You are not logged in.
Did you also add the modules to mkinitcpio?
lsinitcpio /boot/initramfs-linux-zen.img | grep -i nvidia
The nvidia driver is loaded, but the GPU is taken by the modsetting driver.
Since this is an optimus system, ensure that you read https://wiki.archlinux.org/index.php/NVIDIA_Optimus and elaborate on your setup.
As the GPU isn't detected by any driver, also post the output of "lspci"
Offline
Thanks Johnathon for all your work keeping this working. I also have the hardware (NVS 4200M) and am willing to test, but I'm not sure how to do that. If someone who knows how could give a high-level guide or point me in the direction of more info I would really appreciate it. Currently the best I can do is reboot and see if lightdm fails to start.
Offline
I ... am willing to test, but I'm not sure how to do that.
Essentially, install the driver and use it. If it's a hybrid/Optimus setup then there's a link above. If it's a dedicated GPU then install and configure as normal.
Offline
Thanks to a pointer from BlackIkeEagle I have added a patch for kernel 5.12 support (taken from https://build.opensuse.org/package/view … h?expand=1, credit for this one goes to Larry Finger): https://aur.archlinux.org/cgit/aur.git/ … ecb2e5f69b
If anyone wants to test with linux-mainline in advance of the 5.12 release then let me know how it goes!
Last edited by jonathon (2021-04-06 11:19:23)
Offline
PKGBUILDs updated for 390.143 (https://forums.developer.nvidia.com/t/l … pus/175646).
5.11 appears to be supported upstream, I've refreshed the 5.12 patch but it needs to be checked/tested.
Last edited by jonathon (2021-04-23 11:47:45)
Offline
Hi!
I updated today but the dkms modules had problems, here are some logs:
Yay: https://pastebin.com/fGi0HZwG
Dkms: https://pastebin.com/fhLwdcB8
Doesn't look like a big deal though.
I'm on 5.11.16, without the 5.12 patch, it works just fine, thank you.
Last edited by Filip62 (2021-04-23 15:12:46)
Offline
Ack, I need to #include <linux/version.h>. Stand by.
Edit: OK, go.
Last edited by jonathon (2021-04-23 15:46:32)
Offline
Builds and installs and works perfectly on 5.11 now, thanks.
Offline
hi
I can't upgrade to the newest package, as it depends on the older version of nvidia-390xx-utils
ütköző csomagok keresése...
hiba: nem sikerült előkészíteni a tranzakciót (nem sikerült kielégíteni a függőségeket)
:: anvidia-390xx-utils (390.143-3) telepítése megtöri a 'nvidia-390xx-utils=390.141' függőségét, amely nvidia-390xx által szükséges
:: anvidia-390xx-utils (390.143-3) telepítése megtöri a 'nvidia-390xx-utils=390.141' függőségét, amely nvidia-390xx-dkms által szükséges
Command 'sudo pacman --color=always --upgrade --asdeps /home/lenry/.cache/pikaur/pkg/nvidia-390xx-utils-390.143-3-x86_64.pkg.tar.zst' failed to execute.
It translates to: installing nvidia-390xx-utils (390.143-3) breaks the dependency of nvidia-390xx-utils=390.141 which is needed by nvidia-390xx
I can't figure out where and what to modify
Offline
It translates to: installing nvidia-390xx-utils (390.143-3) breaks the dependency of nvidia-390xx-utils=390.141 which is needed by nvidia-390xx
I can't figure out where and what to modify
Whenver something like this happens, you probably need to remove nvidia-390xx first, somewhat like this:
Build new nvidia-390xx-utils
Uninstall old nvidia-390xx
Install new nvidia-390xx-utils
Build new nvidia-390xx
Install new nvidia-390xx
As you can see, there is a window where the old nvidia-390xx is not installed but the new nvidia-390xx is not even built. Also, nvidia-390xx needs to be built for each kernel anew. Therefore, it might be a good idea to have
the old nvidia-390xx as well as
its dependency nvidia-390xx-utils and
the kernel these old packages where built for
pre-packaged and ready to install somewhere in case something goes wrong.
Offline
Lenry wrote:It translates to: installing nvidia-390xx-utils (390.143-3) breaks the dependency of nvidia-390xx-utils=390.141 which is needed by nvidia-390xx
I can't figure out where and what to modify
Whenver something like this happens, you probably need to remove nvidia-390xx first, somewhat like this:
Build new nvidia-390xx-utils
Uninstall old nvidia-390xx
Install new nvidia-390xx-utils
Build new nvidia-390xx
Install new nvidia-390xx
As you can see, there is a window where the old nvidia-390xx is not installed but the new nvidia-390xx is not even built. Also, nvidia-390xx needs to be built for each kernel anew. Therefore, it might be a good idea to have
the old nvidia-390xx as well as
its dependency nvidia-390xx-utils and
the kernel these old packages where built for
pre-packaged and ready to install somewhere in case something goes wrong.
Thank you, it worked perfectly
Offline
Updated to 5.12 today, no issues so far.
Offline
I have been using this for over a week now on 5.12 with no noticeable issues.
Offline
Has anyone got Vulkan working with this driver? I posted in another thread, but basically I can't get Vulkan working on an Optimus system. One reply mentioned checking for /usr/share/vulkan/implicit_layer.d/nvidia_layers.json. I know the package nvidia-utils provides this file, but nvidia-390xx-utils does not provide it. I'm not sure if that's why Vulkan fails.
Offline
Has anyone got Vulkan working with this driver? I posted in another thread, but basically I can't get Vulkan working on an Optimus system. One reply mentioned checking for /usr/share/vulkan/implicit_layer.d/nvidia_layers.json. I know the package nvidia-utils provides this file, but nvidia-390xx-utils does not provide it. I'm not sure if that's why Vulkan fails.
I tried copying nvidia_layers.json from the most-recent nvidia package, modifying the "api_version" to be consistent with the nvidia-390xx package, but this did not work. Does anyone have a hint, or is getting Vulkan to work with nvidia-309xx impossible? nvidia-390xx-utils does provide /usr/share/vulkan/icd.d/nvidia_icd.json, so it looks like it's trying to provide at least some support for Vulkan.
Offline
google "nvidia fermi vulkan"
Offline
google "nvidia fermi vulkan"
Thanks Seth. Looks like my Nvidia GeForce GT 555M is a Fermi card, which doesn't and won't get Vulkan. Oh well.
Offline
I've pushed initial support for kernel 5.13 (adapted from Alberto Milone's patch for Ubuntu).
Feel free to test.
Offline
Well this sucks, This driver has been flawless for me until I tried the new 5.13.0.1-realtime1-1-rt kernel today. It fails installing for this kernel with the message below. I'll see if I can cook up a patch.
CC [M] /var/lib/dkms/nvidia/390.143/build/nvidia-drm/nvidia-drm-gem-nvkms-memory.o
CC [M] /var/lib/dkms/nvidia/390.143/build/nvidia-drm/nvidia-drm-gem-user-memory.o
LD [M] /var/lib/dkms/nvidia/390.143/build/nvidia.o
LD [M] /var/lib/dkms/nvidia/390.143/build/nvidia-uvm.o
LD [M] /var/lib/dkms/nvidia/390.143/build/nvidia-modeset.o
LD [M] /var/lib/dkms/nvidia/390.143/build/nvidia-drm.o
MODPOST /var/lib/dkms/nvidia/390.143/build/Module.symvers
ERROR: modpost: GPL-incompatible module nvidia-drm.ko uses GPL-only symbol '__rt_mutex_init'
ERROR: modpost: GPL-incompatible module nvidia-modeset.ko uses GPL-only symbol '__rt_mutex_init'
ERROR: modpost: GPL-incompatible module nvidia-uvm.ko uses GPL-only symbol '__rt_mutex_init'
ERROR: modpost: GPL-incompatible module nvidia.ko uses GPL-only symbol '__rt_mutex_init'
make[2]: *** [scripts/Makefile.modpost:150: /var/lib/dkms/nvidia/390.143/build/Module.symvers] Error 1
make[2]: *** Deleting file '/var/lib/dkms/nvidia/390.143/build/Module.symvers'
make[1]: *** [Makefile:1762: modules] Error 2
make[1]: Leaving directory '/usr/lib/modules/5.13.0.1-realtime1-1-rt/build'
make: *** [Makefile:81: modules] Error 2
--
Randy DuCharme (Radio AD5GB)
Powered by Open Source software!
Offline
390.144 has just been released. Patches for 5.12 and 5.13 don't seem to be necessary any more, DKMS module seems to compile fine (but I have no way to actually test it works).
Have fun!
Offline
LOL Jonathon! I am just now sitting down to look at it. The whole day went haywire. I'll try the 390.144 release first. It wasn't there this morning. I'd been running DKMS on various 5.12 and 5.13 kernels - just not the realtime one that just came out. Been a long time since I've been in the Linux kernel. 4-5 years ago at least and it was Android at that. There can't be much to i. 390.143 DKMS is running fine on the AUR 5.13-xanmod-cacule kernel right now. If I end up getting into it and come up with something useful, you will have it.
--
Randy DuCharme (Radio AD5GB)
Powered by Open Source software!
Offline
I can confirm that 390.144 works fine on kernel 5.13.x, requiring no patches. Update went fine here on Arch, and I've been able to use the .run file from Nvidia on Slackware without patching, as was necessary before 390.144. I have an ancient-ish machine (Intel Core i7 K 875) with a GeForce GT 430 card.
Why not enjoy a delightful Royal Crown cola?
Offline
I can confirm that 390.144 works fine on kernel 5.13.x, requiring no patches. Update went fine here on Arch, and I've been able to use the .run file from Nvidia on Slackware without patching, as was necessary before 390.144. I have an ancient-ish machine (Intel Core i7 K 875) with a GeForce GT 430 card.
I've got a GeForce GT 450 card in an AMD FX-9590 system. Yes DKMS 390.144-1 works great on every other kernel I've used until I tried the latest AUR linux-rt kernel ( linux-rt 5.13.0.1.realtime1-1 ). I'm trying to shave as much latency as possible and the Preempt-RT patched kernels work absolutely fantastic on my Raspberry-Pi's.
I'm just now sitting back down to work on this again. I'd be very surprised if you installed linux-rt and did not have the same problem. If you do decide to play though, I'd love to hear how you make out with it. It could very well be the kernel and not the NVidia driver. Wish I had a few different NVidia adapters to play with actually.
--
Randy DuCharme (Radio AD5GB)
Powered by Open Source software!
Offline
Hello guys,
I have a bit of a problem with new kernel and package. I am using the 5.13.4-arch1-1 kernel from the repo. I updated yesterday. DMKS went fine. The driver seems to be loaded, and there are no problems in dmesg. However I am getting "(EE) No devices detected." and "(EE) no screens found(EE)" in the X logs.
I have GeForce GT 430. It is actually a virtual machine with VGA passthrough, but I doubt that is a problem.
Best regards.
Last edited by Gruntz (2021-07-23 08:06:11)
Offline
[ 241.915] (--) PCI:*(0@0:2:0) 1013:00b8:1af4:1100 rev 0, Mem @ 0xc0000000/33554432, 0xc348a000/4096, BIOS @ 0x????????/131072
[ 241.915] (--) PCI: (0@0:8:0) 10de:0de1:0000:0000 rev 161, Mem @ 0xc2000000/16777216, 0x800000000/134217728,
00:02.0 VGA compatible controller: Cirrus Logic GD 5446
00:08.0 VGA compatible controller: NVIDIA Corporation GF108 [GeForce GT 430] (rev a1)
You basically created a hybrid graphics setup with the cirrus vga contorller (added by qemu) as primary card and the nvidia card as secondary .
I doubt that's what you want, change the way you start the VM to ensure the nvidia card is the only videocard it gets.
Last edited by Lone_Wolf (2021-07-23 11:25:37)
Disliking systemd intensely, but not satisfied with alternatives so focusing on taming systemd.
clean chroot building not flexible enough ?
Try clean chroot manager by graysky
Offline