[Intel-gfx] [PATCH] drm/i915: Initialize HDMI connectors as HDMI, not DVI.

[Adam Jackson]

On Fri, 2009-09-25 at 09:58 +0800, Zhenyu Wang wrote:
> On 2009.09.24 11:16:53 -0400, Adam Jackson wrote:
> > On Thu, 2009-09-24 at 09:10 +0800, Zhenyu Wang wrote:
> > > This is correct when 'HDMI' digital port is really made as HDMI port on board,
> > > but we've seen boards which actually link up with DVI-D port. ;)
> > 
> > That's fine.  Labeling it as HDMI in RANDR will mean users will know
> > they can use a DVI->HDMI converter and be able to send audio over that
> > port.
> 
> maybe the ideal solution would be to fix drm connector type assign, which should
> be removed from connector init time, and provide new connector type setting function,
> as the real physical connection type is normally known at connector probe time, e.g HDMI/DVI,
> TV, etc.

That seems like a lot of complexity to expose to the user for no real
gain.  People know HDMI is a superset of DVI.  If the connector can send
HDMI signals, we should let people know that.

If Intel VBIOSes had a real connector table - like the rest of the
civilized world - then we could reliably report the physical connector
[1].  They don't, so the best we can do is report the signaling
capabilities honestly.  We're already calling SDVO HDMI "HDMI"; let's
call integrated HDMI "HDMI" too.

> For HDMI, we already have drm_detect_hdmi_monitor() helper.

This is a property of the sink, not of the source.  Why would you change
the source's name based on a property of the sink?

[1] - Well, no, not really, since you can never really trust physical
connector tables, the OEM is both capable of lying and has no incentive
to tell the truth.

- ajax
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://lists.freedesktop.org/archives/intel-gfx/attachments/20090925/1849338f/attachment.sig>