kms howto - is there one?

Felix Miata mrmazda at earthlink.net
Fri Mar 19 12:09:25 UTC 2010


On 2010/03/19 03:12 (GMT-0400) Adam Williamson composed:

> On Fri, 2010-03-19 at 02:49 -0400, Felix Miata wrote:

>> FWIW, I do now have the 845G with F13 running 2048x1536, but the amount of
>> work required to set it up is just nuts compared to how automatic it was
>> before auto configuration development "progressed" up to now. The xorg.conf
>> and Xorg.0.log resulting:
>> http://fm.no-ip.com/tmp/Linux/Xorg/xorg.conf.08-t2240-i845G-f13-2048x1536x144x24bpp-withxorgconf
>> http://fm.no-ip.com/tmp/Linux/Xorg/xorg.0.log-t2240-i845G-f13-2048x1536x144x24bpp-withxorgconfgmodelines

> Erm. There's exactly three lines in there that are actually necessary
> for what you're doing:

> Modeline "2048x1536_60.00"  267.25  2048 2208 2424 2800  1536 1539 1543 1592 -hsync +vsync
> HorizSync    30-107
> Modes      "2048x1536" "1920x1440" "1856x1392" "1792x1344" "1600x1200" "1400x1050" "1280x960" "1152x864" "1024x768"

> The rest are commented out or irrelevant.

As far as I am concerned, commented lines do not exist except at such time I
wish to perform a spec switch, which due to CRS disease would be needlessly
difficult in their absence. Adding or removing a "#" is much easier than
remembering syntax or why. :-)

In current F13 (unless using mga), modes without modelines in xorg.conf and
not autoprobed are missing from the krandrtray mode select list.

> In fact, even the Modes line may not be necessary, looking at the X log.

Without both modes and modelines in xorg.conf, only the anachronistic
1024x768 and worse are available modes. Xorg.0.log lies. The "probed" modes
above 1024x768 are not there unless present in xorg.conf, unless I'm using
mga, or Factory, Knoppix, Lucid, Cooker or previous Fedora releases.

> (Although I note "VertRefresh  56-61" - don't you get headaches?!)

The only headaches I get are from having to play with the device's physical
controls every time another driver comes up with yet another set of
auto-generated mode specifications that haven't been entered in the device's
memory, resulting in errant sizing and/or positioning of the output. By
limiting refresh to that narrow range the auto-generation usually matches a
generic spec that's already in the device's memory, resulting in correct
centering and sizing of the output.

To be clear, my displays get subjected to many different gfxcards, and many
different operating system brands (DOS, Windows & OS/2, in addition to the
previously mentioned major distros) and versions, from a bunch of different
computers, pretty much at random. There's no way any of my displays has
enough memory for every mode it's ever been subjected to. This is good,
because it tests the desktop environment's ability to cope with a variety of
hardware, quite unlike modern LCDs with their working DDC/EDID virtually
forcing their use only with a single native mode.

> This would not be a problem on any monitor which correctly communicates
> its correct HorizSync range to the system. In the case where the monitor
> doesn't have functioning EDID or DDC, *every* choice X can make is
> 'wrong' in some sense, so it doesn't make much sense to complain about
> the particular wrongness of the particular choice we go with.

Again I think you're missing the major point that this is _new_ behavior.
Only in F13 (so far, using intel & radeon at least, but not when using mga)
does what worked previously work no longer. Until now, and for those for whom
lowfi resolution is not acceptable, explicit modelines in xorg.conf hadn't
been necessary with broken/missing DDC/EDID for many many moons.

And BTW, actually switching modes without restarting X must be done by typing
xrandr commands. Attempting to select another mode from krandrtray corrupts
everything such that CAD or the reset button (or I suppose remote login) are
required to recover.
-- 
"The wise are known for their understanding, and pleasant
words are persuasive." Proverbs 16:21 (New Living Translation)

 Team OS/2 ** Reg. Linux User #211409

Felix Miata  ***  http://fm.no-ip.com/


More information about the test mailing list