Kirill A. Korinsky <kir...@korins.ky> writes:
> On Tue, 21 May 2024 18:38:39 +0100, > Dave Voutila <d...@sisu.io> wrote: >> >> Can you reproduce this and get details on which process panics? It's not >> clear what the vm cpu usage has to do with this panic, if anything. > > I'll try. May you suggest that command / output can be useful in the case > I've reproduced the issue? If you manage to reproduce it, it would be helpful to know which process suffered the fault (show proc). Some details on the uvm system (show uvmexp) and current register states (show regs) too. Might help to know what else is scheduled on each cpu: show all procs /o > > Anyway, usually at some point, after vmctl start docker or doas reboot > inside the guest, the host starts to lag and in the top I see ~30% CPU usage > by Xorg and some chrome's proccesses. Load average was 6 if I recall right. > > Switching to Chrome requires significant amount of time (couple of minutes), > and open its menu to shutdown requries also a lot of time, and I see how it > draws the white box for menu, and draws menu content. > > The crash had happened when I've clicked the exit from chrome, and it, I > guess, starts to saves its sate on the disk. > > Anything else, expect X11 and chrome, seems "normal". It's hard to isolate vmm/vmd issues as bugs in vmm can cause failures in other systems (uvm, vfs, etc.). vmm also has the ability to stress those systems in ways that aren't normally stressed by other programs in base. The more information, the better, because these bugs can be very tricky.