On Tue, 11 Feb 2020 18:55:16 +0100
Paolo Bonzini <pbonz...@redhat.com> wrote:
> Even though MSR_IA32_UCODE_REV has been available long before Linux 5.6,
> which added it to the emulated MSR list, a bug caused the microcode
> version to revert to 0x100000000 on INIT. As a result, processors other
> than the bootstrap processor would not see the host microcode revision;
> some Windows version complain loudly about this and crash with a
> fairly explicit MICROCODE REVISION MISMATCH error.
>
> [If running 5.6 prereleases, the kernel fix "KVM: x86: do not reset
> microcode version on INIT or RESET" should also be applied.]
>
> Reported-by: Alex Williamson <alex.william...@redhat.com>
> Signed-off-by: Paolo Bonzini <pbonz...@redhat.com>
> ---
> target/i386/kvm.c | 7 +++++--
> 1 file changed, 5 insertions(+), 2 deletions(-)
Fixed! Thanks,
Alex
> diff --git a/target/i386/kvm.c b/target/i386/kvm.c
> index 6ef291d580..69eb43d796 100644
> --- a/target/i386/kvm.c
> +++ b/target/i386/kvm.c
> @@ -105,6 +105,7 @@ static bool has_msr_smi_count;
> static bool has_msr_arch_capabs;
> static bool has_msr_core_capabs;
> static bool has_msr_vmx_vmfunc;
> +static bool has_msr_ucode_rev;
>
> static uint32_t has_architectural_pmu_version;
> static uint32_t num_architectural_pmu_gp_counters;
> @@ -2056,6 +2057,9 @@ static int kvm_get_supported_msrs(KVMState *s)
> case MSR_IA32_VMX_VMFUNC:
> has_msr_vmx_vmfunc = true;
> break;
> + case MSR_IA32_UCODE_REV:
> + has_msr_ucode_rev = true;
> + break;
> }
> }
> }
> @@ -2696,8 +2700,7 @@ static void kvm_init_msrs(X86CPU *cpu)
> env->features[FEAT_CORE_CAPABILITY]);
> }
>
> - if (kvm_arch_get_supported_msr_feature(kvm_state,
> - MSR_IA32_UCODE_REV)) {
> + if (has_msr_ucode_rev) {
> kvm_msr_entry_add(cpu, MSR_IA32_UCODE_REV, cpu->ucode_rev);
> }
>
