[OpenIndiana-discuss] Nvidia 515 problem

Wed Nov 9 16:53:40 UTC 2022

Am 09.11.22 um 17:34 schrieb Stephan Althaus:
> On 11/9/22 13:40, Udo Grabowski (IMK) wrote:
>>
>>
>> On 09/11/2022 13:04, Stephan Althaus wrote:
>>> On 11/9/22 13:02, Udo Grabowski (IMK) wrote:
>>>>
>>>>
>>>> On 09/11/2022 12:22, Udo Grabowski (IMK) wrote:
>>>>>
>>>>>
>>>>> On 09/11/2022 12:14, Udo Grabowski (IMK) wrote:
>>>>>>
>>>>>>
>>>>>> On 09/11/2022 12:08, Stephan Althaus wrote:
>>>>>>> On 11/9/22 11:32, Marcel Telka wrote:
>>>>>>>> On Wed, Nov 09, 2022 at 11:22:01AM +0100, Udo Grabowski (IMK)
>>>>>>>> wrote:
>>>>>>>>>
>>>>>>>>> On 09/11/2022 10:21, Stephan Althaus wrote:
>>>>>>>>>> On 11/9/22 10:15, Marcel Telka wrote:
>>>>>>>>>>> On Wed, Nov 09, 2022 at 08:44:58AM +0100, Stephan Althaus
>>>>>>>>>>> wrote:
>>>>>>>>>>>> Hello!
>>>>>>>>>>>>
>>>>>>>>>>>> Yesterday i tried to use nvidia driver version 515 with my
>>>>>>>>>>>> M1200M,
>>>>>>>>>>>> but got an error in the Xorg.0.log:
>>>>>>>>>>>>
>>>>>>>>>>>>
>>>>>>>>>>>> [   234.097] (II) Loading sub module "glxserver_nvidia"
>>>>>>>>>>>> [   234.097] (II) LoadModule: "glxserver_nvidia"
>>>>>>>>>>>> [   234.098] (WW) Warning, couldn't open module
>>>>>>>>>>>> glxserver_nvidia
>>>>>>>>>>>> [   234.098] (II) UnloadModule: "glxserver_nvidia"
>>>>>>>>>>>> [   234.098] (II) Unloading glxserver_nvidia
>>>>>>>>>>>> [   234.098] (EE) NVIDIA: Failed to load module
>>>>>>>>>>>> "glxserver_nvidia" (module
>>>>>>>>>>>> does not exist, 0)
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0): Failed to initialize the GLX
>>>>>>>>>>>> module; please
>>>>>>>>>>>> check in your X
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0):     log file that the GLX
>>>>>>>>>>>> module has been
>>>>>>>>>>>> loaded in your X
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0):     server, and that the
>>>>>>>>>>>> module is
>>>>>>>>>>>> the NVIDIA
>>>>>>>>>>>> GLX module.  If
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0):     you continue to encounter
>>>>>>>>>>>> problems, Please
>>>>>>>>>>>> try
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0): reinstalling the NVIDIA driver.
>>>>>>>>>>>> [   234.098] (II) NVIDIA: The X server does not support
>>>>>>>>>>>> PRIME Render
>>>>>>>>>>>> Offload.
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(GPU-0): Failed to initialize the
>>>>>>>>>>>> NVIDIA
>>>>>>>>>>>> graphics
>>>>>>>>>>>> device!
>>>>>>>>>>>> [   234.098] (EE) NVIDIA(0): Failing initialization of X
>>>>>>>>>>>> screen
>>>>>>>>>>>> [   234.098] (II) UnloadModule: "nvidia"
>>>>>>>>>>>>
>>>>>>>>>>>> i do not find glxserver_nvidia in the running version 470:
>>>>>>>>>>>>
>>>>>>>>>>>> $ grep -i glx /var/log/Xorg.0.log
>>>>>>>>>>>> [    90.930] (II) LoadModule: "glx"
>>>>>>>>>>>> [    90.932] (II) Loading
>>>>>>>>>>>> /usr/lib/xorg/modules/extensions/amd64/libglx.so
>>>>>>>>>>>> [    90.986] (II) Module glx: vendor="NVIDIA Corporation"
>>>>>>>>>>>> [    90.986] (II) NVIDIA GLX Module 470.141.03 Thu Jun 30
>>>>>>>>>>>> 18:38:53 UTC
>>>>>>>>>>>> 2022
>>>>>>>>>>>> [    93.742] (II) Initializing extension GLX
>>>>>>>>>>>> [    93.742] (II) Indirect GLX disabled.
>>>>>>>>>>>>
>>>>>>>>>>>> Does anyone has a hint on how to solve this?
>>>>>>>>>>> There is no '*glxserver_nvidia*' file in the Solaris driver
>>>>>>>>>>> distribution
>>>>>>>>>>> https://us.download.nvidia.com/solaris/515.76/NVIDIA-Solaris-x86-515.76.run
>>>>>>>>>>>
>>>>>>>>>>> so I'm not sure why they expect such file (module) to exist.
>>>>>>>>>>> Maybe this
>>>>>>>>>>> is a bug in their package?  Either they forgot to add the
>>>>>>>>>>> module into
>>>>>>>>>>> the package, or they mistakenly require it when they
>>>>>>>>>>> shouldn't? Maybe
>>>>>>>>>>> you could try to contact NVIDIA.
>>>>>>>>>>>
>>>>>>>>>>>
>>>>>>>>>>> Regards.
>>>>>>>>>>>
>>>>>>>>>> Hello!
>>>>>>>>>>
>>>>>>>>>> Looking a bit around i think its the GLX module that is not
>>>>>>>>>> "NVIDIA"
>>>>>>>>>>
>>>>>>>>>> [   234.053] (II) LoadModule: "glx"
>>>>>>>>>> [   234.055] (II) Loading
>>>>>>>>>> /usr/lib/xorg/modules/extensions/amd64/libglx.so
>>>>>>>>>> [   234.071] (II) Module glx: vendor="X.Org Foundation"
>>>>>>>>>> [   234.071]     compiled for 1.19.7, module version = 1.0.0
>>>>>>>>>> [   234.071]     ABI class: X.Org Server Extension, version 10.0
>>>>>>>>>>
>>>>>>>>>> and the glxserver is a method within that or something similar..
>>>>>>>>>> ...
>>>>>>>>> # nm -D
>>>>>>>>> NVDAgraphics/reloc/X11/lib/modules/extensions/NVIDIA/amd64/libglx.so.1|fgrep
>>>>>>>>>
>>>>>>>>> -i glx
>>>>>>>>>
>>>>>>>>> [6]     |            14550384| 24|OBJT |GLOB |0 |22
>>>>>>>>> |glxserver_nvidiaModuleData
>>>>>>>>>
>>>>>>>>> That entrypoint is not present in , e.g., 295.40 . So probably a
>>>>>>>>> packaging error by NVIDIA.
>>>>>>>> I thought initially the same, but:
>>>>>>>>
>>>>>>>> $ /usr/bin/nm -D
>>>>>>>> nvidia-470/NVIDIA-Solaris-x86-470.141.03/NVDAgraphics/reloc/X11/lib/modules/extensions/NVIDIA/amd64/libglx.so.1|fgrep
>>>>>>>> -i glx
>>>>>>>> nvidia-470/NVIDIA-Solaris-x86-470.141.03/NVDAgraphics/reloc/X11/lib/modules/extensions/NVIDIA/amd64/libglx.so.1:
>>>>>>>>
>>>>>>>> [55]    |            10124656| 35|FUNC |GLOB |0 |12    
>>>>>>>> |__glXNvHandShake
>>>>>>>> [60]    |            14546160| 24|OBJT |GLOB |0 |22    
>>>>>>>> |glxModuleData
>>>>>>>> [7]     |            14546032| 24|OBJT |GLOB |0 |22    
>>>>>>>> |glxserver_nvidiaModuleData
>>>>>>>> $ /usr/bin/nm -D
>>>>>>>> nvidia/NVIDIA-Solaris-x86-515.76/NVDAgraphics/reloc/X11/lib/modules/extensions/NVIDIA/amd64/libglx.so.1|fgrep
>>>>>>>> -i glx
>>>>>>>> nvidia/NVIDIA-Solaris-x86-515.76/NVDAgraphics/reloc/X11/lib/modules/extensions/NVIDIA/amd64/libglx.so.1:
>>>>>>>>
>>>>>>>> [55]    |            10128960| 35|FUNC |GLOB |0 |12    
>>>>>>>> |__glXNvHandShake
>>>>>>>> [61]    |            14550512| 24|OBJT |GLOB |0 |22    
>>>>>>>> |glxModuleData
>>>>>>>> [6]     |            14550384| 24|OBJT |GLOB |0 |22    
>>>>>>>> |glxserver_nvidiaModuleData
>>>>>>>> $
>>>>>>>>
>>>>>>>> and 470.141.03 seems to work okay (at least this is my
>>>>>>>> understanding from OP).
>>>>>>>>
>>>>>>>>
>>>>>>> 470 works. From the running machine with 470 i follow the GLX
>>>>>>> module from Xorg.0.log like this:
>>>>>>>
>>>>>>> $ grep -i glx /var/log/Xorg.0.log
>>>>>>> [    90.930] (II) LoadModule: "glx"
>>>>>>> [    90.932] (II) Loading
>>>>>>> /usr/lib/xorg/modules/extensions/amd64/libglx.so
>>>>>>> [    90.986] (II) Module glx: vendor="NVIDIA Corporation"
>>>>>>> [    90.986] (II) NVIDIA GLX Module  470.141.03  Thu Jun 30
>>>>>>> 18:38:53 UTC 2022
>>>>>>> [    93.742] (II) Initializing extension GLX
>>>>>>> [    93.742] (II) Indirect GLX disabled.
>>>>>>> ,,,
>>>>>
>>>>> Which shows the correct vendor, while the non-working says Xorg.
>>>>> Just curious: Did you trigger a 'bootadm update-archive' by 'init 6'
>>>>> or just 'reboot' ?
>>>>
>>>> Not the culprit, 'reboot' also updates the boot archive.
>>>>
>>>> _______________________________________________
>>>> openindiana-discuss mailing list
>>>> openindiana-discuss at openindiana.org
>>>> https://openindiana.org/mailman/listinfo/openindiana-discuss
>>>
>>> Don't know for sure. This afternoon, I will boot the BE, do a
>>> 'bootadm update-archive' annd reboot again to see what happens..
>>
>> You can check then what the Xorg process is actually linking to:
>> 'ps -ef|grep X' pick the PID, then
>> 'ls -l /proc/<PID>/path/'
>> should show where it grabs the libglx.so library.
>>
>> _______________________________________________
>> openindiana-discuss mailing list
>> openindiana-discuss at openindiana.org
>> https://openindiana.org/mailman/listinfo/openindiana-discuss
>
> Hello!
>
> I did boot the BE with the nvidia 515 driver into single mode and did
> a pkg update && init 6.
>
> Xorg works and i gt the login screen from lightdm, GLX is used from
> NVIDIA:
>
> [   394.470] (II) LoadModule: "glx"
> [   394.471] (II) Loading
> /usr/lib/xorg/modules/extensions/amd64/libglx.so
> [   394.501] (II) Module glx: vendor="NVIDIA Corporation"
> [   394.501] (II) NVIDIA GLX Module  515.76  Mon Sep 12 19:15:55 UTC 2022
> [   396.450] (II) Initializing extension GLX
> [   396.450] (II) Indirect GLX disabled.
>
> When i login, X closes, and i get really strange things on the
> appearing text console when i type on my (usb) keyboard

I think I got the same problem. You could try to login into another
session type, eg. xterm only (and then manually start twm), or whatever
window manager you have additionally installed. I was able to have
working sessions with enlightenment and twm. There seem to be something
used by Mate's window manager that segfaults (but doesn't leave a
coredump in my case). At least Xorg.log.old is telling me so.

Andreas