If it's anything like HAP features then I'd suspect it was intended for debugging and very customer-specific cases. I could totally see them accounting for a situation where sharing branch prediction across execution units caused performance issues.
They're pretty encrypted. What surprised me was not adding MSRs to flush the branch predictor (I, er, predicted they'd do that), but that they have a chicken bit to isolate hyperthreading branch predictors. That suggests they vaguely thought this might happen to some extent.
-
-
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
How does isolating _hyperthreading_ branch preds help you? Ignoring OOE for a sec*, I thought Meltdown bug didn't require a separate insn stream; spec executing a single stream/thread was enough. * OOE fetches multiple insns at once in most impls, but that's different I think?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Are there new instructions, or just new MSRs? Is there any public documentation?
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The prospect of abusing these bugs have been known for a very, very long time. https://marc.info/?l=openbsd-misc&m=118296441702631&w=2 …
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.