View previous topic :: View next topic |
Author |
Message |
RayDude Advocate
Joined: 29 May 2004 Posts: 2056 Location: San Jose, CA
|
Posted: Sun May 27, 2018 3:43 pm Post subject: nvidia driver 396.24 with gentoo-sources-4.16.1 [resolved] |
|
|
I've been running gentoo on my Ryzen 5-1600 for over a year now. I have an NVidia card and it's also been fine.
But with the kernel upgrade to 4.16.9 (I've also tried 4.16.12) I'm getting errors when sddm runs.
Code: | [ 96.663] (==) NVIDIA(0): No modes were requested; the default mode "nvidia-auto-select"
[ 96.663] (==) NVIDIA(0): will be used as the requested mode.
[ 96.663] (==) NVIDIA(0):
[ 96.663] (II) NVIDIA(0): Validated MetaModes:
[ 96.663] (II) NVIDIA(0): "DFP-1:nvidia-auto-select,DFP-2:nvidia-auto-select"
[ 96.663] (II) NVIDIA(0): Virtual screen size determined to be 5760 x 2160
[ 96.669] (--) NVIDIA(0): DPI set to (95, 94); computed from "UseEdidDpi" X config
[ 96.669] (--) NVIDIA(0): option
[ 96.669] (II) UnloadModule: "modesetting"
[ 96.669] (II) Unloading modesetting
[ 96.669] (--) Depth 24 pixmap format is 32 bpp
[ 96.691] (II) NVIDIA: Using 24576.00 MB of virtual memory for indirect memory
[ 96.691] (II) NVIDIA: access.
[ 99.693] (EE) NVIDIA(GPU-0): Failed to initialize DMA.
[ 99.694] (EE) NVIDIA(0): Failed to allocate push buffer
[ 99.817] (EE)
Fatal server error:
[ 99.817] (EE) AddScreen/ScreenInit failed for driver 0
[ 99.817] (EE)
[ 99.817] (EE)
Please consult the The X.Org Foundation support
at http://wiki.x.org
for help.
[ 99.817] (EE) Please also check the log file at "/var/log/Xorg.0.log" for additional information.
[ 99.817] (EE)
[ 99.819] (EE) Server terminated with error (1). Closing log file. |
Googling shows that the problem has existed for a while. There's a perfect thread from 2015 on nvidia's dev site with no answer.
Has anyone seen this?
Thanks in advance.
Update: I had previously adjusted the kernel settings to add some AMD CPU settings. I deliberately used the previously working .config from 4.16.7 without any modifications and when the kernel built I got these depmod warnings:
Code: | depmod: WARNING: /lib/modules/4.16.12-gentoo/video/nvidia-drm.ko needs unknown symbol sme_me_mask
depmod: WARNING: /lib/modules/4.16.12-gentoo/video/nvidia-uvm.ko needs unknown symbol sme_me_mask
depmod: WARNING: /lib/modules/4.16.12-gentoo/video/nvidia.ko needs unknown symbol sme_me_mask
depmod: WARNING: /lib/modules/4.16.12-gentoo/video/nvidia.ko needs unknown symbol sme_active |
I've seen this before... This is a hint at something.
I'm going to boot to this kernel and then reinstall the nvidia driver and see if it works again. _________________ Some day there will only be free software.
Last edited by RayDude on Sun Jun 03, 2018 4:23 pm; edited 2 times in total |
|
Back to top |
|
|
axl Veteran
Joined: 11 Oct 2002 Posts: 1144 Location: Romania
|
Posted: Sun May 27, 2018 3:57 pm Post subject: |
|
|
just a wild guess but did you blacklist nouveau, uvesafb and vga16fb?
could you provide an output to lsmod and dmesg? |
|
Back to top |
|
|
RayDude Advocate
Joined: 29 May 2004 Posts: 2056 Location: San Jose, CA
|
Posted: Sun May 27, 2018 4:16 pm Post subject: |
|
|
So I restored the original .config file from 4.16.7 to the 4.16.12 folder, recompiled the kernel, reboot, reinstalled the nvidia blob and it's working. That means when I tried to enable some AMD feature of the CPU it caused problems. I was also disabling what I perceived to be intel only features and perhaps AMD is using them as well?
Does anyone have a working Ryzen .config file for 4.16 they can post for me? I'd like to compare notes.
I think it's worrisome that I don't know how to configure the kernel for my Ryzen...
This all started because my system has become unstable in the last few weeks. It lasts about 24 hours and then X freaks out.... I've been trying to figure out what's behind that... _________________ Some day there will only be free software. |
|
Back to top |
|
|
RayDude Advocate
Joined: 29 May 2004 Posts: 2056 Location: San Jose, CA
|
Posted: Sun Jun 03, 2018 4:22 pm Post subject: |
|
|
For those who find this thread after the fact.
It was indeed the nouveau driver that was loading.
I'm not sure how it got enabled in 4.16.12, but it did.
I ended up grabbing Ubuntu 18.04 default config, tweaking it a bit, disabling fb_nvidia and nouveau and then the nvidia driver started working again. _________________ Some day there will only be free software. |
|
Back to top |
|
|
axl Veteran
Joined: 11 Oct 2002 Posts: 1144 Location: Romania
|
Posted: Mon Jun 04, 2018 3:13 am Post subject: |
|
|
happy i fixed your box.
you can always use a kernel parameter to block a module.
$your_kernel_string modprobe.blacklist="nouveau", or "nvidia". you can use the same kernel with both drivers with a bit of bootloader whichcraft. |
|
Back to top |
|
|
RayDude Advocate
Joined: 29 May 2004 Posts: 2056 Location: San Jose, CA
|
Posted: Wed Jun 20, 2018 7:20 pm Post subject: |
|
|
axl wrote: | happy i fixed your box.
you can always use a kernel parameter to block a module.
$your_kernel_string modprobe.blacklist="nouveau", or "nvidia". you can use the same kernel with both drivers with a bit of bootloader whichcraft. |
Thanks much. _________________ Some day there will only be free software. |
|
Back to top |
|
|
|