On Thu, Feb 13, 2025 at 05:50:57PM +0000, Patrick Bellasi wrote:
> The "should be set identically across all processors in the system" makes me
> wondering if using the "KVM's user_return approach" proposed here is robust
> enough. Could this not lead to the bit being possibly set only on some CPU
> but not others?
That's fine, we should update that paper.
> If BpSpecReduce does not prevent training, but only the training from being
> used, should not we keep it consistently set after a guest has run, or until an
> IBPB is executed?
After talking with folks internally, you're probably right. We should slap an
IBPB before clearing. Which means, I cannot use the MSR return slots anymore.
I will have to resurrect some of the other solutions we had lined up...
Stay tuned.
Thx.
--
Regards/Gruss,
Boris.
https://people.kernel.org/tglx/notes-about-netiquette
