Bug#243575: X configuration should allow to let out monitor frequencies and allow DDC to get them.
On Thu, Nov 11, 2004 at 07:47:17AM -0800, Daniel Stone wrote:
> On Thu, Nov 11, 2004 at 03:24:18PM +0100, Sven Luther wrote:
> > On Thu, Nov 11, 2004 at 05:23:19AM -0800, Daniel Stone wrote:
> > > amd64, i386 and powerpc have working DDC probe support (the former two
> > Nope, powerpc has not, it seems to be only some way of getting the information
> > from the openfirmware probed data, which will work if the graphic card does
> > indeed provide it, which is probably only the case on apple with the apple
> > standard cards. In particular, it fails on my pegasos, and probably also on
> > all IBM RS6k machines.
> Well, for the general case, PowerPC has EDID support. There are working
> mechanisms to probe EDID data on machines that support it, which happen
> to comprise the *vast* majority of deployed PowerPC machines. Agreed?
As we are going to deploy an amount of pegasos machines running debian in the
thousands nextly, i would dispute this *vast* majority claim. Also, i believe
that read-edid is not built for ppc right now, or at least not in sarge. As
debian-installer complains about not being able to install it.
> > > via VESA's VBE, the latter via /proc), and there are very, very, very,
> > > very, very, very, very few monitors out there that people actually use
> > > that don't have working DDC/EDID. Trust me, I've been there, and found
> > > out.
> > He, i have encountered this case at least 5 times this past month. Ranging
> > from cheap monitors, which altough they claim to do DDC/EDID, they only send
> > bogus data to the card, one case was even missing the Ranges fields, or
> > age-old big sun monitors, or even people behind cheap KVM switches, which
> > don't relay the DDC pins. All those i have encountered.
> The ranges field is not a problem. KVMs are a problem, and this is a
Err, how do you detect the allowed frequencies without Ranges entry ?
> problem. You can blacklist known bad resolutions (IIRC I have
> blacklisted about four), or compare them against a whitelist of known
> good resolutions. There are many good ways to get around this.
Yep, and many bad ones probably too.
> It won't work exhaustively correctly in every case, but it's really an
> exceptionally sensible default.
Well, my main grip is that, when the user of the box is installing it, and
wants to let the field empty, he should be allowed to do it, since there are
good chances that he knows what to do. Also defaulting to empty modelines for
flat panels is a sensible default, since most flat panels would have working
DDC support, and giving them default CRT frequency ranges is not a good idea.
> > Now, X clearly knows how to get the DDC info on most plateforms, should it not
> > be easiest to have some kind of X based tool which probes the DDC info, and
> > reports back if it worked. Hell i guess running X in -probeonly, and parsing
> > the XFree86.0.log file should be a cheap way of doing that.
> Cheap but crap, yes.
Do you see a better way which would allow yo do real DDC/EDID probing without
relaying on the bios to do it ?