Re: TDX migration model

["Serge E. Hallyn" <serge@hallyn.com>]

On Thu, Nov 05, 2026 at 05:59:01PM +0200, Artem Bityutskiy wrote:
> [I dropped a bunch of people from CC, sorry about that, just restored the CC
> list]
> 
> On Thu, 2026-11-05 at 08:58 -0600, Serge Hallyn (AMD) wrote:
> > On Wed, Nov 04, 2026 at 03:51:45PM +0200, Artem Bityutskiy wrote:
> > > Hi Peter, community,
> > > 
> > > Here is the TDX migration model doc that I mentioned in the CoCo migration
> > > uAPI thread:
> > > 
> > > https://github.com/dedekind/dedekind-share/blob/master/2026/tdx-live-migration.md
> > > 
> > > The doc focuses on TDX only. It does not cover CoCo in general or
> > > Linux/KVM. 
> > > 
> > > The goal is to lay out what the TDX module offers in terms of ABI and
> > > migration flow today. Hopefully it is easier to read than the TDX specs,
> > > which are much more detailed. Feedback is very welcome.
> > > 
> > > The uAPIs that we suggested were modeled around this flow and ABI. But
> > > we kept general CoCo VMs in mind as much as we could.
> > > 
> > > The TDX live migration PoC implementation that we mentioned earlier follows
> > > the flows and uses the ABIs described in the doc.
> > > 
> > > Thanks!
> > > 
> > > P.S. From the ARM talk and hallway conversations at Plumbers, I understood
> > > that the ARM live migration flow is very similar to TDX. For SNP guests, we
> > > have not heard enough yet.
> > > 
> > > Thanks,
> > > Artem.
> > 
> > Thanks, Artem!
> > 
> > So, looking at the section '### How the Setup Session Works', is it safe
> > to assume that you are using the qemu migration fd for the message
> > transfers?  So basically qemu loops over KVM_MIGRATE_SETUP calls, which
> > is directed to platform-specific handlers, which in the tdx case calls
> > TDH.MIG.SETUP, and this loop continues until TDX_SUCCESS?  If so, great,
> > that mirrors pretty exactly what we were looking at for SNP.
> 
> Hi,
> 
> Yes, that's the flow. I’m not sure about the migration fd part, I believe we
> pass the setup session data via ioctl buffers. I am sure Kishen will can
> clarify.

If you mean pass via ioctl buffers from tdx->kvm-> back to qemu, which passes
it over the migration fd to the other side, then that's what I expected.
Othewise I'll be interested to the details :)

thanks,
-serge