On Tue, Feb 28, 2023 at 07:55:09PM +0200, Zhi Wang <zhi.wang.linux@xxxxxxxxx> wrote: > On Mon, 27 Feb 2023 15:49:14 -0800 > Isaku Yamahata <isaku.yamahata@xxxxxxxxx> wrote: > > > 2) Move > > > > > > > + apic_base_msr.data = APIC_DEFAULT_PHYS_BASE | LAPIC_MODE_X2APIC; > > > > + if (kvm_vcpu_is_reset_bsp(vcpu)) > > > > + apic_base_msr.data |= MSR_IA32_APICBASE_BSP; > > > > + apic_base_msr.host_initiated = true; > > > > > > to: > > > > > > void kvm_lapic_reset(struct kvm_vcpu *vcpu, bool init_event) > > > { > > > struct kvm_lapic *apic = vcpu->arch.apic; > > > u64 msr_val; > > > int i; > > > > > > if (!init_event) { > > > msr_val = APIC_DEFAULT_PHYS_BASE | MSR_IA32_APICBASE_ENABLE; > > > > > > /* here */ > > > if (is_td_vcpu(vcpu)) > > > msr_val = xxxx; > > > if (kvm_vcpu_is_reset_bsp(vcpu)) > > > msr_val |= MSR_IA32_APICBASE_BSP; > > > kvm_lapic_set_base(vcpu, msr_val); > > > } > > > > No. Because I'm trying to contain is_td/is_td_vcpu in vmx specific and not use > > in common x86 code. > > > > I guess so. Centeralizing the initialization would be the nice and greatly > improve the readablity of the code. Maybe adding a new callback in kvm x86_ops > like .get_default_msr_val instead. Finally I can eliminate cpuid/APIC BASE MSR, and move it user space VMM, qemu. -- Isaku Yamahata <isaku.yamahata@xxxxxxxxx>