Radeon 9600 XT trouble radeon.hw_i2c

[Alex Deucher]

On Sat, Jan 8, 2011 at 1:09 AM, Patrick Haller
<patrick.haller at haller-consult.com> wrote:
> Hi,
>
> I've got two ViewSonic 930 LCDs attached to the DVI-0 and VGA-0 port of a
> Sapphire Radeon 9600 XT. Using GIT master of xf86-video-ati on xorg-1.9.3 RC
> 2 on Linux kernel 2.6.37 (gentoo).
>
> Experimenting with radeon.hw_i2c=1 caused the DVI-0 LCD to go black during
> boot. To identify the cause of the problem, I exchanged the identical panels
> with the effect that the 2nd LCD also went dead on DVI-0. Restarts and
> operating system switches didn't help.
>
> Eventually, both panels kept working on VGA-0 with initial 1024x768 and
> dmesg showed that the EDIDs of both panels were corrupt.
> radeon.hw_i2c=1 seemingly has overwritten substantial parts of the the LCDs
> EDID EEPROM with 0xff. (i2cdump)
>
> To my great relief, I was able to manually (using i2cset and a VP930
> reference EDID block) restore both EDID EEPROMS and got the panels working
> fine again.
> I'd not be too fond of retrying the whole procedure in order to ensure that
> hw_i2c really was the culprit.
>
> Have I overlooked documented warning of using hw_i2c on this old board or
> has anyone else experienced such?

It worked fine for me on all cards (r1xx-r5xx) and monitors I have.

Alex

>
>
> Thanks,
> Patrick
>