Latest from Proxmox: with | with
⚠️ Starting with version 6.2.16-13-pve (expected around 18/Sep/2023), the Relaxable RMRR patch is now included with the Proxmox VE kernel. You don't need a patched kernel to enable the feature anymore. Just follow the configuration steps below to enable it. See Proxmox Bug 4707 for more details.
Kernel series 5.15 (Proxmox 7.4) still needs to be patched and it will continue to be released here.
This project aims to provide an easy way to build kernels for Proxmox VE with a particular set of patches.
At the moment, these are the patches applied during build:
- Relax Intel RMRR: relax
RMRRs on Intel platforms to allow certain PCIe devices to be passed through
to VMs (credit to kiler129).
This patch works around the dreaded
Device is ineligible for IOMMU domain attach due to platform RMRR requirement. Contact your platform vendor.
message when trying to passthrough certain PCIe devices to a VM.
There are two options:
-
Trigger the Build pve kernel (in container) workflow.
This workflow will build a new kernel with the current set of patches applied and produce artifacts that can be downloaded. It will run on a 2-core VM in GitHub and it will take between 2h30m and 3h to complete.
-
Build it locally
Use the build.sh script to build the kernel locally with the current set of patches applied. Because you are building everything locally, you can customize the set of patches you want before building.
In all cases, kernel builds are done using docker to contain the dependencies and make cleanup easier.
Regardless of how you built the kernel, you'll end up with a few of .deb files. To install them you need to:
- Go to the releases tab and pick appropriate packages
- Download all applicable
*.deb
s packages to your machine- If you don't use your Proxmox machine for building software, you really only need the
pve-kernel-<version>-pve-relaxablermrr_<version>_amd64.deb
file.
- If you don't use your Proxmox machine for building software, you really only need the
- Install the package(s) using
dpkg -i *.deb
in the folder where you downloaded the debs - (OPTIONAL) Verify the kernel works with the patch disabled by rebooting and checking if
uname -r
shows a version ending with-pve-relaxablermrr
- Configure the kernel
By default, after the kernel is installed, the patch will be inactive (i.e. the kernel will
behave like no patch was applied). To activate it you have to add relax_rmrr
to the intel_iommu
option on your Linux boot args.
In most distros (including Proxmox) you do this by:
-
Opening
/etc/default/grub
(e.g. usingnano /etc/default/grub
) -
Editing the
GRUB_CMDLINE_LINUX_DEFAULT
to include the option:- Example of old line:
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on iommu=pt intremap=no_x2apic_optout"
- Example of new line:
GRUB_CMDLINE_LINUX_DEFAULT="quiet intel_iommu=on,relax_rmrr iommu=pt intremap=no_x2apic_optout"
- Side note: these are actually options which will make your PCI passthrough work and do so efficiently
- Example of old line:
-
Running
update-grub
-
Rebooting
To verify if the the patch is active execute dmesg | grep 'Intel-IOMMU'
after reboot.
You should see a result similar to this:
root@sandbox:~# dmesg | grep 'Intel-IOMMU'
[ 0.050195] DMAR: Intel-IOMMU: assuming all RMRRs are relaxable. This can lead to instability or data loss
root@sandbox:~#
- kiler129: provider of the Relax Intel RMRR patch. Kiler129 provides lots of good info on the why and how the patch works.
- roforest: provides the basis for the GitHub workflows implemented here.