[Intel-gfx] DVI/HDMI regression due to 30ad48b7334a2eb2edf22f6c91f7b3f22a22a837

[Jesse Barnes]

On Tue, 16 Jun 2009 17:25:55 -0700
Jesse Barnes <jbarnes at virtuousgeek.org> wrote:

> Zhenyu's commit to add IGDNG support broke my G45 configuration.  It
> looks like my config breaks when the SDVO_NULL_PACKETS_DURING_VSYNC
> bit is set in the SDVO mode reg (I have a dual head config with HDMI
> and DVI).  Not setting the bit allows both displays to work.  Setting
> it causes a corrupted mode on my HDMI attached display and a blank
> screen on the DVI attached one.
> 
> Any ideas here Zhenyu?

Also, the connector init for HDMI should be HDMIA rather than DVID,
shouldn't it?  Our SDVO/HDMI init might still mess that up though, and
label a DVI output as HDMI...

-- 
Jesse Barnes, Intel Open Source Technology Center