qemu implementation
>
> What is it that we're doing wrong in the qemu implementation?

The comment is referring to the fact that enabling write access to
that area causes the existing contents to be discarded, and SeaBIOS
needs to copy the contents elsewhere, enable the write, and then copy
the contents back.  A real machine shadow implementation wouldn't
require that.  Though, I don't think it's a big deal to handle this
in SeaBIOS.

>If it's  
> the lack of PCI option loading, then I guess we're okay post my changes  
> provided that SeaBIOS can take care of what it needs in the e-segment.

Yes - I think your changes are fine.

-Kevin