Comment 13 for bug 2068738

Revision history for this message
Matthew Ruffell (mruffell) wrote : Re: Kernel update 5.15.0-112 might cause severe problems with specific AMD GPUs

Hi everyone,

Mario, that does indeed look like the relevant fix on the stable mailing list.
It reverts this commit:

commit 3c7e53c0d4b43ffe6e7715414b5f2b3177881ecd ubuntu-jammy
Author: Yifan Zhang <email address hidden>
Date: Tue Sep 28 15:42:35 2021 +0800
Subject: drm/amdgpu: init iommu after amdkfd device init
Link: https://git.launchpad.net/~ubuntu-kernel/ubuntu/+source/linux/+git/jammy/commit/?id=3c7e53c0d4b43ffe6e7715414b5f2b3177881ecd

It arrived in 5.15.150 stable, applied in 5.15.0-112-generic.

Interestingly enough, this was once applied during the 5.15 development cycle
in 2021, in 5.15-rc5:

commit 714d9e4574d54596973ee3b0624ee4a16264d700
Author: Yifan Zhang <email address hidden>
Date: Tue Sep 28 15:42:35 2021 +0800
Subject: drm/amdgpu: init iommu after amdkfd device init
Link: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=714d9e4574d54596973ee3b0624ee4a16264d700

Bizarre. It seems to have been removed back then as well.

Anyway, it seems it is on track to be once again reverted in the stable mailing list thread that Mario linked.

https://<email address hidden>/

To make sure this is the commit that is causing you issue, I have built you test kernels with the above revert commit applied.

There are builds for 22.04 and 20.04 HWE.

I only just uploaded them, so give them 3 hours from this message to build before trying to install. You can also check build status here:
https://launchpad.net/~mruffell/+archive/ubuntu/lp2068738-test

Please note this package is NOT SUPPORTED by Canonical, and is for TESTING
PURPOSES ONLY. ONLY Install in a dedicated test environment.

Instructions to Install (On a focal or jammy system):
1) sudo add-apt-repository ppa:mruffell/lp2068738-test
2) sudo apt update
3) sudo apt install linux-image-unsigned-5.15.0-112-generic linux-modules-5.15.0-112-generic linux-modules-extra-5.15.0-112-generic linux-headers-5.15.0-112-generic
4) sudo reboot
5) uname -rv
Look for "5.15.0-112.122+TEST2068738v20240610b1".

Can you boot into this kernel and let me know if it fixes the problem? If it does, we will chime in upstream, and get this included, and we will land it in the next Ubuntu kernel release.

Just give the kernels about 3 hours to build first.

Thanks,
Matthew