|
|
#1 | |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
My Iiyama ProLite E2773HDS works perfectly on Windows 7, but whenever I just plug it in via HDMI on Linux (x64) the X server immediately crashes, regardless of my xorg.conf present or none at all. If I try to log in from ngetty (to xmonad) the desktop sometimes flashes for less then a second, then it's back to ngetty again.
The same crash will happen one time after X is started without this second monitor, thereafter at least the primary one works again. Aside from the memory positions, the backtrace in Xorg.0.log is always the same. xorg-server 1.12.1, nvidia 295.40, glibc 2.15 (w/ Arch Linux patchset -10), kernel 3.3.2, 9600M GT Occurred also with previous versions. Attached nvidia-bug-report created with no xorg.conf present (used X server 1.12.0.902, kernel 3.3.1). Last edited by misc; 04-14-12 at 06:07 AM. |
|
|
|
|
|
|
#2 | |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
Any comments, devs? This situation is really lousy, I'm doing most of my work with Linux.
edit: Explicitly disabling composite changes nothing. |
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
Explicitly disabling Xinerama doesn't change anything either.
However, the display works fine with nouveau. Thus I'm tempted to say that this indeed is a bug in your driver, or at least the interaction between it and xorg (d'uh). (Won't use nouveau though since I need vdpau/etc. acceleration and power management. But sweet goodness, KMS makes the console look so much nicer on a FHD display.) edit: With no relevant source changes, same for X server 1.12.1 and kernel 3.3.2. |
|
|
|
|
|
#4 | |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
With both (and only both) RenderAccel set to False and the first Metamodes "1920x1080,NULL;" the X Server will at least properly start (and still crash if the HDMI cable is plugged in/out).
However, it shows a corrupted oversized image (example) if there's another metamode given, and nvidia-settings will not detect the second display in "X Server Display Configuration", though it does list it as DFP-1 under GPU 0. Also, that monitor will remain in standby. edit: RenderAccel may remain true, apparently. Correction: The monitor is listed properly in Display Configuration, I just had to actually select it in "Model:". The crash then occurs once it's enabled, regardless of the chosen resolution (eg Auto). Last edited by misc; 04-16-12 at 10:58 AM. |
|
|
|
|
|
|
#5 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
If just tried it with a DVI:HDMI (monitor:laptop) cable, and while the crash is exactly the same if the cable if plugged in/out with X running (same with the different monitor it belongs to), this way the display at least works.
I do not want this workaround to become permanent, though. |
|
|
|
|
|
#6 | |
|
NVIDIA Corporation
Join Date: Dec 2010
Posts: 259
|
Please attach dmidecode o/p.
|
|
|
|
|
|
|
#7 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
dmidecode 2.12-cvs
|
|
|
|
|
|
#8 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
So, what does it look like? Any status?
|
|
|
|
|
|
#9 |
|
NVIDIA Corporation
Join Date: Dec 2010
Posts: 259
|
NVIDIA internal Bug Id 966827 to track this issue.
|
|
|
|
|
|
#10 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
Yay, thanks.
Is it possible for unprivileged users like me to check that bug id's progress? |
|
|
|
|
|
#11 |
|
Registered User
Join Date: Apr 2012
Posts: 11
|
Gone with 302.07 (using a fresh X Server compile from git). Hotplugging works fine, too.
|
|
|
|
![]() |
| Thread Tools | |
|
|