On Fri, Sep 28, 2018 at 12:43:19PM +0200, Sebastian Andrzej Siewior wrote:
> Intel SDM says for CPUID function 0DH, sub-function 0:
>
> | • ECX enumerates the size (in bytes) required by the XSAVE instruction for
> an
> | XSAVE area containing all the user state components supported by this
> | processor.
> | • EBX enumerates the size (in bytes) required by the XSAVE instruction for
> an
> | XSAVE area containing all the user state components corresponding to bits
> | currently set in XCR0.
>
> Signed-off-by: Sebastian Andrzej Siewior <bige...@linutronix.de>
> ---
> target/i386/cpu.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index f24295e6e4bb7..00c18444c0bac 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -4178,7 +4178,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index,
> uint32_t count,
> *ecx = xsave_area_size(x86_cpu_xsave_components(cpu));
> *eax = env->features[FEAT_XSAVE_COMP_LO];
> *edx = env->features[FEAT_XSAVE_COMP_HI];
> - *ebx = *ecx;
> + *ebx = xsave_area_size(env->xcr0);
Reviewed-by: Eduardo Habkost <ehabk...@redhat.com>
For reference, KVM isn't affected by this bug because KVM kernel code updates
CPUID data at runtime, at:
int kvm_update_cpuid(struct kvm_vcpu *vcpu)
{
[...]
best = kvm_find_cpuid_entry(vcpu, 0xD, 0);
if (!best) {
[...]
} else {
[...]
best->ebx = xstate_required_size(vcpu->arch.xcr0, false);
}
[...]
}
> } else if (count == 1) {
> *eax = env->features[FEAT_XSAVE];
> } else if (count < ARRAY_SIZE(x86_ext_save_areas)) {
> --
> 2.19.0
>
--
Eduardo
