Bugzilla – Bug 103139
Installation ends with blank screen: wrong X configuration
Last modified: 2005-08-17 09:21:20 UTC
I have an Acer Ferrari 4005 notebook. The installation correctly identified the notebook's display and graphics chipset (though it chose 1024x768 vs. the 1680x1050 that the display supports), but didn't enable output on the correct display. I had to manually add: Option "MonitorLayout" "LVDS,AUTO" .. to the "Device" section in my xorg.conf for it to work. I haven't yet tested if an external display is usable.
Please, provide all your Ferrari y2logs :) http://www.suse.de/~sh/Bugreporting-faq/index.html#id2518603
Created attachment 45401 [details] y2logs speedracer:/var/log # tar zcf y2logs.tgz YaST2/ Hopefully this will contain everything from the start of installation. Otherwise, I can reinstall to get fresh copies.
Hmm, you are using only one channel from the device but need to specify a MonitorLayout ? sounds weird to me. Matthias did you test such a notebook ? I don't have that one
I cannot read YaST2 logs. Marcus, is there any hardware information and X log files in there which I haven't noticed? I don't have an Acer as well. Jeff, can you add /var/log/Xorg.0.log and /etc/X11/xorg.conf, too?
Forgot to re-CC Marcus.
Created attachment 45597 [details] Working xorg.conf This is the working xorg.conf, with the Option "MonitorLayout" "LVDS,AUTO" line enabled. Without the MonitorLayout line - it doesn't appear to output on any display. I have two other LCDs on my desk, and hooked the DVD-D and analog d15 cables to each in order to determine just _where_ the output was going. All screens remained blank, even at a much less aggressive 1024x768.
Created attachment 45598 [details] Xorg.0.log from working xorg.conf
Created attachment 45599 [details] Xorg.0.log from broken xorg.conf Here's the log for the broken xorg.conf - quite a bit different.
Please always attach log and configuration files as text/plain. Somehow they ended up here as text/x-log and application/octet-stream, both of which is bad. Do you get any output on the internal display without "MonitorLayout" if the external monitor is *not* plugged in? You know that output device selection depends on whether a monitor is actually plugged in or not. Please note that you did not configure dualhead, thus you will never get output on both monitors (or at least it is unlikely that you will get it, and if you get it it is even more unlikely that the image will be crisp).
There were no external displays attached during boot/installation/detection. I only attached them to attempt some debugging as to where the video signal was actually going. Also, since the primary display wasn't configured correctly, I wasn't really thinking that a dual head display would work either, so I didn't try it.
In that case your test was perfect. Note that there are quite a number of (laptop internal) LCD displays that cannot display other resolutions and frequencies than their native one. Though I doubt this is the case here, when everything works with the MonitorLayout added. Did you ever try to run 9.3 or is it a brand new laptop? I *assume* that the BIOS of the laptop is broken WRT reporting plug association. I really don't understand the monitor detection output (- broken + working): (II) RADEON(0): Primary: - Monitor -- CRT + Monitor -- LVDS Connector -- VGA DAC Type -- Primary TMDS Type -- NONE DDC Type -- MONID (II) RADEON(0): Secondary: Monitor -- NONE Connector -- LVDS DAC Type -- Unknown TMDS Type -- NONE DDC Type -- MONID The card appearantly thinks it has an LVDS output as secondary output, and an CRT as primary... Additionally, the 'HorizSync 28-60' is wrong (is the internal panel a 'SEC 1024' or has this been guessed wrong), because the 1680x1050 mode is already at 64kHz. Did SaX detect the monitor correctly? Does the unmodified xorg.conf work with a correct LCD model selected manually? Right now this is wild guessing, and we don't have one of these laptops in house. But I was just told by Bodo, that we're about to get one.
I seriously doubted that the SEC 1024 LCD detection was correct, at the time I didn't look too deeply into what LCD it actually was. Acer's web site wasn't exactly forthcoming with that information. I just spent a bit of time tracking down what the LCD actually is, and I suspect that it is the Samsung LTN154P1. This is just a Samsung part number. I believe it is used in my notebook, the Acer Travelmate 8100, the Dell Inspiron 6000d, and others. I'm not sure what the "correct" selection would be since it is just an OEM part number. The native resolution of this display is WSXGA+ (1680x1050) However, this might help you, unless it's already in the y2logs: speedracer:~ # hwinfo --monitor 32: None 00.0: 10000 Monitor [Created at monitor.87] Unique ID: rdCR.S9fSVRSgyG8 Hardware Class: monitor Model: "SAMSUNG Monitor" Vendor: SEC "SAMSUNG" Device: eisa 0x0000 Size: 33x21 cm Config Status: cfg=new, avail=yes, need=no, active=unknown Also, this bit in the SaX.log was interesting to me, with respect to the CRT selection: 17-Aug 14:42:16 <I> [ Sysp: XStuff detection data Card0 => DDC : SEC0000 Card0 => Name : SAMSUNG Monitor Card0 => Vendor : SEC Card0 => Primary : 01-00-0 Card0 => Chipset : <undefined> Card0 => Vsync : 75 Card0 => Hsync : 40 Card0 => Dacspeed : 220 Card0 => Memory : 16384 Card0 => Current : 01-00-0 Card0 => RawDef : None Card0 => Option : None Card0 => Extension : None Card0 => Module : radeon Card0 => Display : CRT Card0 => Size : 33x21
That's what I would call a perfectly broken BIOS :-( Can you try configure X with sax, using a LCD 1680x1050@60Hz monitor and check whether it works w/o the MonitorLayout option? I doubt it, but it would be interesting to know.
No luck. That's not really surprising though, since the Option MonitorLayout directive appears to be for the radeon driver, independent of the display. Or at least that's my take on it.
Yes, it actually influences display detection. It was worth a try, though, to eliminate the other problem (bad scan ranges). Sorry, right now we cannot do anything for broken BIOSes. I hope that in the near future the radeon driver implements a better (heuristic) display detection routine, but I don't have the register descriptions of radeon chips (yet), so I can't do anything myself. SaX cannot supply the MonitorLayout option, too, because we currently have no way to detect the situation you're in. In fact, the situation you're in is not completely clear anyway... Closing this as WONTFIX right now, please feel free to reopen if something changes in the future. I will reopen myself when we get an Acer 4005 and I find something new.