xf86-video-intel: unexpected phenomenon on XV texture adapter with no scaling

[Roland Scheidegger]

On 22.06.2009 19:11, Krzysztof Halasa wrote:
> Roland Scheidegger <sroland at tungstengraphics.com> writes:
> 
>> With a very quick look at the r600 code, I suggest trying out the
>> attached patch to test my theory about half pixel offsets in hardware.
>> This could mess though with EXA acceleration, so if you see a bit odd
>> corruption don't be surprised :-).
> 
> I can now confirm this patch, applied to current Fedora 11
> xorg-x11-drv-ati-6.12.2-14 driver, fixes this Xv issue on RV6xx.
> 
> I can't see any EXA corruption.
> 
> Thanks a lot.
> 
> I guess it should be aplied to the repository?
I think it's probably the right thing to do, but I'm a bit worried there
might be side effects in other parts (composite), though it's possible
it would be actually more correct there too. I've never really looked
into the r6xx side of things in the driver, so I'd like some opinion
from someone more familiar with that code. Alex?

Roland

> 
>> --- a/src/r6xx_accel.c
>> +++ b/src/r6xx_accel.c
>> @@ -974,7 +974,7 @@ set_default_state(ScrnInfoPtr pScrn, drmBufPtr ib)
>>      EREG(ib, PA_SU_POLY_OFFSET_FRONT_OFFSET,      0);
>>  
>>      EREG(ib, PA_SU_LINE_CNTL,                     (8 << PA_SU_LINE_CNTL__WIDTH_shift)); /* Line width 1 pixel */
>> -    EREG(ib, PA_SU_VTX_CNTL,                      ((2 << PA_SU_VTX_CNTL__ROUND_MODE_shift) |
>> +    EREG(ib, PA_SU_VTX_CNTL,                      ((2 << PA_SU_VTX_CNTL__ROUND_MODE_shift) | PIX_CENTER_bit |
>>  						   (5 << QUANT_MODE_shift))); /* Round to Even, fixed point 1/256 */
>>      EREG(ib, PA_SU_POLY_OFFSET_CLAMP,             0);
>>  
>