On Thu, Jul 01, 2021 at 03:08:22PM +0100, Will Deacon wrote: > On Tue, Jun 15, 2021 at 02:39:46PM +0100, Fuad Tabba wrote: > > Add trap handlers for protected VMs. These are mainly for Sys64 > > and debug traps. > > I think one big thing missing from this patch is some rationale around > which features are advertised and which are not. Further, when traps > are enabled later on, there doesn't seem to be one place which drives the > logic, so it's quite hard to reason about. > > So I think we need both some documentation to describe the architectural > environment provided to protected VMs, but also a way to couple the logic > which says "We hide this feature from the ID registers because of foo" > with the logic that says "And here is the control bit to trap this feature". I think it would be better to have documentation that says "We expose this feature because foo". If the feature doesn't have any rationale to be exposed, then it's just not. It may also make sense to document features that should never be exposed, if there are features like that, in order to ensure nobody comes along and exposes them later. Thanks, drew _______________________________________________ kvmarm mailing list kvmarm@xxxxxxxxxxxxxxxxxxxxx https://lists.cs.columbia.edu/mailman/listinfo/kvmarm
