Radeon mem map fix (#2): Need regression tests please

[Simon Strandman]

Benjamin Herrenschmidt skrev:

>On Mon, 2005-12-19 at 17:07 +1100, Benjamin Herrenschmidt wrote:
>
>  
>
>>I think I've get something in shape for getting the radeon memory map
>>finally sane, removing all sort of hacks, and possibly fixing all sort
>>of weird bugs. It works best with the very latest radeon DRM fixes to
>>the same area but might well improve things all by itself
>>    
>>
>
>  .../...
>
>And here's a new rev. of the patch that fixes a couple more things.
>  
>
It seems like the patch causes some regressions. I added them to my 
unofficial gentoo xorg ebuild used by some people on the gentoo forums 
and got these responses:

"emerged rc4, now I have garbage in place of mouse cursor graphics and 
garbage in firefox 1.5
and yes, I have radeon mobility..."

"No, I had the same problem with my Radeon 7000 which worked fine before 
this latest bump. In addition to the cursor thing it actually 
hard-locked the machine after a few mins of X use but worked fine in 
console-login mode."

Disabling the patch fixed the issues in both cases. I told them to 
recompile with debug and send in their logs. I don't have any radeons 
myself to test with.

-- 
Simon Strandman <simon.strandman at telia.com>