Re: [Qemu-devel] [PATCH 07/21] target-i386: convert cpuid features into properties

[Igor Mammedov <imammedo@redhat.com>]

On Thu, 16 Aug 2012 15:10:50 -0300
Eduardo Habkost <ehabkost@redhat.com> wrote:

> On Wed, Aug 15, 2012 at 06:13:27PM +0200, Igor Mammedov wrote:
> > Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> > --
> > v2:
> >   * replaced mask/ffs tricks by plain 'for (bit = 0; bit < 32; bit++)'
> >     as suggested by Eduardo Habkost
> > ---
> >  target-i386/cpu.c | 101 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  1 file changed, 101 insertions(+)
> > 
> > diff --git a/target-i386/cpu.c b/target-i386/cpu.c
> > index 37ba5ef..440e724 100644
> > --- a/target-i386/cpu.c
> > +++ b/target-i386/cpu.c
> > @@ -606,6 +606,101 @@ static int check_features_against_host(x86_def_t *guest_def)
> >      return rv;
> >  }
> >  
> > +static bool is_feature_set(const char *name, const uint32_t featbitmap,
> > +                                  const char **featureset)
> > +{
> > +    uint32_t bit;
> > +
> > +    for (bit = 0; bit < 32; ++bit) {
> > +        if (featureset[bit] && !altcmp(name, NULL, featureset[bit])) {
> > +            if (featbitmap & (1 << bit)) {
> > +                return true;
> > +            }
> > +        }
> > +    }
> > +    return false;
> > +}
> > +
> > +static void x86_cpuid_get_feature(Object *obj, Visitor *v, void *opaque,
> > +                                         const char *name, Error **errp)
> > +{
> > +    X86CPU *cpu = X86_CPU(obj);
> > +    CPUX86State *env = &cpu->env;
> > +    bool value = true;
> > +
> > +    if (!is_feature_set(name, env->cpuid_features, feature_name) &&
> > +       !is_feature_set(name, env->cpuid_ext_features, ext_feature_name) &&
> > +       !is_feature_set(name, env->cpuid_ext2_features, ext2_feature_name) &&
> > +       !is_feature_set(name, env->cpuid_ext3_features, ext3_feature_name) &&
> > +       !is_feature_set(name, env->cpuid_kvm_features, kvm_feature_name) &&
> > +       !is_feature_set(name, env->cpuid_svm_features, svm_feature_name)) {
> > +        value = false;
> > +    }
> > +
> > +    visit_type_bool(v, &value, name, errp);
> > +}
> > +
> > +static void x86_cpuid_set_feature(Object *obj, Visitor *v, void *opaque,
> > +                                         const char *name, Error **errp)
> > +{
> > +    X86CPU *cpu = X86_CPU(obj);
> > +    CPUX86State *env = &cpu->env;
> > +    uint32_t mask = 0;
> > +    uint32_t *dst_features;
> > +    bool value;
> > +
> > +    visit_type_bool(v, &value, name, errp);
> > +    if (error_is_set(errp)) {
> > +        return;
> > +    }
> > +
> > +    if (lookup_feature(&mask, name, NULL, feature_name)) {
> > +        dst_features = &env->cpuid_features;
> > +    } else if (lookup_feature(&mask, name, NULL, ext_feature_name)) {
> > +        dst_features = &env->cpuid_ext_features;
> > +    } else if (lookup_feature(&mask, name, NULL, ext2_feature_name)) {
> > +        dst_features = &env->cpuid_ext2_features;
> > +    } else if (lookup_feature(&mask, name, NULL, ext3_feature_name)) {
> > +        dst_features = &env->cpuid_ext3_features;
> > +    } else if (lookup_feature(&mask, name, NULL, kvm_feature_name)) {
> > +        dst_features = &env->cpuid_kvm_features;
> > +    } else if (lookup_feature(&mask, name, NULL, svm_feature_name)) {
> > +        dst_features = &env->cpuid_svm_features;
> > +    } else {
> > +        error_set(errp, QERR_PROPERTY_NOT_FOUND, "", name);
> > +        return;
> > +    }
> 
> Some feature names are duplicated on feature_names and
> ext_feature_names. On AMD CPU models, we have to set both, on Intel
> models we need to set the bits only on cpuid_features.
> 
> Maybe it's better to:
> 
> 1) eliminate the duplication and set the names only on feature_name
>    array;
> 2) At the end of CPU initialization, set the features on
>    cpuid_ext2_features as copies of the corresponding cpuid_features
>    bits if CPU vendor == AMD (or, maybe, if some boolean
>    "ext2_features_aliases" flag is set, to make it not
>    vendor-dependent).

So far I was trying to keep original behavior and keep scope of this series
only on moving CPU features into properties. I'll rebase this series on top
of what you proposed in a week when I'm back from vacation.
 
> 
> > +
> > +    if (value) {
> > +        *dst_features |= mask;
> > +    } else {
> > +        *dst_features &= ~mask;
> > +    }
> > +}
> > +
> > +static void x86_register_cpuid_properties(Object *obj, const char **featureset)
> > +{
> > +    uint32_t bit;
> > +
> > +    for (bit = 0; bit < 32; ++bit) {
> > +        if (featureset[bit]) {
> > +            char *feature_name, *save_ptr;
> > +            char buf[32];
> > +            if (strlen(featureset[bit]) > sizeof(buf) - 1) {
> > +                abort();
> > +            }
> > +            pstrcpy(buf, sizeof(buf), featureset[bit]);
> > +            feature_name = strtok_r(buf, "|", &save_ptr);
> > +            while (feature_name) {
> > +                object_property_add(obj, feature_name, "bool",
> > +                                x86_cpuid_get_feature,
> > +                                x86_cpuid_set_feature, NULL, NULL, NULL);
> > +                feature_name = strtok_r(NULL, "|", &save_ptr);
> 
> What happens when object_property_add() is called twice with the same
> feature name?
Just extra item in list (pure waste). Thanks for pointing out.
I've fixed it (not yet pushed) by checking if property already exists before
adding it.

> 
> > +            }
> > +        }
> > +    }
> > +}
> > +
> >  static void x86_cpuid_version_get_family(Object *obj, Visitor *v, void *opaque,
> >                                           const char *name, Error **errp)
> >  {
> > @@ -1824,6 +1919,12 @@ static void x86_cpu_initfn(Object *obj)
> >      object_property_add(obj, "tsc-frequency", "int",
> >                          x86_cpuid_get_tsc_freq,
> >                          x86_cpuid_set_tsc_freq, NULL, NULL, NULL);
> > +    x86_register_cpuid_properties(obj, feature_name);
> > +    x86_register_cpuid_properties(obj, ext_feature_name);
> > +    x86_register_cpuid_properties(obj, ext2_feature_name);
> > +    x86_register_cpuid_properties(obj, ext3_feature_name);
> > +    x86_register_cpuid_properties(obj, kvm_feature_name);
> > +    x86_register_cpuid_properties(obj, svm_feature_name);
> >  
> >      env->cpuid_apic_id = env->cpu_index;
> >  
> > -- 
> > 1.7.11.2
> > 
> > 
> 
> -- 
> Eduardo
> 


-- 
Regards,
  Igor