Summary: | radeon MergedFB stuck in "clone" mode | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Product: | xorg | Reporter: | Joe Ammond <joe.ammond> | ||||||||
Component: | Driver/Radeon | Assignee: | Benjamin Herrenschmidt <benh> | ||||||||
Status: | RESOLVED FIXED | QA Contact: | |||||||||
Severity: | major | ||||||||||
Priority: | high | CC: | alexdeucher, benh, hhgarnes, jag, mitch, stian_web | ||||||||
Version: | 6.99.99.904 (7.0 RC4) | ||||||||||
Hardware: | x86 (IA32) | ||||||||||
OS: | Linux (All) | ||||||||||
Whiteboard: | |||||||||||
i915 platform: | i915 features: | ||||||||||
Attachments: |
|
Description
Joe Ammond
2006-01-11 02:10:20 UTC
Created attachment 4313 [details]
xorg.conf file
xorg.conf file. This configuration worked under 6.8.2, and no longer does on
7.0 RC4.
Created attachment 4314 [details]
Logfile with external monitor attached
Logfile with the external monitor attached.
Created attachment 4315 [details]
Picture of laptop and external monitor
This is a picture taken with my camera of the laptop LCD and the external
monitor. The GDM screen shows on both displays in "clone" mode.
I'm having exactly the same problem with my radeon 9200 card. Seems to be a regression in the 7.0 drivers. I posted my query to the alias - no response. Thanks, I got a similar report from a PowerMac user, I'll investigate, I may have hw to reproduce. Please kick me if you don't hear from me by next week :) assign to me I'll look into this as well as soon as I move to 7.0/modular on my radeon box. What's odd is I'm been using pre-modular cvs (aka 6.9ish) with mergedfb with no problems for ages. I saw the same problem when I added an older version of the radeon memory map patch to debian unstable's xorg 6.9.0.dfsg.1-4, so maybe this has something to do with the memory map fixes. I recently loaded OpenSUSE 10.1 beta 6 on this laptop, and the problem is in that distribution as well. The server info from that is: X Window System Version 6.9.0 Release Date: 21 December 2005 X Protocol Version 11, Revision 0, Release 6.9 Build Operating System: SuSE Linux [ELF] SuSE Current Operating System: Linux bauhaus 2.6.16-rc5-git2-2-default #1 Tue Feb 28 09:16:17 UTC 2006 i686 Build Date: 26 February 2006 I use MergedFB on my T42 (Radeon Mobility 7500). It worked with X.org 6.8.2, it works with 7.0 (Ubuntu Breezy and now Dapper). It worked fine with 6.8.2, but with 7.0 I see some weirdness: * When I start gdm, I get a large virtual desktop, but both monitors show the left part of it (I know the right part exists because I can move my mouse deep into it). The external monitor (a LCD connected to the VGA port) is all wobbly and wavy. * When I log in, I get 1024x768 clone mode. The external monitor is wobbly and wavy. * I can use xrandr to get 1024x768-1280x1024 or 1024x768-1024x768. The external monitor displays a sharp picture. * I can use xrandr to switch back to 1024x768 clone mode. The external monitor displays a sharp picture. * If I use Ctrl+Alt+F1 to switch into a text console and then Ctrl+Alt+F7 to switch back, while in a 1024x768-1280x1024 screen, I once again see the left part of the large desktop on both monitors. The external monitor is wobbly and wavy. I have to use xrandr to switch to some other mode and then switch back again to get correct picture. My xorg.conf is available at http://mg.pov.lt/xorg.conf (BTW when I used 6.8.2, I used MergedFBLayout (1024x768-1024x768) from my config file; now I switched to MergedFB2Layout (1024x768-1280x1024), because now I can use MergeNonRectangular.) After reading the comments from Marius, I started looking at xrandr. I've discovered a couple of things: 1) If I leave out a 'MetaModes' line, then I see two resolutions in xrandr: 2800x1050 and 1400x1050. Upon login, it shows 2800x1050 with the display cloned. 2) I can switch to 1400x1050 and the display stays cloned. 3) Switching back to 2800x1050 turns off the clone of the display and gives the MergedFB I had under 6.8.2. If I switch to a text VT the back to X, I'm back at step 1 above. The display thinks it is 2800x1050 but the screens are cloned. (In reply to comment #11) > After reading the comments from Marius, I started looking at xrandr. I've > discovered a couple of things: > > 1) If I leave out a 'MetaModes' line, then I see two resolutions in xrandr: > 2800x1050 and 1400x1050. Upon login, it shows 2800x1050 with the display cloned. > > 2) I can switch to 1400x1050 and the display stays cloned. > > 3) Switching back to 2800x1050 turns off the clone of the display and gives the > MergedFB I had under 6.8.2. > > If I switch to a text VT the back to X, I'm back at step 1 above. The display > thinks it is 2800x1050 but the screens are cloned. Sounds like we are missing the call to adjustframe() during serverinit and enterVT. I assume Crtl-Alt-+/- behaves the same as xrandr? (In reply to comment #12) > (In reply to comment #11) > > After reading the comments from Marius, I started looking at xrandr. I've > > discovered a couple of things: > > > > 1) If I leave out a 'MetaModes' line, then I see two resolutions in xrandr: > > 2800x1050 and 1400x1050. Upon login, it shows 2800x1050 with the display cloned. > > > > 2) I can switch to 1400x1050 and the display stays cloned. > > > > 3) Switching back to 2800x1050 turns off the clone of the display and gives the > > MergedFB I had under 6.8.2. > > > > If I switch to a text VT the back to X, I'm back at step 1 above. The display > > thinks it is 2800x1050 but the screens are cloned. > > > Sounds like we are missing the call to adjustframe() during serverinit and > enterVT. I assume Crtl-Alt-+/- behaves the same as xrandr? > Hmmm... looks like that should be taken care of. perhaps we need to call it again after messing with the memory controller? Hi Benjamin, you requested a friendly kick in comment 5 if nothing happened after a week. Please find yourself gently kicked. I have a similar problem that I haven't fully figured out yet. I have a MetaModes with 2 modes: First is a clone mode of the size of my laptop panel, and second is a dual head modes with my laptop on the left and the external display on the right. Switching with xrandr between those works fine. However, the mode I obtain at server statup is weird... I get a mode with a virtual resolution covering both screens (dual head like, thus panning around when the cursor hits the borders) but with the second head showing a clone of the primary one. That is a kind of bastard of my first and my second mode Using xrandr -s 1 will switch to full dual head and works fine (first screen stops panning, external display now shows the right part of the desktop, etc...) and switching back to inital mode with xrandr -s 0 also works fine, that is it gives me what i wanted initially, not what the server did: virtual is now the size of the laptop panel and the external display clones it. Bug 6165 which appears similar to this bug is reportedly fixed. Does that fix affect this bug's status? Thanks! -jag Sorry for delay, but i've been out of the country. I'd like to confirm that as per comment 16, the fix for bug 6165 that was committed does indeed also *fix* this bug. So now i'm happily running 7.0 (from a cvs checkout today) in mergedfb mode with both my screens correctly active, in correct positions and sync. I'm ok to close this bug ! Closing |
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.