Re: [PATCH] docs: outline some guidelines for security classification

[Peter Xu <peterx@redhat.com>]

On Mon, Jul 20, 2026 at 05:27:00PM +0100, Daniel P. Berrangé wrote:
> We should likely distinguish live migration from snapshots
> a little.
> 
> In live migration if loading aborts, we still have the source
> VM running fine.
> 
> With snapshots if you launch a new QEMU and restore from snapshot
> there was no source VM that's relevant.
> 
> With snapshots if you have a running QEMU and rollback to a
> snapshot (with loadvm), and QEMU crashes, you've lost your
> currently running QEMU.

Yes, it's a different case, but IMHO not different from triaging security
issues.

loadvm on top of live QEMU is almost saying, "quit this qemu, and start
that snapshot VM instead, without bothering re-starting another process,
because they have the same hardware layout".

So it's not the loadvm process that destroyed the "source QEMU", it's the
user decision.  That's the case with/without loadvm being successful.

Thanks,

-- 
Peter Xu