New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Segfaults when booting Debian 11 installer
I'm trying to install Debian 11 from ISO via Netboot.xyz on a VPS from Wishosting (4 vCPU cores Ryzen 9 5950X, 8 GB RAM, 500 GB NVMe)
The Debian 10 (buster) installer works fine. However both the Debian 11 (bullseye) and testing (bookworm) installers crash with either a segfault or a kernel panic on boot.
Segfault:
Panic:
The commonality in all the segfaults and panics seems to be this error: Code: unable to access opcode bytes at RIP 0x.........
Has anyone seen this before with Debian 11? I'm not sure which kernel the bookworm installer currently uses, but bullseye uses 5.10. Maybe 5.10 has some incompatibilities with something on the node?
Comments
I remember there was something funky with ryzen and kernels under 5.11 but not sure if related.
I'm going to just use debian 9 for future deployments, with a different kernel that I compiled myself. Programs/libraries will be updated manually, as needed, and put into my internal repo.
@tinyweasel rereg - how does that help the OP?
It's my thoughts on the subject.
To also append on this subject, why doesn't OP try and debootstrap debian 11 (with an updated kernel) from a debian 10 live environment?
Thanks! Do you have a bug link for that?
I just installed Debian 10, upgraded to testing/bookworm which uses a 5.14 kernel, rebooted, and it worked! So it may have indeed been the issue you're referring to. I don't think the installer itself is using the newer kernel yet.
It looks like there's a backported 5.14 kernel (albeit a non Secure Boot signed one) available in
bullseye-backports: https://packages.debian.org/bullseye-backports/linux-image-5.14.0-0.bpo.2-amd64-unsigned so in theory I could try using Bullseye plus the backported kernel.That's a good idea too!
https://www.techradar.com/au/news/amd-users-should-immediately-switch-to-the-latest-linux-511-kernel
Not exactly a bug report, but it's got some info.
I came across a similar issue with this when using Proxmox on a Ryzen, it'd be a nightmare sometimes with crashes, kernel panics, etc. Turns out that there was a github with a newer pve kernel based off 5.11+ with improvements specifically for AMD.
https://github.com/fabianishere/pve-edge-kernel
I know you're a Proxmox guy too, so if you're heading down the road of installing PVE, check that out. I haven't touched Ryzen in "that kinda way" for months now, so I can't say whether Debian 11 would improve that or not from experience, but hopefully it helps you find your way to where you need to be.
Meh, virt-manager or a qemu wrapper is better, really. For LXC, I can see why; it's a real pain with the shoddy debug output, etc.
Fuck off weasel.
I tried Proxmox once, but don't actually use it. I'm not using nested virt on any of my VPSes.
Having a bad day? You might wanna stop PMSing so hard, for no reason. If my presence can anger you, then you might want to get help.
I'm not angry and I'm having a great day, thanks for asking!
I'm just letting you know that I'm onto you. Even if it seems like some of the mods and others are buying into it, I'm not.
Since it seems to trigger you a little, and I'm not going to waste more of my energy on a weasel, I'll leave you be - I won't engage you. If you've come back to "turn a new leaf", whatever, as long as the bs stopped who cares.
Welcome back, I guess?
Back on topic... I just confirmed that the backported kernel (
linux-image-5.14.0-0.bpo.2-cloud-amd64-unsigned) works fine tooSo:
Debian 10 Installer = Works
Debian 10 (4.19 kernel) = Works
Debian 11 with backported kernel (5.14) = Works
Debian 12 with stock kernel (5.14) = Works
Debian 11 Installer = Broken
Debian 12 Installer = Broken
Some time ago (many years), I've had a similar problem, one debian version installed, another did not, I was using netboot ISOs.
After a long struggle, I have managed to install and update the newer version and then started to get crashes too. It turned out it was a RAM issue, one of the sticks was faulty but very small fault which was probably only critical in one installer.
I am not saying this is the case, the error indicates CPU issue, just a thought in case you have similar issues, consider this too.
Just in case it's a specific thing for netboot.xyz, does https://github.com/bohanyang/debi work?
It was crashing during the boot of debian-installer's Linux kernel, so I don't think it's specific to netboot.xyz. The script you linked to uses the same kernel (https://github.com/bohanyang/debi/blob/master/debi.sh#L761).