Re: [Dri-devel] MGA obscuring bug

[Jeff H. <jha...@va...>]

Geoff Reedy wrote:

> Are any of the devlopers looking at this bug, or are they are working on
> developing the radeon driver?  If noone is working on it I'd be willing to
> try and track it down, but I don't know enough about the way DRI, DRM, and
> Mesa work together.  If someone could offer some advice on how I might go
> abou tracking this bug down I'll try and figure out what's going wrong.  And
> once again, its not a bug with glut or glu, since a program written purely
> with the GLX stuff exhibits the same behavior.
> 
This is a bug with handling the zero cliprect case correctly.  Currently 
we just discard the buffers when the window is completely obscured.  I 
think a possible fix is to render these buffers with a zero area 
cliprect (the 3Dfx driver did this at one point.)  Another fix is to get 
the DRM and the MGA card to agree on buffer aging when we discard a ton 
of buffers (This is the harder fix, and turns out to be a very complex 
case to handle with the current kernel code.)

Currently I am very busy with other things, but I do know about the 
bug.  If you want to look at the bug it would be appreciated.  I would 
suggest handling the zero cliprect case in the client side driver.  This 
should probably fix the problem.

-Jeff