Leniwcowaty
Leniwcowaty
UBUniversal Blue
Created by Leniwcowaty on 3/23/2025 in #🛟bazzite-help
Massive slowdowns/lags in plasma, potentially related to amdgpu, no errors, no logs, no indicators
For 3 months now I've been struggling with a problem, where sometimes my NIC would drop connection. I swapped to a PCIe NIC (actually I tested 3 other NICs), but the only thing that changed was that instead of dropping connection, the whole system would slow down, have massive lags/stutters and in terms of connection, the ping response times would skyrocket. Tried changing NIC buffers, no luck. THIS IS NOT NETWORKING ISSUE. This was happening on Mint 22.1, Fedora 41, Fedora 42 Beta, Aurora and now Bazzite. I absolutely ruled out any problems with my network - all cables were changed, tested 3 routers, 2 switches, tested the same cable with 4 other devices (2 laptops, SteamDeck and MacBook). THIS IS NOT NETWORKING ISSUE. I've tested unplugging all USB devices, messing around with IOMMU, power states in BIOS, tried disabling problematic modules (r8169, wl and mt7921e). No luck. The only thing that persists now is my GPU. I've noticed that during slowdowns, a kworker/##:#-events would climb to the top of the top (hehe), while not actually hogging any resources. Followed this rabbit hole, and found: https://bbs.archlinux.org/viewtopic.php?id=299561 https://community.frame.work/t/kworker-events-unbound-stuck-at-100-cpu-laptop-slows-to-1-fps/63464 https://gitlab.freedesktop.org/drm/amd/-/issues/3647 All of these mention setting kernel module argument amdgpu.dcdebugmask=0x10, which seems to fix this issue for everyone. Tried using rpm-ostree kargs --append-if-missing="amdgpu.dcdebugmask=0x10. During boot I can see, that this is appended to LINUX_CMDLINE in Grub, and does show up in dmesg, as seen here:
[ 0.000000] Command line: BOOT_IMAGE=(hd0,gpt2)/ostree/default-c9427fe8724148d53c9a121fe5e9e6c793917c3beb6df98bfdccb7b399d9d5dc/vmlinuz-6.13.6-103.bazzite.fc41.x86_64 rhgb quiet root=UUID=b1363291-e46c-4b3c-8cf0-1c8f5ddd96f3 rootflags=subvol=root rw ostree=/ostree/boot.1/default/c9427fe8724148d53c9a121fe5e9e6c793917c3beb6df98bfdccb7b399d9d5dc/0 bluetooth.disable_ertm=1 preempt=full amdgpu.ppfeaturemask=0xfff7ffff amdgpu.dcdebugmask=0x10 amdgpu.sg_display=0
[ 0.000000] Command line: BOOT_IMAGE=(hd0,gpt2)/ostree/default-c9427fe8724148d53c9a121fe5e9e6c793917c3beb6df98bfdccb7b399d9d5dc/vmlinuz-6.13.6-103.bazzite.fc41.x86_64 rhgb quiet root=UUID=b1363291-e46c-4b3c-8cf0-1c8f5ddd96f3 rootflags=subvol=root rw ostree=/ostree/boot.1/default/c9427fe8724148d53c9a121fe5e9e6c793917c3beb6df98bfdccb7b399d9d5dc/0 bluetooth.disable_ertm=1 preempt=full amdgpu.ppfeaturemask=0xfff7ffff amdgpu.dcdebugmask=0x10 amdgpu.sg_display=0
EDIT: The dcdebugmask actually is setting properly and disabling PSR, so something other's at play... Can somebody help me?
2 replies
UBUniversal Blue
Created by Leniwcowaty on 2/20/2024 in #🛟bazzite-help
"Bad shim signature" when rebasing to Bazzite
Sooo... As far as I'm concerned, the whole idea of Universal Blue is, that you can essentially use UBlue, then if you want rebase to Bluefin, and if you feel like it, freely rebase to Bazzite. Or any other image. So I tried that - installed Bluefin from official UBlue ISO, than rebased to Bazzite stable (the default one). After rebooting, when I select Bazzite in grub, I get:
error: ../../grub-core/kern/efi/sb.c:182: bad shim signature.
error: ../../grub-core/loader/i386/efi/linux.c:258: you need to load the kernel first.
error: ../../grub-core/kern/efi/sb.c:182: bad shim signature.
error: ../../grub-core/loader/i386/efi/linux.c:258: you need to load the kernel first.
While Bluefin boots normally. Is this intended behaviour? Or I'm just stupid and it's not like I can rebase Bluefin or any other image to any different image? As for my hardware - ThinkPad E14 Gen2, i7-1165G7
27 replies