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

[Roland Scheidegger]

Roland Scheidegger wrote:
> I've tried the patch with my radeon 9000pro, and it does not work there 
> neither. Issues are pretty similar, moving (3d) windows around leaves 
> artifacts (and eventually even a gpu lockup), xawtv overlay still 
> doesn't work, but on this setup I got the mouse pointer back at least.
I got the patch to work when disabling this:
* The above doesn't necessarily work. For example, I've seen machines
* with 128Mb configured as 2x64Mb apertures. I'm now _always_ setting
* RADEON_HOST_PATH_CNTL.
       OUTREGP (RADEON_HOST_PATH_CNTL, RADEON_HDP_APER_CNTL,
                  ~RADEON_HDP_APER_CNTL);
(which was previously done only on some chip families).
I _think_ this is not correct on all cards as the apertures may not be 
configured correctly (and X doesn't set them up  neither, if those 
correspond to the RADEON_CONFIG_APER registers). For the record, fglrx 
doesn't set RADEON_HDP_APER_CNTL on my card neither.
The regs on this 9000pro read as follow:

RADEON_CONFIG_APER_0_BASE 0xe8000000
RADEON_CONFIG_APER_1_BASE 0xe4000000
RADEON_CONFIG_APER_SIZE   0x04000000

I don't really know what to make of these numbers without docs, but 
assuming "straight" encoding both apertures would be 64MB apart, and the 
aperture size would be 64MB (is this per aperture or both together?). 
This card only has 64MB ram so if this HDP_APER_CNTL bit makes it use 2 
64MB apertures that would certainly be wrong.
If someone wants to tackle that I can easily provide complete reg dumps.

Roland