Multihead support and preparing for 0.8.0

[Ray Strode]

Hi,

>> In order to get the monitor layout I needed to move from using /dev/fb
>> to libdrm for setting up the scan out buffers.  I think that's the
>> right long term move, but it did add a bit of complexity to the code.
>> Even though we use libdrm now, it's still all done in software.  We
>> don't do any gpu acceleration; just software blits from a shadow frame
>> buffer to the scanout buffer.
>
> What does it mean when using vesa framebuffer as a fallback (when KMS is
> not available or not working properly) ?
When it falls back to /dev/fb you get the old behavior (with wrong
looking multi-head).

--Ray