On Wed, Oct 18, 2023, Xiaoyao Li wrote: > On 10/18/2023 1:28 AM, Sean Christopherson wrote: > > On Fri, Oct 13, 2023, Xiaoyao Li wrote: > > > Fix the kvm_gfn_to_hva_cache_init() to use the correct size though KVM > > > only touches fist 8 bytes. > > > > This isn't a fix. There's actually meaningful value in precisely initializing the > > cache as it guards against KVM writing into the padding, e.g. this WARN would fire: > > > > if (WARN_ON_ONCE(len + offset > ghc->len)) > > return -EINVAL; > > > > So it's a bit odd, but I would prefer to keep the current behavior of mapping only > > the first 8 bytes. > > > > Here's what I'm thinking to clean up the enabled field (compile tested only, > > haven't touched the docs other than the obvious removal): > > It looks better. > > Will you send out a formal patch yourself? or leave it to me? Your call, I don't have a preference. Just let me know which option you choose.
