Search results

I am running Kernel Version Linux 6.8.12-2-pve (2024-09-05T10:03Z) I am going to remove CUDA. I am not using that on the host.

I guess I need to remove the old kernel then :( as I think the repos are configured correctly They are yes? Anyway haha I am such a newb :( How to remove the old kernel? :) I am not sure I dare to do that as I am known to break systems all too well :(

interesting point. I am running PVE 8.2.7. So I am not sure why I got these updates :( I am unsloving this thread and changing the title to reflect the following question. How do I make sure I am using the latest kernel?

after the cups incident I would like to remove as much unneeded stuff as possible. Just now I saw the following (security) updates available Do I need those or are they save to delete? I am running non free drivers though in case that matters.

marked the thread as solved again. it turns out that my laptop running proxmox 8.2.2 did have secure boot enabled yet somehow the boot process states "booting in insecure mode" Also going forward I will try forcing each new installation of proxmox to use grub rather than systemd-boot

Sorry guys to unsolve and necro this thread but I am confused again. My secure boot is disabled and am using UEFI. Re-executing '/usr/sbin/proxmox-boot-tool' in new private mount namespace.. System currently booted with uefi EDAE-F8E1 is configured with: grub (versions: 6.8.4-3-pve...

I recently installed PVE 8.2 from the new iso available. And as per usual I used a zfs mirror to install on. However this time around the boot loader seems to be grub rather than systemd boot. Last I was aware of is that when using zfs the installer defaults to systemd boot and not grub...

something like this is what you want to see on the host when you run dmesg | grep amdgpu [ 5.084765] amdgpu 0000:06:00.0: enabling device (0006 -> 0007) [ 5.087448] amdgpu 0000:06:00.0: amdgpu: Fetched VBIOS from VFCT [ 5.087458] amdgpu: ATOM BIOS: 113-CEZANNE-017 ATOM BIOS: ***** And...

none of it during your reinstall and getting the amdgpu driver to load correctly on the host. once you have that up and running then it becomes time to think about passing it through

in your case it is best to start over. as disabling IOMMU is not required in order to not blacklist or load vfio drivers. So better use a clean slate and document well what you are doing or use a tutorial from proxmox it self like https://pve.proxmox.com/wiki/PCI(e)_Passthrough

yes

I can't make that call for you :) I for one am starting over again pretty soon. But I do not know how much you have already tweaked your system.

I am unaware of any tutorial you have used. but it does not matter. just undo anything you did and see if you can get the amdgpu driver to load normally.

dont run any vm that has that igpu passed through. remove any blacklists remove the vfio-driver loading for the igpu. just like in the tutorials but then in reverse. and then show the output of dmesg | grep amdgpu you will find the name of the vbios file being used if the motherboard supplies...

My advice would be is to first not blacklist it and not pass it through and also not load vfio drivers at any stage. let your host system figure out which vbios file to use. or do you get greeted with an error message that a VBIOS file could not be found in your host's dmesg? I have had that...

although it threw an error this mount -t evivarfs I was able to run the proxmox-boot-tool now

thank you leesteken. it did hint me more into the right direction. I was able to use the proxmox iso debug installer which I exited to enter a shell then I was able to update the /etc/kernel/cmdline using the steps found below...

I run the latest proxmox with root on zfs I locked myself out of the system by having added too many kernel cmdline parameters (/etc/kernel/cmdline) related to the disabling of the display. So I tried using the rescue boot option on the most recent pve iso but it errors out stating it can not...

I see you are still blacklisting the amdgpu driver? or perhaps not blacklisting it but having the vfio-pci being used for your igpu instead. the idea is to have it load normally to be able to see in the dmesg which BIOS rom file is loaded as supplied by the firmware on the motherboard. But...

ahh well you know the age old saying. give a man enough to drink and he will start reading up out of sheer boredom :) So in order to work from the top down you indeed need to revert a few steps back in order for me to be able to help out. If things never worked on the host then I am not sure I...