On Wed, 2024-05-08 at 07:38 -0700, Sean Christopherson wrote: > > Ok, thanks for clarification. So it's more of a strategic thing to move more > > zapping logic into userspace so the logic can change without introducing > > kernel > > regressions. > > You're _really_ reading too much into my suggestion. As above, my suggestion > was very spur of the momemnt. I haven't put much thought into the tradeoffs > and > side effects. I'm not taking it as a mandate. Just trying to glean your insights. That said, I'm really on the fence and so leaning on your intuition as the tie breaker. For TDX's usage a struct kvm bool seems simpler code wise in KVM, and for userspace. But the zapping logic as ABI problem seems like a reasonable thing to think about while we are designing new ABI. Of course, it also means KVM has to be responsible now for safely zapping memory from a variety of userspace algorithms. So it somewhat makes KVM's job easier, and somewhat makes it harder. The real issue might be that that problem was never debugged. While there is no evidence it will affect TDXs, it remains a possibility. But we can't do the zap roots thing for TDX, so in the end the ABI design will not affect TDX exposure either way. But making it a normal feature will affect exposure for normal VMs. So we are also balancing ABI flexibility with exposure to that specific bug.
