Red Hat Bugzilla – Bug 109542
graphical installer uses wrong X settings on laptop external monitor
Last modified: 2007-11-30 17:10:33 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (compatible; Konqueror/3.1; Linux)
Description of problem:
On my Dell Inspiron 8100 laptop with external monitor, graphical
install fails if the external monitor is enabled.
Anaconda detects my video card (NVidia GeForce 2 Go) and external
monitor (Dell Ultrascan P991, a rebranded trinitron) correctly. I am
guessing that it is maybe trying to use the X settings for the lcd
on the external montior.
I can toggle the display from external to lcd and back with a key on
the keyboard. When I enable the lcd, I get static there as well. If
I switch to a virtual console on the lcd, it is garbled but
recognizable as text. If I then switch back into X, it is fine. If I
switch back to the console again, I no longer get garbled text, but
moving static instead. Switching the external monitor back on at any
point results in an on-screen display reporting "out of scan range"
There are no problems if I install with the external monitor turned
off the whole time.
This bug prevents install from proceeding, and would prevent a user
who doesn't know how to kill X from performing a soft reboot.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Do graphical install with external monitor turned on
Actual Results: Static on external monitor
I haven't been able to retrieve the X logs. Maybe I can if I switch
to the lcd and kill X and then hit control-s during shutdown, but I
haven't tried that yet.
Can you attach the /tmp/XF86Config and /tmp/anaconda.log from having
both the external monitor attached and not? You shouldn't have to
actually complete the install, just get to where X has started and you
can access tty2.
Created attachment 95891 [details]
logfiles for external and builtin cases
I'm uploading a zipped tarball with subdirectories called crt and lcd. Each
contains X.log, XF86Config.test, and anaconda.log. Crt is the external monitor
(broken X config) and lcd is the builtin display (working). Let me know if you
want these files uploaded some other way.
I diffed the crt and lcd directories and the only difference is a
timestamp in X.log.
LCD is without the external monitor attached/powered on? I don't see
how it could possibly probe a CRT without it actually being there...
You're right, I probably just had it toggled off from the laptop
keyboard. This causes the external monitor to go to sleep, but it
might still be probed. When I get a chance I'll try it with the
monitor powered off.
I noticed a similar but slightly different problem on my FC2-test1
The machine in question has an onboard SiS video card, and an addition
NVidia card installed. The NVidia is set as the primary video and is
the only one used ( always had issues with the SiS in the past ).
Anaconda detected the SiS card, and went on to use this for the X11
install which failed, doing a text install worked, but X11 was setup
for XF86Config, and thus failed again. Manually editing the
XF86Config for nvidia use allowed me to actually use the machine thou.
If this is different, should I open another issue for it?
This is all done differently in our current code, so I'm going to hope
that it's no longer a problem. If you continue to experience this in
Fedora Core 3, please reopen or file a new report with details of what
you see there.