On 11.03.2024 13:40, Vaishali Thakkar wrote:
> @@ -698,11 +698,11 @@ nsvm_vcpu_vmentry(struct vcpu *v, struct cpu_user_regs
> *regs,
> /* Convert explicitely to boolean. Deals with l1 guests
> * that use flush-by-asid w/o checking the cpuid bits */
> nv->nv_flushp2m = !!ns_vmcb->tlb_control;
> - if ( svm->ns_guest_asid != ns_vmcb->_guest_asid )
> + if ( svm->ns_asid != ns_vmcb->_asid )
> {
> nv->nv_flushp2m = 1;
> hvm_asid_flush_vcpu_asid(&vcpu_nestedhvm(v).nv_n2asid);
> - svm->ns_guest_asid = ns_vmcb->_guest_asid;
> + svm->ns_asid = ns_vmcb->_asid;
> }
>
> /* nested paging for the guest */
> @@ -1046,7 +1046,7 @@ nsvm_vmcb_prepare4vmexit(struct vcpu *v, struct
> cpu_user_regs *regs)
> /* Keep it. It's maintainted by the l1 guest. */
>
> /* ASID */
> - /* ns_vmcb->_guest_asid = n2vmcb->_guest_asid; */
> + /* ns_vmcb->_asid = n2vmcb->_asid; */
Unlike in the earlier patch, where I could accept the request to switch
to using accessor functions as scope-creep-ish, here I'm pretty firm
with my request to stop their open-coding at the same time. Unless of
course there's a technical reason the accessors cannot be used here.
Jan
