Bryan Cantill was given my phone number by someone. He continued accusing me at https://lobste.rs/s/zwkuza/intel_cpus_might_leak_information_about
He did not phone. I'm waiting for a call, or an apology. It says a lot when someone working for a gigantic company has to posture like that. I wonder if all *HIS* illumos clients are protected from all consequences of what heise mentioned around 2 months ago. Reports said there are apparently 8 problems, this is 1. 7 to go. Or more, because I suspect there are 2-3 generic problems which have a vast number of consequences. Has Illumos deployed all fixes? Maybe that is where the anger comes from. There is a note that Bryan thinks he could have gotten Intel to allow us access.... why did Intel not listen to him. So since OpenBSD is not in the embargo, we can fix the problems as we figure them out. We are willing to over-fix problems we guess at, and if they are not real, we will back out the solutions. The situation is that our users are vulnerable. We'll do the best we can to get ahead of the game, because otherwise we find out in late August because our users don't matter compared to Illumus or FreeBSD users. Suggest non-response from Intel. So far we suspect there is a collection of TLB effects, a collection of speculative instruction effects including basically being greedy about any instruction-result address being used to feed the caches and load via the TLB. Including eviction problems. Suspect there are workarounds subject to the limitations of the micro-architure, and trying to impliment this. R.N. Ibbett and N.P. Topham of 1989 providing guidance. Though I did receive 2 copies of a paper today.... Not from Bryan though, he's busy trying to get Intel to give us access, while slandering me on forums. Can he do both things at once?
