Arch Linux User Repository

Pretty sure, delete the tar ball and re-download :p

@graysky, are you sure about b2sum checksum for 'more-uarches-20210610.tar.gz'? I downloaded it manually and checksum didn't match from PKGBUILD: 30d1df754608bb423cbc99c2097ad521baa091b9a3b39df4bd5c2d50c57eec54d8fa0e4a4a04b847c3d1b87ba682cadc8db45fabeefdc9ad7caaf8e77b96e41a tmp/kernel_compiler_patch-20210610.tar.gz

If the 'https://github.com/graysky2/kernel_compiler_patch' git repository is yours would it be feasible to use signed tags to accompany releases? In such cases signed tag would help others to simply change checksum in PKGBUILD. If it is too troublesome then don't bother with it.

According to this:

https://www.infoq.com/news/2021/01/docker-engine-cgroups-logging/

the current version of docker should work with cgroups v2. So I guess it's some interplay between the CK patchset and cgroups v2 that breaks docker. Sadly, I got no reply from CK on his blog...

@artafinde your suspect was correct. Setting the kernel parameter you suggested makes docker work again with linux-ck. I'm going to report this on CK's blog. Thanks.

I suspect your issues has something to do with cgroups v2. Try to replicate the issue after setting (temporarily) systemd.unified_cgroup_hierarchy=0 as a kernel param which forces the cgroup v1 behavior.

@snack had your issue and also a system freeze with the latest release. First time after almost a year of using CK. Went back to Zen.

@zerophase I'm not sure. For sure it's something inside the CK patchset (since the stock and many other patched kernels work fine), and it's not MuQSS (see e.g. the comment by Steven Barret on CK's blog stating that Liquorix with MuQSS works fine). I'd need to understand if there is some hope for this to be fixed, otherwise I'll have to ditch linux-ck.

@snack It has been a bug for awhile. I tried reporting it about a year ago, when trying to author a custom arch install disk, and got some explanation from a guy that this is not a CK issue. (Wasn't CK)

I cannot start docker containers when using the current (5.12) linux-ck-skylake kernel. I get this error:

$ docker run --rm -it -p 80:80 ckulka/baikal:nginx
docker: Error response from daemon: failed to create shim: OCI runtime create failed: container_linux.go:380: starting container process caused: process_linux.go:545: container init caused: process_linux.go:508: setting cgroup config for procHooks process caused: load program: invalid argument: unknown.

With the stock Arch kernel it works, and seemingly also with other patched kernels like Liquorix (although I didn't try by myself).

I know that it is usually suggested to post to CK's blog for this kind of issue, and in fact I did it but from there I got the advice to post to AUR...

graysky> GCC 11 reorders things and since GCC 11 is in [core], I updated the PKGBUILD.

That was a culprit! I updated PKGBUILD, rebuild the package (linux-ck-sandybridge) and was able to boot with linux-ck-sandybridge kernel.