Uh... OpenBSD is susceptible to Retbleed, which doesn't involve SMT behavior, only branch prediction state on a single CPU. The very subject under discussion seems to invalidate your point. OpenBSD, like everyone else, made a call not to patch this vulnerability proactively because it didn't seem exploitable. And like the rest of us, they were wrong (a little -- Retbleed remains a very slow channel, but it's real).