On Mon, May 13, 2024 at 06:53:24PM +0200, Paolo Bonzini wrote:
> On 5/13/24 17:19, Nathan Chancellor wrote:
> > > +static int snp_begin_ext_guest_req(struct kvm_vcpu *vcpu)
> > > +{
> > > + int vmm_ret = SNP_GUEST_VMM_ERR_GENERIC;
> > > + struct vcpu_svm *svm = to_svm(vcpu);
> > > + unsigned long data_npages;
> > > + sev_ret_code fw_err;
> > > + gpa_t data_gpa;
> > > +
> > > + if (!sev_snp_guest(vcpu->kvm))
> > > + goto abort_request;
> > > +
> > > + data_gpa = vcpu->arch.regs[VCPU_REGS_RAX];
> > > + data_npages = vcpu->arch.regs[VCPU_REGS_RBX];
> > > +
> > > + if (!IS_ALIGNED(data_gpa, PAGE_SIZE))
> > > + goto abort_request;
> >
> > [...]
> >
> > > +abort_request:
> > > + ghcb_set_sw_exit_info_2(svm->sev_es.ghcb, SNP_GUEST_ERR(vmm_ret, fw_err));
> > > + return 1; /* resume guest */
> > > +}
> >
> > This patch is now in -next as commit 32fde9e18b3f ("KVM: SEV: Provide
> > support for SNP_EXTENDED_GUEST_REQUEST NAE event"), where it causes a
> > clang warning (or hard error when CONFIG_WERROR is enabled) [...]
> > Seems legitimate to me. What was the intention here?
>
> Mike, I think this should just be 0?
Hi Paolo,
Yes, I was just about to submit a patch that does just that:
https://github.com/mdroth/linux/commit/df55e9c5b97542fe037f5b5293c11a49f7c658ef
Sorry for the breakage,
Mike
>
> diff --git a/arch/x86/kvm/svm/sev.c b/arch/x86/kvm/svm/sev.c
> index c7a0971149f2..affb4fb47f91 100644
> --- a/arch/x86/kvm/svm/sev.c
> +++ b/arch/x86/kvm/svm/sev.c
> @@ -3911,7 +3911,6 @@ static int snp_begin_ext_guest_req(struct kvm_vcpu *vcpu)
> int vmm_ret = SNP_GUEST_VMM_ERR_GENERIC;
> struct vcpu_svm *svm = to_svm(vcpu);
> unsigned long data_npages;
> - sev_ret_code fw_err;
> gpa_t data_gpa;
> if (!sev_snp_guest(vcpu->kvm))
> @@ -3938,7 +3937,7 @@ static int snp_begin_ext_guest_req(struct kvm_vcpu *vcpu)
> return 0; /* forward request to userspace */
> abort_request:
> - ghcb_set_sw_exit_info_2(svm->sev_es.ghcb, SNP_GUEST_ERR(vmm_ret, fw_err));
> + ghcb_set_sw_exit_info_2(svm->sev_es.ghcb, SNP_GUEST_ERR(vmm_ret, 0));
> return 1; /* resume guest */
> }
> Paolo
>
>
