Noah writes:
> vmm crashes during boot after upgrading a VM from Ubuntu 18 to Ubuntu 20. > Host is running 6.8 with all syspatches > > vmd -dvvv output provides a log entry of: > vcpu_run_loop: vm 7 / vcpu 0 run ioctl failed: Bad address > > and this coincides with a kernel message: > vmx_fault_page: uvm_fault returns 14, GPA=0xfe001818, rip=0xffffffffb98a31b4 It's a known issue. Newer Ubuntu kernels are built with an Intel PMC driver that probes mmio registers on (IIRC) Intel Skylake and newer systems. vmm(4) and vmd(8) do not support mmio like this at the moment. There's no way to disable the probing as it's not a kernel module, so no combination of linux boot args will help. There are some hacky patches that would work if you were on -current, but unless you want to stay on bleeding edge, it's probably a bigger hastle than it's worth. This is in my backlog, however...I have a Kaby Lake system impacted by this. -Dave