On Wed, Apr 26 2023 at 08:59, Mark Rutland wrote: > On Tue, Apr 25, 2023 at 09:51:12PM +0200, Thomas Gleixner wrote: >> If not then it's just yet another way of DoS which is an "acceptable" >> attack as it only affects availability but not confidentiality. > > Sure. > > My thinking is that this is an attack against the *integrity* of the guest > (since the vCPU that gets unpasued may write to memory), and so it's > potentially more than just a DoS. > > I only mention this because I'd like to account for that on arm64, and if other > architectures also wanted to handle that it might make sense to have some > common infrastructure to track whether CPUs are potentially still within the > kernel. Fair enough.
