Knowledge about this type of attack dates back to the 90s (possibly earlier), but it isn't entirely clear whether the engineers who developed the "protection ch

Knowledge about this type of attack dates back to the 90s (possibly earlier), but it isn't entirely clear whether the engineers who developed the "protection check after [speculative] load" were aware of that. I would argue though that "check after load" should have smelled bad to people intimiately familiar with CPU design. It should be noted though that at the time neither sharing processors with strangers / across trust boundaries nor executing arbitrary crap in a VM were common activities. Memory protection and such were mostly viewed as a technique to increase reliability, not to provide actual security.