Open vieladams opened 5 years ago
Same issue here. I believe the kernel doesn't match the linux headers. If I try to install linux headers using dom0: sudo qubes-dom0-update linux-headers-$(uname -r)(or sudo qubes-dom0-update linux-headers$(uname -r)), I get: No match for argument: linux-headers-4.14.74-1.pvops.qubes.x86_64 Error: Unable to find a match. I'm getting the devel and headers version: dom0# rpm -qa | grep -E "kernel-devel|kernel-headers"
the oputput: Kernel-devel-4.19.82-1 (from a previous upgrade, I guess this one should be OK if you are not updating dom0, this is a match with latest kernel available on stable repo) and: kernel-headers-4.13.16-100 My suggestion is to try to downgrade your kernel version to 4.13.16-100 and retry to install the nvidia again...I will try it and get back with the results.
Try qubes-dom0-update kernel-devel-$(uname -r)
Update:
I've updated the kernel to 4.19.82-1, to match the kernel-devel-4.19.82-1, the same error pup up. I am unable to find the kernel-headers to match the kernel. Again, I believe that's the only reason Nvidia fails.
I've also updated to latest kernel(5.3**) available in repo and make it work(blacklisting the nouveau), but lightdm crashes on login. Is there any method to connect to the internet using TY2(command line)?
Need to mention that my bios doesn't have legacy option(I have a alienware area 51m i9 w Nvidia rtx2080). I can install only Qubes Release 4.0.1(Kernel 4.14.74-1), without issues, but the only resolution available is 800x600. I've been tried xrandr to force the resolution to 1920x1080 with no luck. I've been used Qubes for years now and I really like it, I don't want to move to other distribution. My previous machine was a MSI stealth 65r(i7 RTX1080) and I've managed to make it work properly after changing the uefi to legacy and blacklisting nouveau. Any help will be highly appreciated. Thanks
Update: following marmarek advice,
dom0 ~]$ sudo qubes-dom0-update kernel-devel-$(uname -r)
...
kernel-devel-4.14.74-1.pvops.qubes.x86_64.rpm 149 kB/s | 12 MB 01:23
--------------------------------------------------------------------------------
Total 147 kB/s | 12 MB 01:24
Complete!
The downloaded packages were saved in cache until the next successful transaction.
You can remove cached packages by executing 'dnf clean packages'.
Qubes OS Repository for Dom0 129 MB/s | 133 kB 00:00
No package kernel-devel-4.14.74-1.pvops.qubes.x86_64 available.
Error: Unable to find a match.
It can't install it using dom0 terminal. However, the package was downloaded and I've done a manual installation:
dom0: sudo dnf install /var/lib/qubes/updates/rpm/kernel-devel-4.14.74-1.pvops.qubes.x86_64.rpm
Now I have:
dom0 ~]$ uname -r
Output:
4.14.74-1.pvops.qubes.x86_64
kernel-devel-4.14.74-1
dom0 ~]$ rpm -qa | grep -e kernel-devel
Output:
kernel-devel-4.14.74-1.pvops.qubes.
dom0 ~]$ rpm -qa | grep -e kernel-headers
Output:
kernel-headers-4.13.16-100.fc25.x86_64
So far so good. The kernel matches the kernel-devel. Next step, kernel-headers.
dom0: sudo qubes-dom0-update kernel-headers-$(uname -r)
dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-headers-$(uname -r)
dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-security-testing kernel-headers-$(uname -r)
dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-unstable kernel-headers-$(uname -r)
with output:
No match for argument: kernel-headers-4.14.74-1.pvops.qubes.x86_64
Error: Unable to find a match.
I will give a try to install it now.
There is no kernel-headers package, the right one is kernel-devel
You're right. The installer started, I didn't get kernel-headers error again, but the install crashed again. After a look in nvidia-installation.log, a package was missing: elfutils-libelf-devel so: dom0: sudo qubes-dom0-update elfutils-libelf-devel everything was smooth again, init 3 and restart the installation of Nvidia. another crash, with: ld: cannot find /tmp/selfgz6575/NVIDIA-Linux-x86_64-440.31/kernel/nvidia/linux_nvswitch.o: No such file or directory /usr/src/kernels/4.14.74-1.pvops.qubes.x86_64/scripts/Makefile.build:574: re$ make[3]: [/tmp/selfgz6575/NVIDIA-Linux-x86_64-440.31/kernel/nvidia.o] Er$ make[3]: Target '__build' not remade because of errors. /usr/src/kernels/4.14.74-1.pvops.qubes.x86_64/Makefile:1527: recipe for targ$ make[2]: [module/tmp/selfgz6575/NVIDIA-Linux-x86_64-440.31/kernel] Err$ make[2]: Target 'modules' not remade because of errors. make[2]: Leaving directory '/usr/src/kernels/4.14.74-1.pvops.qubes.x86_64' Makefile:146: recipe for target 'sub-make' failed make[1]: [sub-make] Error 2 make[1]: Target 'modules' not remade because of errors. make[1]: Leaving directory '/usr/src/kernels/4.14.74-1.pvops.qubes.x86_64' Makefile:81: recipe for target 'modules' failed make: [modules] Error 2 ERROR: The nvidia kernel module was not created. ERROR: Installation has failed. Please see the file '/var/log/nvidia-instal$.
No luck at all. This is the output:
Is there any command line to connect to the internet directly from the terminal? I want to try to update dom0 to latest kernel also update the kernel-devel and give another try. I cannot use the graphic interface since lightdm is crashing but I can login trough Ctrl Alt F2. Or any chance to give dom0 internet access directly?
Finally manage to install it but I have other issues now. The lighdm doesn't start. It is stuck at: starting hold until boot process finishes up. However, nvidia is finally installed. Steps: Updated kernel and kernel-devel to latest using unstable repositories. I will take a look at nvidia installer log and go ahead from there. I believe it has something to do with xorg.conf. hopefully I'll sort it.
can you please post your xorg.conf if it worked ?
@vieladams - Yes, as soon as I get back home
Again, managed to update and match the kernel and kernel-devel to 5.4.0-0. With both installed, I've disabled the nouveau in xen.cfg. I've downloaded the Nvidia driver in sys-firewall and transfer it to dom0 reboot, Ctrl+F2(lightdm si dead, for some reasons...) I've been tried to install it using the old method: https://groups.google.com/forum/#!topic/qubes-users/v26zXkiNElg without success. Next step, chmod +x nvidia.run sudo ./nvidia run
The Nvidia was installed successfully.(at least, this was the last message).
restarted and it doesn't work. The image is flashing like something is crashing on lightdm at login and the text is coming back over and over again.
Have I missed something?
This is the xorg.0.log
Nvidia installer log/:
and Xorg.conf:
I feel I'm very close....
Does anyone know if the issue may be caused by lightdm? Maybe it was broken somehow during the installation of nvidia?
Again, is there any possibility to connect to the internet trough sys-net and sys-firewall from dom0 terminal? I will start over and install KDE, as a backup for lighten, and after nvidia installer finished, I will switch to KDE environmen. Finger crossed
As default, after a fresh installation, my xorg.conf.d folder shows the graphic card is a radeon. (10-radeon.conf). My cards are Intel and nvidia. Any suggestions? If I change the conf file to Intel conf one or nvidia, can I force a higher resolution? My default resolution is 800x600. I've been tried to force change it using xrandr with no luck
Unsuccessfull..again. I guess I need to wait until the next release of dom0 with fc29...
Update: I've managed to download and install Nvidia correctly Using RPM fusion. ISO: Qubes-20200113, Dom0-Fedora 31 Kernel: 5.4.5-1.qubes.x86_64.
My steps: Downloaded the iso and wrote it to USB. nomodeset(instead disabling nouveau) this is the only way and since I have grub access in EFI, I don't have to edit the iso.
Downloaded and imported RPM fusion Keys from: http://rpmfusion.org/keys?action=AttachFile&do=get&target=RPM-GPG-KEY-rpmfusion-nonfree-fedora-31 and http://rpmfusion.org/keys?action=AttachFile&do=get&target=RPM-GPG-KEY-rpmfusion-free-fedora-31
move it to dom0 and import it: sudo rpmkeys --import /path/to/signing/key
Downloaded rpmfusion repos and moved to dom0:
https://download1.rpmfusion.org/free/fedora/rpmfusion-free-release-31.noarch.rpm https://download1.rpmfusion.org/nonfree/fedora/rpmfusion-nonfree-release-31.noarch.rpm
Install the rpm's: sudo rpm -i /path/to/file/*.rpm --nodeps
modified the repo files: $release with 31 and the architecture wit X86_64
and according to RPMfusion:
sudo qubes-dom0-update akmod-nvidia # rhel/centos users can use kmod-nvidia instead sudo qubes-dom0-update xorg-x11-drv-nvidia-cuda #optional for cuda/nvdec/nvenc support sudo dnf update -y
Everything works perfectly without any error, here is the output:
Installed:
akmod-nvidia-3:440.44-1.fc31.x86_64
elfutils-debuginfod-client-0.178-7.fc31.x86_64
perl-IO-Socket-IP-0.39-440.fc31.noarch
perl-IO-Socket-SSL-2.066-7.fc31.noarch
perl-Mozilla-CA-20180117-6.fc31.noarch
akmods-0.5.6-24.fc31.noarch
dwz-0.12-11.fc31.x86_64
efi-srpm-macros-4-3.fc31.noarch
egl-wayland-1.1.4-1.fc31.x86_64
elfutils-0.178-7.fc31.x86_64
elfutils-libelf-devel-0.178-7.fc31.x86_64
fakeroot-1.24-2.fc31.x86_64
fakeroot-libs-1.24-2.fc31.x86_64
fpc-srpm-macros-1.2-2.fc31.noarch
gc-7.6.4-6.fc31.x86_64
gdb-minimal-8.3.50.20190824-26.fc31.x86_64
ghc-srpm-macros-1.5.0-1.fc31.noarch
gnat-srpm-macros-4-10.fc31.noarch
go-srpm-macros-3.0.8-4.fc31.noarch
grubby-8.40-36.fc31.x86_64
guile22-2.2.6-2.fc31.x86_64
kernel-devel-5.4.13-201.fc31.x86_64
kmodtool-1-37.fc31.noarch
libatomic_ops-7.6.10-2.fc31.x86_64
libglvnd-opengl-1:1.1.1-5.fc31.x86_64
libvdpau-1.3-1.fc31.x86_64
make-1:4.2.1-15.fc31.x86_64
mesa-libGLES-19.2.8-1.fc31.x86_64
mesa-vulkan-drivers-19.2.8-1.fc31.x86_64
nim-srpm-macros-3-1.fc31.noarch
nvidia-settings-3:440.44-1.fc31.x86_64
ocaml-srpm-macros-5-6.fc31.noarch
openblas-srpm-macros-2-6.fc31.noarch
patch-2.7.6-11.fc31.x86_64
perl-Data-Dumper-2.174-440.fc31.x86_64
perl-Digest-1.17-439.fc31.noarch
perl-Digest-MD5-2.55-439.fc31.x86_64
perl-Encode-4:3.02-440.fc31.x86_64
perl-File-Temp-1:0.230.900-439.fc31.noarch
perl-Getopt-Long-1:2.51-1.fc31.noarch
perl-HTTP-Tiny-0.076-439.fc31.noarch
perl-MIME-Base64-3.15-439.fc31.x86_64
perl-Net-SSLeay-1.88-3.fc31.x86_64
perl-Pod-Escapes-1:1.07-439.fc31.noarch
perl-Pod-Perldoc-3.28.01-442.fc31.noarch
perl-Pod-Simple-1:3.39-2.fc31.noarch
perl-Pod-Usage-4:1.69-439.fc31.noarch
perl-Storable-1:3.15-442.fc31.x86_64
perl-Term-ANSIColor-4.06-440.fc31.noarch
perl-Term-Cap-1.17-439.fc31.noarch
perl-Text-ParseWords-3.30-439.fc31.noarch
perl-Time-Local-2:1.280-439.fc31.noarch
perl-URI-1.76-5.fc31.noarch
perl-libnet-3.11-440.fc31.noarch
perl-podlators-1:4.12-2.fc31.noarch
perl-srpm-macros-1-33.fc31.noarch
python-srpm-macros-3-51.fc31.noarch
qt5-srpm-macros-5.13.2-1.fc31.noarch
redhat-rpm-config-142-1.fc31.noarch
rpm-build-4.15.1-1.fc31.x86_64
rpmdevtools-8.10-9.fc31.noarch
rust-srpm-macros-13-1.fc31.noarch
vulkan-loader-1.1.114.0-1.fc31.x86_64
xemacs-filesystem-21.5.34-34.20190323hgc0ed7ef9a5a1.fc31.noarch
xorg-x11-drv-nvidia-3:440.44-2.fc31.x86_64
xorg-x11-drv-nvidia-cuda-libs-3:440.44-2.fc31.x86_64
xorg-x11-drv-nvidia-kmodsrc-3:440.44-2.fc31.x86_64
xorg-x11-drv-nvidia-libs-3:440.44-2.fc31.x86_64
zlib-devel-1.2.11-20.fc31.x86_64
zstd-1.4.4-1.fc31.x86_64
Complete! System up to date: sudo qubes-dom0-update
this is how it looks ( /etc/default/grub file:) GRUB_TIMEOUT=5 GRUB_DISTRIBUTOR="$(sed 's, release .*$,,g' /etc/system-release)" GRUB_DEFAULT=saved GRUB_DISABLE_SUBMENU=true GRUB_TERMINAL_OUTPUT="gfxterm" GRUB_CMDLINE_LINUX="rd.driver.blacklist=nouveau modprobe.blacklist=nouveau nvidia-drm.modeset=1 rd.lvm.lv=qubes_dom0/root rd.lvm.lv=qubes_dom0/swap plymouth.ignore-serial-consoles nomodeset rhgb quiet" GRUB_DISABLE_RECOVERY="true" GRUB_THEME="/boot/grub2/themes/qubes/theme.txt" GRUB_CMDLINE_XEN_DEFAULT="console=none dom0_mem=min:1024M dom0_mem=max:4096M iommu=no-igfx ucode=scan smt=off" GRUB_DISABLE_OS_PROBER="true" GRUB_ENABLE_BLSCFG=true
and regenerated grub configuration:
grub2-mkconfig -o /boot/efi/EFI/qubes/grub.cfg with output:
Generating grub configuration file ... Found theme: /boot/grub2/themes/qubes/theme.txt Found linux image: /boot/vmlinuz-5.4.5-1.qubes.x86_64 Found initrd image: /boot/initramfs-5.4.5-1.qubes.x86_64.img Found linux image: /boot/vmlinuz-5.4.5-1.qubes.x86_64 Found initrd image: /boot/initramfs-5.4.5-1.qubes.x86_64.img done
I will restart it now. Wish me luck
It crashed after the video card is loaded(not 100% because the processes are working but the screen doesn't display anything) I've erased nomodeset and regenerated the grub. No luck at this point. I will try to install proprietary driver instead
Hi there, any update?
Did you get it to work?
you don't need to install nvidia anymore. Just upgrade to kernel-latest(5.6...) from current-testing repo. Install qubes normally and once installation is done: dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-latest -y also don't forget to exclude kernel from next updates dom0: sudo qubes-dom0-update --exclude=kernel -y
@ol7278: Why don't you need to install nvidia anymore? I've been following this and had similar issues, and never got it working. I know Qubes does work without the nvidia proprietary drivers, but mine randomly crashes when I don't disable nouveau. Would much prefer to have the nvidia working rather than the onboard Intel video. Even watching videos can be patchy otherwise.
@qdhaf2hy Have you tried recent kernel-latest
as @ol7278 suggested? I too have a testing machine with nvidia GPU that was unusable with nouveau before (hangs and crashes at boot etc), but with recent kernel it just works.
Yes, indeed, it works flawlessly even with the latest Nvidia RTX. No issues at all!
@ol7278 So.. to use my NVIDIA card (instead of built-in Intel graphics card), all I have to do is to upgrade kernel-latest
?
I can see in lspci
(among others):
00:02.0 VGA compatible controller: Intel Corporation 4th Gen Core Processor Integrated Graphic Controller (rev 06)
01:00.0 3D controller: NVIDIA Corporation GM107M [GeForce GTX 960M] (rev a2)
I take it means GeForce will be used for accelerating 3D?
@marmarek: So I've tried just the kernel-latest
, which installs 5.5.7-1 from memory, and that does seem to stop the constant NVIDIA crashes for me.
@ioleo: Not sure if Qubes is actually using the NVIDIA GPU, or whether it is using the Intel Integrated GPU, though.
@ioleo install the latest kernel from testing repository, dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-latest -y the latest kernel is 5.6.4-1 and yes, it takes nvidia
@ol7278 cool, great to know! :) off to install Gaia Sky in some qube :D
BTW. didn't need to enable testing repo, I already have 5.5.7-1
installed
Is it possible to run qubes+nvidia with 4K 120Hz monitor? I'm trying various combinations and no luck unfortunatelly. This is the setup I have:
So, at original installation, qubes is unable to render anything greater than 4K/60Hz. I tried adding custom modelines and forcing them, but without luck. So I followed @ol7278 guide and installed most recent kernel drivers, and updated dom0. After restarting, the monitor went black with "no display". I attached another one, and checked the console, it appeared that 4K/120Hz was detected properly and set up by X automatically. However I had to downgrade back to 60Hz to actually see anything. Xrandr also detects max refresh ratio properly, and shows 4K/119.91Hz. I can safely exclude monitor problem, as Windows 10 renders 4k/120Hz without any issues. I tried to check the Xorg.0.log
but didn't find anything meaningful there.
Is there another place I can check?, I'm also happy to provide more info if needed.
@ioleo install the latest kernel from testing repository, dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-latest -y the latest kernel is 5.6.4-1 and yes, it takes nvidia
When launching Qubes, I get about 2 frames per second and only a single (wrong) resolution. When following along here https://www.qubes-os.org/doc/nvidia-troubleshooting/ to get an xorg.conf.new with ShadowFB and nouveau, I can not start Xorg "No screens found".
I feel like I am getting either incomplete or contradictory instructions.
What are the steps I need to take to get Qubes into a semi-usable state with an Nvidia RTX 2080?
I have tried sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-latest -y. I get told: Package kernel-latest ...5.8.16-1... already installed, but uname -r shows 4.19.152-1
Where can I change that?
How do I get (efi? xen?) to show a boot menu like grub does instead if immediately booting into the default kernel?
Dom0: sudo dnf remove kernel-latest ...5.8.16-1... Restart Dom0: sudo qubes-dom0-update --enablerepo=qubes-dom0-current-testing kernel-latest -y. Restart
I always remove all other kernel versions previously installed after updating to latest. Dom0: sudo thunar And go to boot to see the previous kernels And you remove it with Dom0: sudo dnf remove kernel-version..
thanks @ol7278 I will try that.
In the meantime I have simply changed the default
in xen.cfg. It boots. X starts and is smoother, but it is stuck on a single resolution... so I still have more to figure out. sigh..
For the record, I am booting qubes from another disk/grub using
menuentry 'Qubes' {
search --fs-uuid XXXX-XXXX --set=root
chainloader /EFI/qubes/xen.efi
}
Update: It works now. There was a nouveau.modset=0
in the xen.cfg left over from installation. Removing that and setting the default to 5.8.16-1 solved it.
It would still be nice to choose a kernel at boot time instead of editing default in xen.cfg, but that's a minor quibble.
Sorry I don't understand it completely. I am running 2 sli nivida 980m card and like to get the nvidia driver instead of the nouveau working. And all I need to is to update to a newer kernel like 5.9.1-4 ? But then I need to install the driver and disabel nouveau, or I will it work automatically? I only don't need to install the package from rpm-fusion or? Thank you very much:)
It should work automatically
On Sat, 13 Mar 2021, 16:02 rastababy, @.***> wrote:
Sorry I don't understand it completely. I am running 2 sli nivida 980m card and like to get the nvidia driver instead of the nouveau working. And all I need to is to update to a newer kernel like 5.9.1-4 ? But then I need to install the driver and disabel nouveau, or I will it work automatically? I only don't need to install the package from rpm-fusion or? Thank you very much:)
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/QubesOS/qubes-issues/issues/4610#issuecomment-798482395, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMLZSZOFG7PPJP6RLACDNXDTDN5AFANCNFSM4GKRCKOA .
The community guide has been updated to give instructions to manually install the proprietary driver on R4.1
https://github.com/Qubes-Community/Contents/blob/master/docs/configuration/install-nvidia-driver.md
Qubes OS version:
Qubes Release 4.0 (R4.0)
Affected component(s):
dom0
Steps to reproduce the behavior:
after following the steps mentioned in https://groups.google.com/d/topic/qubes-users/v26zXkiNElg/discussion Nvidia installer failed on creating the kernel module and gives an error
Expected behavior:
https://www.qubes-os.org/doc/install-nvidia-driver/
Actual behavior:
General notes:
Related issues:
https://github.com/QubesOS/qubes-issues/issues/2526