XIO

xWire

Posts from every feed QUEST has found across the web — blogs, news, podcasts, channels and more.

Fedora Discussion - Latest posts

Latest posts

"Failed to allocate NvKmsKapiDevice" with latest 7.3.4 kernel

Thank you for updating that info. This sounds as if the nvidia driver for the 7.3.4 kernel may have been corrupted and thus not loading while booting with that kernel. Try running sudo akmods --rebuild --force and after that has completed successfully reboot. The module should then load properly It can be verified with lsmod | grep nvidia and if that returns several lines with nvidia it would confirm the nvidia driver is properly loaded. On my f45 system with an RTX 3050 GPU and 7.3.4 kernel I…

SELinux doesn't allow QEMU/KVM to use 3D acceleration of Nvidia properly

Rostislav Krasny: This is very strange limitation. What is the root or the main cause of it? The hypervisor, under which the VM operates, is not capable of directly interacting with the nvidia GPU used by the host. The VM only sees a virtual GPU which is layered over the actual gpu driver managed by the host OS. The kernel of the VM does not actually see the physical GPU used by the host.

Boot drive stuck in read-only mode

maeyo: [ 14.675163] BTRFS error (device nvme1n1p3): incorrect extent count for 62307434496; counted 8392, expected 8391 [ 76.196176] BTRFS error (device nvme1n1p3 state EA): failed to run delalloc range, root=257 ino=6075503 folio=0 submit_bitmap=0 start=0 len=4096: -EROFS These btrfs errors would explain why the file system is mounted read only. It does so to prevent additional errors. Those errors could be a result of the memory hardware issue. I would suggest. run memtest86+ for at least…

Local variable doesn't work anymore?

John A. Ward: $ MBSID=‘MBS’+$(date %Y%M%D%Y%H%M%S) That probably would work if it were MBSID=MBS$(date +%Y%M%D%Y%H%M%S) You could verify it with MBSID=MBS$(date +%Y%M%D%Y%H%M%S) && echo $MBSID I doubt that the actual formatting seen is what you expected. Check the info in man date to see why it does not appear as desired. You probably would want date +%Y%m%d%H%M%S

SELinux doesn't allow QEMU/KVM to use 3D acceleration of Nvidia properly

Jeff V: Do you have 2 physical GPUs? No, I have only one GPU - Nvidia Quadro P1000 Jeff V: As I understand it, the only way to properly get hardware acceleration on an nvidia GPU is to have the OS using that card to have the nvidia driver running. What this means is that the VM can only load a driver for that GPU if the user has passed the GPU through to the VM and thus the host can no longer access that GPU. The net result is that there must be 2 different physical GPUs installed with one used…

"Failed to allocate NvKmsKapiDevice" with latest 7.3.4 kernel

Lars Hansson: nvidia-smi output for 7.3.3 (abbreviated) | NVIDIA-SMI 620.70.06 KMD Version: 620.70.06 CUDA UMD Version: 13.5 | | 0 NVIDIA GeForce RTX 3050 … Off | 00000000:01:00.0 Off | N/A | | N/A 38C P8 3W / 80W | 25MiB / 4096MiB | 0% Default | Please, when posting command output post the full output and then do so as preformatted text. It would look like this ``` your pasted text here and formatted as seen on screen ``` You may either enter the 3 backticks on the line before and on the line…

SELinux doesn't allow QEMU/KVM to use 3D acceleration of Nvidia properly

Do you have 2 physical GPUs? As I understand it, the only way to properly get hardware acceleration on an nvidia GPU is to have the OS using that card to have the nvidia driver running. What this means is that the VM can only load a driver for that GPU if the user has passed the GPU through to the VM and thus the host can no longer access that GPU. The net result is that there must be 2 different physical GPUs installed with one used by the host and the other used by the VM. If the nvidia GPU…

SELinux doesn't allow QEMU/KVM to use 3D acceleration of Nvidia properly

UPDATE: I finally managed to enable 3D acceleration. I did it by changing the default cgroup_device_acl in /etc/libvirt/qemu.conf into following: cgroup_device_acl = [ "/dev/null", "/dev/full", "/dev/zero", "/dev/random", "/dev/urandom", "/dev/ptmx", "/dev/userfaultfd", "/dev/kvm", "/dev/dri/renderD128", "/dev/nvidiactl", "/dev/nvidia0", "/dev/nvidia-modeset", "/dev/nvidia-uvm", "/dev/nvidia-uvm-tools" ] But there are still issues with both 3D acceleration and SELinux. The SELinux spams me with…

"Failed to allocate NvKmsKapiDevice" with latest 7.3.4 kernel

the nivida driver no longer loads with the latest 7.3.4 kernel. The error in dmesg is: [drm:nv_drm_dev_load [nvidia_drm]] *ERROR* [nvidia-drm] [GPU ID 0x00000100] Failed to allocate NvKmsKapiDevice The nvidia driver is 620.70.06 and it works just fine with the 7.3.3 kernel. nvidia-smi output for 7.3.4: No devices were found nvidia-smi output for 7.3.3: +-----------------------------------------------------------------------------------------+ | NVIDIA-SMI 620.70.06 KMD Version: 620.70.06 CUDA…

Apps launching in unpredictable locations

A weird problem recently cropped up. I’ve been running Fedora 44 KDE daily, and after a system upgrade a couple weeks ago (about kernel 7.11 or 7.13 ??), apps started launching in random and unpredictable locations. System Settings|Window Management|Window Behavior|Movement & Placement have been set on ‘Centered’ for months, and apps usually launched in the middle of the screen (with a couple exceptions). But now they launch at random locations, and often mostly off-screen, with only just…

Fails to wakeup from sleep and randomly reboots after upgrading to F45

Only the monitor. With the machine itself not set to sleep, the monitor is turned off after 10 minutes and everything else keeps running, so the resets on 7.3.3 happened while the system was awake. If it keeps running a few more days on 7.3.4 without a reset, it was most likely a regression in 7.3.3 that 7.3.4 fixed. Drafted with AI, reviewed by me.

Looking for sponsor / review: yara-x (v1.23.0, Rust non-crate, RHBZ#2574288)

Hi everyone! I am looking for a Fedora packager sponsor (FE-NEEDSPONSOR) and reviewer for yara-x 1.23.0, VirusTotal’s official next-generation YARA implementation written in Rust. Links Package Review: 2574288 – Review Request: yara-x - malware identification and classification tool Copr Build (fedora-rawhide-x86_64 & aarch64 green): Making sure you're not a bot! Spec URL: https://download.copr.fedorainfracloud.org/results/jramirez/yara-x/fedora-rawhide-x86_64/11342732-yara-x/yara-x.spec SRPM…

How do I clear the update cache in Fedora 43?

Jaimie Blackstone: seems no REAL way to get Fedora from checking and building this cache… While it’s kind of like using a sledgehammer to stick a thumbtack into a corkboard, I presume it should work. Testing on F45 seems to confirm it: Open a terminal’ Execute sudo dnf remove gnome-software and follow the prompts; Use dnf from the terminal/command line if/when you wish to check for/apply updates; If you check for updates and do not wish to apply them, use dnf clean all to clear the cache. It’s…

Broken dependencies F44. ffmpeg and libavcodec

Villy Kruse: Some user had to learn the hard way that editing the .conf file won’t survive the upgrade to Fedora 45. That’s not strictly true. In Fedora 45, editing files in the /usr/share/dnf5/repos.d directory will absolutely work – unless you also use dnf config-manager <enable|disable> <repoid> Instead of modifying the repo file, dnf config-manager adds an entry to /etc/dnf/repos.override.d/99-config_manager.repo instead. I’d love to hear why someone thought it would be a good idea to have…

Call for Fedora 46 Test Days

The Fedora 46 schedule [1] is winding up and it’s time we started thinking about what we’d want to have test days for. There are several changes accepted already for F45 and the window for proposals is still open so more may come. You can find the list of accepted Changes here: https://fedoraproject.org/wiki/Releases/46/ChangeSet Please take some time and look through the list and see if there’s anything you’d be interested in testing or if there’s something you think should get some testing…