Arch Linux User Repository

Official binaries of linux-lqx, linux-lqx-headers, and linux-lqx-docs are now available: https://wiki.archlinux.org/index.php/Unofficial_user_repositories#liquorix

Signing key import instructions: sudo pacman-key --keyserver hkps://keyserver.ubuntu.com --recv-keys 9AE4078033F8024D && sudo pacman-key --lsign-key 9AE4078033F8024D

@Anarconda, are you building with the AUR or using the binary package? I'm currently booted into the package built through CI.

Also, you can boot into another kernel and check logs from your previous boot with journalctl -b -1 (add -k for just kernel logs). If you're not using systemd, I'm assuming it'll be in a bunch of files in /var/log.

EDIT: It appears I need to restore a commit: https://aur.archlinux.org/cgit/aur.git/commit/?h=linux-lqx&id=c622099099d1b33802af6cd052ade11d8b527df0

New version is already on the way if you prefer binary build.

@damentz It doesn't boot. Even when system is accessible from another tty the messages stop after show that net is up. No display manager, just stop there.

Can ya'll try the latest release? I made a few changes:

1. Enable some missing DRBG algorithms (hash and counter block mode)
2. Enabled the latent entropy GCC plugin to force more entropy at boot
3. Re-enabled kernel debugging with BTF (required as per https://bugzilla.kernel.org/show_bug.cgi?id=218320)
4. Enabled module compression with ZSTD, switched kernel image to ZSTD compression as well

I can verify the hid_bpf and bpf-lsm errors are gone, but I don't have any systems triggering the jitterentropy error to confirm if my changes helped.

@damentz: I can confirm what @vlad1.96 posts.

┌─[13:07:43]─[root@Gaia-E]
└──> amarcord #  dmesg | grep bpf
[    0.373796] LSM: initializing lsm=capability,yama,bpf
[    0.783663] hid_bpf: error while preloading HID BPF dispatcher: -22
[    2.891418] systemd[1]: bpf-lsm: Failed to load BPF object: No such process

and I'm also having this "problem":

─[13:07:39]─[root@Gaia-E]
└──> amarcord #  dmesg | grep jitter
[    0.602768] jitterentropy: Initialization failed with host not compliant with requirements: 9

With kernel versions 6.6.8, 6.6.9 and 6.6.10 not tests with previous versions.

Thanks in advance.

@damentz Not yet, I'll be looking for a way to find the source of the problem. Now I can only share a part of the code on linux-xen and linux-lqx in common places. There are differences there.

I don't know if it's related yet, but maybe it will lead to some thoughts.

linux-zen https://pastebin.com/XNPH6WRE

linux-lqx https://pastebin.com/kvAX7Jzu

UPD

I checked on an already compiled kernel from liquorix.net (to exclude my configuration changes) and the warning persists

@glorious-yellow: Liquorix uses PDS from Project-C: https://gitlab.com/alfredchen/linux-prjc. However, the maintainer appears to be absent, so I've been maintaining necessary changes and optimizations in Zen Kernel here: https://github.com/zen-kernel/zen-kernel/commits/6.6/prjc

@vlad1.96: That's odd, do you have any kernel output that explains why it fails? The BPF configuration compared with linux-zen is for the most part identical aside from support for kprobe error injection. All the options for BPF as an LSM module and HID_BPF are enabled.

Can you tell me which patches or modified configurations of linux-lqx (in relation to linux-zen) affect bpf-lsm and hid_bpf?

For about half a year now(i've never seen this before), one warning and one error:

archlinux kernel: hid_bpf: error while preloading HID BPF dispatcher: -22

arch systemd[1]: bpf-lsm: Failed to load BPF object: No such process

There is no such thing on linux and linux-zen.

PS Thank you for your hard work

What scheduler is used in this patchset? It works very differently than the current upstream EEVDF.

It looks like that patch may have other side effects, and it was also shot down later in the thread: https://lists.freedesktop.org/archives/amd-gfx/2023-October/100334.html

I'd suggest monitoring the two open issues that are related and let me know if they come up with a complete solution:

• https://gitlab.freedesktop.org/drm/amd/-/issues/2991
• https://github.com/ROCm/ROCm/issues/2596

Can this patch be added? https://lists.freedesktop.org/archives/amd-gfx/2023-October/100298.html

Currently running clinfo or using DaVinci Resolve would lead to kernel bug; this patch would work around that for now.