|
|
#1 | |
|
Registered User
Join Date: Oct 2010
Posts: 3
|
I'm trying to run some GL apps through ssh (indirect rendering). The card is a 285GTX, drivers are 260.12.06. Glxgears and blender work, but others (bzflag and etracer, for example) crash with
Code:
X Error of failed request: BadLength (poly request too large or internal Xlib length error) Major opcode of failed request: 136 (GLX) Minor opcode of failed request: 2 (X_GLXRenderLarge) Serial number of failed request: 14605 Current serial number in output stream: 14648 I'm attaching nvidia-bug-report.log.gz, glxinfo output and Xorg.log. Thanks for any help |
|
|
|
|
|
|
#2 | |
|
NVIDIA Corporation
Join Date: Mar 2005
Posts: 2,487
|
Your glxinfo.log file shows that you're using the Mesa version of libGL on the cilent side. Does the problem go away if you use the NVIDIA libGL instead? This is likely to be a bug in the Mesa indirect GLX protocol handling.
|
|
|
|
|
|
|
#3 |
|
Registered User
Join Date: Oct 2010
Posts: 3
|
Hi AaronP, thanks for your answer. I tried copying /usr/lib/nvidia-current/* to the server (application server, so the glx client), but this makes every gl program segfault. Maybe this is because the server has no nvidia video card or driver (actually, it is a virtual machine)? Is there another way of using nvidia's libgl? Thanks
|
|
|
|
|
|
#4 | |
|
NVIDIA Corporation
Join Date: Mar 2005
Posts: 2,487
|
You need to make sure you copy over the correct TLS library too, which usually lives in /usr/lib/tls, but given that your libGL was in the wrong place (/usr/lib/nvidia-current instead of the standard /usr/lib), you may find it somewhere else.
The NVIDIA client-side libraries should work for indirect GLX rendering even on systems without NVIDIA hardware. |
|
|
|
|
|
|
#5 |
|
Registered User
Join Date: Oct 2010
Posts: 3
|
Thanks, that worked
The tls libs on my system are still in /usr/lib/nvidia-current, but in the initial copy I didn't include subdirectories and that messed up everything. Now it is working properly. Also a big thank you to the NVIDIA engineer who decided to implement (although unofficially) OpenGL 2 through GLX. |
|
|
|
![]() |
| Thread Tools | |
|
|