See #615, #4015 and #12597 as well as Debian #324500 for some history. See my hardware / software configuration in #12597. Opening a new bug because the problem reported in the latter is fixed and this is about dualhead. To recap: VGA == primary, DVI == secondary. Originally, DVI wouldn't start. Using latest git fixed this. In clone mode, no mouse pointer on VGA (latest git didn't fix this :-/) (Btw, #4015 seems to be fixed, too. Clone mode runs the 19" monitor on VGA with 85Hz and the TFT on DVI with 75Hz so that seems settled.) Now, running "xrandr --output VGA --left-of DVI": * DVI monitor is now black. * switching to text mode and back shows that some kind of Dualhead has indeed initialized. I get background on DVI and all application windows on VGA. XFCE stuff (panel etc.) is on DVI, though. AND: the mouse is visible on DVI but when I click, the mouse clicks go to the application windows in VGA (I can not get clicks to register on the XFCE panel on the DVI screen.) The mouse seems to have some invisible space to the *right* of DVI (remember, VGA is to the *left* of DVI, so this shouldn't be the case...) xrandr --output VGA --same-as DVI restores the clone mode, again blanking the DVI until I switch to text mode and back. (This is without matroxfb this time, btw.) I guess http://www.intellinuxgraphics.org/dualhead.html causes the same library/driver calls as xrandr so I won't try this now. I will if you think it helps debugging. Another log and conf attached (xorg.conf still unchanged, haven't looked at the log yet.) And then bedtime. Thanks for the work you're doing here! -- vbi
Created attachment 11786 [details] xorg.conf
Created attachment 11787 [details] Xorg.0.log, have called xrandr --output twice as described
Hmm, a note about the black monitor after calling xrandr: output is *not* switched off, it's black + mouse pointer. I'm not sure if this is actually an X problem or an XFCE problem. But that the mouse can't move to the left monitor and that mouse clicks end up on the wrong display is certainly not XFCE, is it?
Thanks, I pondered filing a bug for this :D For the record, this is http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=444114
I haven't figured out yet what the mode setting does to the output that makes it go black, but I hope I can spend some time on that during the weekend. Does disabling the HWCursor change anything for you? Option "HWCursor" "Off" I don't think it should, but I'm not sure right now.
Ok. Using "HWCursor" "Off" did the trick with the mouse setting and I could finally get working dualhead (non-clone) with mouse on both heads. Cool. Only at 1280 on both screens atm, though. I haven't tried just setting modes with xrandr, but I have tried to do everything at once next. See attached conf and log. The result is both screens dark (no mouse either, but displays didn't go to powersave so output is not entirely switched off), and switching to text console didn't switch on output (computer didn't crash, though. Typing blindly works just fine.) Haven't analyzed this further, gotta go. Won't be around much over the weekend. cheers -- vbi
Created attachment 11809 [details] xorg.conf: try everything at once (dualhead)
Created attachment 11810 [details] Xorg.0.log from the dualhead config.
Looking at the log: seems to be truncated. X crashed? Not sure. (II) MGA(0): Output DVI using monitor section FP767 (**) MGA(0): Option "PreferredMode" "1280x1024" (II) MGA(0): Output VGA using monitor section CM772 (**) MGA(0): Option "PreferredMode" "1440x1080-203" (**) MGA(0): Option "LeftOf" "FP767" in any case reflects true situation, and the edid info for BNQ seems to be for the correct monitor (BNQ == FP767) gotta go, really.
Created attachment 11846 [details] Xorg.0.log (working config)
Created attachment 11847 [details] xorg.conf (working config)
Yay! All I can say: it works. Thanks for being patient with me & thanks for the work you're putting into this. Diff against the last configuration: switched the mnonitors (analog on the left is now on the DVI plug, TFT on the VGA plug.) and fiddled around with the relevant options, see again the latest xorg.conf and log file, and I updated mga_drv to the latest git version (b45d585f07c7da283b6ded2d5f294d097391b942 -- "Set SYNC_CTL to 0xcc when CRTC2 is active.") I had LeftOf at the beginning, but this placed the TFT at offset 1280/0 instead of 1440/0, I guess because the switch to PreferredMode happens only after the display is assembled. Personally, I consider this bug fixed, but I'll leave that up to you. Both monitors now initialize properly after cold boot, but mouse pointer still is not visible on VGA (IIRC) without HWConf Off. cheers -- vbi
(In reply to comment #0) > Now, running "xrandr --output VGA --left-of DVI": > * DVI monitor is now black. > * switching to text mode and back shows that some kind of Dualhead has indeed > initialized. I get background on DVI and all application windows on VGA. I finally found the source of this, it's fixed in fda9377e5e45b4fc3190bad03e081cadc96fe231. I'll assume this is the primary bug this issue is about, so I'm resolving it. The hardware cursor issues are known, but not fixable at this point (I *will* get to it at some point, but don't expect a fix soon). Can you open a separate bug for the other issues you described here (if they are still valid)?
Hi Tilman, > I finally found the source of this, it's fixed in > fda9377e5e45b4fc3190bad03e081cadc96fe231. > I'll assume this is the primary bug this issue is about, so I'm resolving it. is there a chance to create a patch for this fix against the 1.4.[89] series? I also got the not activated second G550 head on a Ubuntu Hardy; for a more detailed description have a look here: <https://bugs.edge.launchpad.net/ubuntu/+source/xserver-xorg-video-mga/+bug/154231> I expect propagation of the 1.9.100 series to a stable distribution release will take some months (or years?), so waiting for this is no real solution. I have already studied the differences between 1.4.8 an 1.9.100 for several hours but got no solution. Guido
Like the first comment on that Launchpad bug says, 1.4.x doesn't even support DVI, so this specific fix cannot be backported. You could backport DVI support to 1.4.x, but I'd rather not do it... btw, 1.9.x probably won't propagate to distros ever, as it's quite buggy.
(In reply to comment #15) > Like the first comment on that Launchpad bug says, 1.4.x doesn't even support > DVI, so this specific fix cannot be backported. > You could backport DVI support to 1.4.x, but I'd rather not do it... btw, 1.9.x > probably won't propagate to distros ever, as it's quite buggy. Thanks for your fast reply. I was not aware of the fact that this fix applies only to DVI signal driven connectors. My not activated monitor is connected to a DVI output, but uses the analog signal. As this bug was not in the X version Ubuntu 6.06 uses, there may be an other solution to fix it for analog driven connectors. Guido
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.