[Bug 206579] KVM with passthrough generates "BUG: kernel NULL pointer dereference" and crashes

[bugzilla-daemon@bugzilla.kernel.org]

https://bugzilla.kernel.org/show_bug.cgi?id=206579

--- Comment #52 from Anthony (anthonysanwo@googlemail.com) ---
(In reply to muncrief from comment #51)
> Created attachment 288207 [details]
> Latest KVM warnings information
> 
> Okay gentlemen, here is an archive with all the latest information I have
> for the continued KVM warnings. To make things as clean as possible I did a
> clean compile of kernel 5.6.0 from the torvalds git. As before the first KVM
> patch was already in the kernel, and I manually checked the source file just
> to be sure. And the second KVM patch applied cleanly. However I included
> both patches so you can check to make sure they're correct. Here's a quick
> description of the four files in the archive:
> 
> dmesg_kvm_warnings.txt - Full dmesg output after a reboot, and then starting
> and logging into the VM.
> 
> Win10-1_UEFI.xml - The VM XML.
> 
> kvm_1.patch - The first KVM patch, which is now already in the kernel
> 
> 
> kvm_2.patch - The second KVM patch that applied cleanly to the kernel.

Yh the second patch is the GA Log one which has been queued for Linux 5.7.

Good news I finally managed to reproduce the same errors both in a test VM and
my current config.

To start my investigate I first made a test VM closely matching your config
minus the PCI devices that can passthrough. After some trial and error, I
managed to find out why I don't see the errors before in my config.

I found when installing Windows the errors would show up 100% of the time just
after it finishes installing and during the reboots to finish the setup. It
also produces the warning around the time a user account has been logged in.
Outside of that, I wasn't able to find any case to produce the warning 100% of
the time. In general, it just happens from time to time when the guest is
running.

The reason I wasn't getting the warnings was that I had both CPU pinning and
-overcommit cpu-pm so when I last tested I never tested with my cpu pinning
disabled.

In general, either CPU pinning or using -overcommit cpu-pm I have found
resulted in no warnings in both my test VM and current config.

Hopefully that helps with debugging the cause.

-- 
You are receiving this mail because:
You are watching the assignee of the bug.