Control: found -1 linux/5.16.12-1~bpo11+1 On Sun, 19 Jun 2022 02:40:22 +0000 Douglas Silva <doug.hs@proton.me> wrote: > The bug was introduced somewhere between kernel version 5.10 and 5.16, and > can be reproduced on Ubuntu and Fedora as well (upstream issue?). Debian > bullseye currently has 5.10.0-15, which works for me. The range between 5.10 and 5.16 is quite large and it would be useful to narrow that down. Via snapshot.debian.org you should be able to get older Debian Backports kernels and testing with them could reduce the range. The following version are worth trying: 5.15.5-2~bpo11+1 5.14.9-2~bpo11+1 I didn't find older versions then that in Bullseye-Backports. (Knowing which versions on Ubuntu/Fedora had the same issue, could help as well.) On Thursday, 23 June 2022 02:24:09 CEST Douglas Silva wrote: > It's not so easy to reproduce. I have been working on it all day, and it > only froze by the end of the day when I left it idle for a couple of > minutes. Maybe the 'idle' part is key to your issue? In the logs you posted (earlier), there were a few things that caught my eye: 1) Several ACPI related errors/issues 2) "DMAR: IOMMU feature XYZ inconsistent" 3) "if necessary, use "pci=nocrs" and report a bug" 4) "Lockdown: swapper/0: hibernation is restricted; see man kernel_lockdown.7" For item 1 and 2, the BIOS would be my 'primary suspect'. Try to see whether an updated BIOS improves things. Item 3 points to a kernel parameter you could try If you grep your logs for 'kernel_lockdown', you'll find quite a few lines. Make sure that's actually intended. If your issue is indeed related to 'idle', which is when usually several power management related 'things' happen, which can include (some form of) hibernation. If that is too restricted, that may give issues. And power management is also related to ACPI/BIOS. That your issue isn't visible with 5.10 _could_ be that later kernel version made issues that were (actually) present before, visible. HTH, Diederik
Attachment:
signature.asc
Description: This is a digitally signed message part.