On Wed, 2020-05-20 at 22:42 +0200, Paolo Bonzini wrote: > On 20/05/20 21:24, Maxim Levitsky wrote: > > Patch 24 doesn't apply cleanly on top of kvm/queue, I appplied it manually, > > due to missing KVM_STATE_NESTED_MTF_PENDING bit > > > > Also patch 22 needes ALIGN_UP which is not on mainline. > > Probably in linux-next? > > Just replace it with ALIGN. (I tested it with memzero_user in > arch/x86/kvm/ for convenience, and the lib/ patch ended up out of sync > with the actual code). That is exactly what I did. > > > With these fixes, I don't see #DE exceptions on a nested guest I try to run > > however it still hangs, right around the time it tries to access PS/2 keyboard/mouse. > > IIRC you said that the bug appeared with the vintr rework, and then went > from hang to #DE and now back to hang? And the hang is reported by L2, > not L1? Yes, and now the hang appears to be deterministic. The initial hang could happen randomally. I remember that once the nested guest got to getty prompt even and hanged on shutdown. Now hang is deterministic when kernel prints something about PS/2. I will re-check this tomorrow. > > In order to debug the hang, a good start would be to understand if it > also happens with vgif=0. This is because with vgif=1 we use VINTR > intercepts even while GIF=0, so the whole thing is a bit more complicated. I will check this tomorrow as well. Best regards, Maxim Levitsky > > Paolo >
