From Bugzilla Helper: User-Agent: Mozilla/4.76 [en] (X11; U; Linux 2.2.16-22smp i686) The installer fails to produce a X setup that will test properly on my IBM A21m with Rage Mobility (as detected). Attempted all other Rage Mobility variants. There is no X server output on any of the VTs so I can't really say what it is bombing out on. Reproducible: Always Steps to Reproduce: 1. starting installer 2. test video config button 3. change video setup 4. wash rinse repeat
Assigning to a developer.
We (Red Hat) should really try to fix this before the next release.
The X output is in a file called /tmp/X.log. Can you send me the contents of this file?
/tmp/X-test.log: (relevant lines) (II) ATI(0): Using digital flap panel interface ... Default modes deleted, [hsync, vrefresh] out of range .. >1024x768 default modes deleted (exeeds panel dimensions) ... Built-in mode "Native panel mode" deleted (vrefresh out of range) (WW) ATI(0): Mode pool is empty ... Fatal server error: no screens found
When removed from the port replicator the system installs without trouble. Port replicator has two ports, standard VGA and DVI. It looks like the installer sees the DVI and doesn't bother with the VGA port. Even when selecting the [crt] monitor manually the X-test.log tells me that X is trying to use the DVI interface. Looks as if the installer prober is doing the same thing as the probed monitor is coming up as "flat panel 1024x768".
It looks like kudzu needs more intelligent probing routines. There isn't time to fix this for the current release, but it's a nice idea. I'll add this to our feature request list and defer it to a future release. Thanks for your report.
I hope this goes in the release notes, it really should. Sounds like (from another bugzilla bug) that someone is having a similar port replicator problem with their Dell.
I see this same problem on Dell Latitude CPx laptops (same video chipset I believe). It would be nice if the X _driver_ handled this vs. kudzu as that's how it really should happen. There is an X option for something like "external_crt" or something like that... this allows the external monitor to work, but otherwise, it will not work with the monitor. I would really love to see XFree86 (I know this isn't YOU guys directly) be able to handle hardware changes better for laptops (I use the build-in pointer and switch to scrollmouse (well, really vie versa, which screws things up). Maybe an option should be used by XFree86 to try one config then another or something.... anyway, don't know if anyone will read this, so I'll stop now ;)
I read it, and I agree. :) There's a lot of things about laptops that I wish Linux handled better...(power management, hibernation, etc.)