Re: [pve-devel] applied: Re: [PATCH qemu] drop patch force-disabling smm

[Wolfgang Bumiller <w.bumiller@proxmox.com>]

On Tue, Aug 24, 2021 at 11:19:52AM +0200, Thomas Lamprecht wrote:
> On 24/08/2021 10:52, Stefan Reiter wrote:
> > On 8/23/21 6:01 PM, Thomas Lamprecht wrote:
> >> On 10/08/2021 15:47, alexandre derumier wrote:
> >>> Le mardi 10 août 2021 à 09:55 +0200, Wolfgang Bumiller a écrit :
> >>>> This drops debian/patches/pve/0005-PVE-Config-smm_available-
> >>>> false.patch
> >>>> (and renumbers the remaining patches)
> >>>>
> >>>>  From what I could gather, this patch was originally added
> >>>> due to issues with old kernels. Now we have users which
> >>>> seem to run into issues *with* the patch.
> >>> yes indeed, this was old kernel with some old processors.
> >>>
> >>>
> >>> Just be carefull with live migration, I wonder if it's not breaking
> >>> migration it you just remove it like that.
> >>> (Maybe it could be better to remove it for qemu > 6.1, it should be
> >>> tested)
> >>>
> >>
> >> Did anybody tested this on live migration (on a system where it would
> >> actually toggle to true)?
> > 
> > I've tested live-migration in both directions (with patch <-> current),
> > Linux and Windows guests on a nested setup (but SMM is emulated in QEMU
> > so nested shouldn't matter AFAIU). All worked without issue.
> > 
> > I'm not sure if there is something specific a guest would need to do to
> > exercise SMM support, but since we're turning it on, not off, I'm pretty
> > sure that at least forward migration should always work.
> 
> sounds sensible

I think this mostly happens on boot, so perhaps migrating to a patched
qemu, *soft*-rebooting the guest and then migrating back might be a
problem.