Re: [PATCH v2 1/6] KVM: x86: Add dedicated storage for guest RIP

[Sean Christopherson <seanjc@google.com>]

On Wed, Apr 15, 2026, Xiaoyao Li wrote:
> On 4/14/2026 11:37 PM, Sean Christopherson wrote:
> > On Tue, Apr 14, 2026, Chang S. Bae wrote:
> > > On 4/14/2026 5:31 AM, Xiaoyao Li wrote:
> > > > Even leave RIP in regs[], what is the problem by just allocating the
> > > > index 16-31 to R16-R31 and making RIP the index 32?
> > > 
> > > But why?
> > > 
> > > Even though the array isn't explicitly labeled as GPRs, that's effectively
> > > how it's being used, and RIP isn't part of that set.
> > > 
> > > I don't think there is any benefit of leaving it in regs[].
> > 
> > +1.  Chang's earlier argument that RIP isn't a proper GPR swayed me over, e.g. RIP
> > doesn't have an architectural index.
> > 
> > Keeping RIP in regs[] saves one line of code in arch/x86/include/asm/kvm_host.h,
> > at the cost of making the code less readable (IMO) and incorrectly suggesting that
> > RIP can be accessed like other regs[].
> 
> I'm not trying to object this patch. Instead, I'm trying to understand the
> justification of the change.
> 
> So I would expected an updated changelog with above justifications
> incorporated.

Noted, I'll expand the changelog for the next version.