I compile a local kernel from the src.rpm, optimized for my hardware. Twice now since I started running 5.6 I've seen a null dereference error. I looked in bugzilla and didn't see this, though there were lots of null dereference bugzillas. The error has only occurred at the end of a dnf update, after the scriptlets are run. So, it could be an error in dnf, I suppose.
The strange thing is, it seems to be bfq causing the problem. And it only hangs one core; after a while, another core sends an nmi to the guilty core, and my system then becomes responsive for a brief time. I can switch from the virtual consoles to X, and even move around in X, but eventually the system hangs, and I have to reset it. I expect that the mess isn't being cleaned up, and eventually the guilty core is rescheduled for a task, and bam!
Is this a known issue? Is it worth opening a bugzilla?
kernel@lists.fedoraproject.org
-
stan