[for-2.9,v2,2/2] i386: Don't override -cpu options on -cpu host/max

The existing code for "host" and "max" CPU models overrides every
single feature in the CPU object at realize time, even the ones
that were explicitly enabled or disabled by the user using
"feat=on" or "feat=off", while features set using +feat/-feat are
kept.

This means "-cpu host,+invtsc" works as expected, while
"-cpu host,invtsc=on" doesn't.

This was a known bug, already documented in a comment inside
x86_cpu_expand_features(). What makes this bug worse now is that
libvirt 3.0.0 and newer now use "feat=on|off" instead of
+feat/-feat when it detects a QEMU version that supports it (see
libvirt commit d47db7b16dd5422c7e487c8c8ee5b181a2f9cd66).

Change the feature property getter/setter to set a
env->user_features field, to keep track of features that were
explicitly changed using QOM properties. Then make the
max_features code not override user features when handling "-cpu
host" and "-cpu max".

This will also allow us to remove the plus_features/minus_features
hack in the future, but I plan to do that after 2.9.0 is
released.

Reported-by: Jiri Denemark <jdenemar@redhat.com>
Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
---
Changes v1 -> v2:
* Don't clear existing bits on env->features (they are already
  supposed to be all zeroes) (Igor)
* Fix typo on comment (Igor)
* Moved feature word getter/setter parameter changes to separate
  patch (Igor)
---
 target/i386/cpu.h |  2 ++
 target/i386/cpu.c | 13 +++++++++----
 2 files changed, 11 insertions(+), 4 deletions(-)

On Mon, 27 Mar 2017 11:48:15 -0300
Eduardo Habkost <ehabkost@redhat.com> wrote:

> The existing code for "host" and "max" CPU models overrides every
> single feature in the CPU object at realize time, even the ones
> that were explicitly enabled or disabled by the user using
> "feat=on" or "feat=off", while features set using +feat/-feat are
> kept.
> 
> This means "-cpu host,+invtsc" works as expected, while
> "-cpu host,invtsc=on" doesn't.
> 
> This was a known bug, already documented in a comment inside
> x86_cpu_expand_features(). What makes this bug worse now is that
> libvirt 3.0.0 and newer now use "feat=on|off" instead of
> +feat/-feat when it detects a QEMU version that supports it (see
> libvirt commit d47db7b16dd5422c7e487c8c8ee5b181a2f9cd66).
> 
> Change the feature property getter/setter to set a
> env->user_features field, to keep track of features that were
> explicitly changed using QOM properties. Then make the
> max_features code not override user features when handling "-cpu
> host" and "-cpu max".
> 
> This will also allow us to remove the plus_features/minus_features
> hack in the future, but I plan to do that after 2.9.0 is
> released.
> 
> Reported-by: Jiri Denemark <jdenemar@redhat.com>
> Signed-off-by: Eduardo Habkost <ehabkost@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>

> ---
> Changes v1 -> v2:
> * Don't clear existing bits on env->features (they are already
>   supposed to be all zeroes) (Igor)
> * Fix typo on comment (Igor)
> * Moved feature word getter/setter parameter changes to separate
>   patch (Igor)
> ---
>  target/i386/cpu.h |  2 ++
>  target/i386/cpu.c | 13 +++++++++----
>  2 files changed, 11 insertions(+), 4 deletions(-)
> 
> diff --git a/target/i386/cpu.h b/target/i386/cpu.h
> index 07401ad9fe..c4602ca80d 100644
> --- a/target/i386/cpu.h
> +++ b/target/i386/cpu.h
> @@ -1147,6 +1147,8 @@ typedef struct CPUX86State {
>      uint32_t cpuid_vendor3;
>      uint32_t cpuid_version;
>      FeatureWordArray features;
> +    /* Features that were explicitly enabled/disabled */
> +    FeatureWordArray user_features;
>      uint32_t cpuid_model[12];
>  
>      /* MTRRs */
> diff --git a/target/i386/cpu.c b/target/i386/cpu.c
> index feefa5b8a4..13c0985f11 100644
> --- a/target/i386/cpu.c
> +++ b/target/i386/cpu.c
> @@ -3373,15 +3373,19 @@ static void x86_cpu_expand_features(X86CPU *cpu, Error **errp)
>      GList *l;
>      Error *local_err = NULL;
>  
> -    /*TODO: cpu->max_features incorrectly overwrites features
> -     * set using "feat=on|off". Once we fix this, we can convert
> +    /*TODO: Now cpu->max_features doesn't overwrite features
> +     * set using QOM properties, and we can convert
>       * plus_features & minus_features to global properties
>       * inside x86_cpu_parse_featurestr() too.
>       */
>      if (cpu->max_features) {
>          for (w = 0; w < FEATURE_WORDS; w++) {
> -            env->features[w] =
> -                x86_cpu_get_supported_feature_word(w, cpu->migratable);
> +            /* Override only features that weren't set explicitly
> +             * by the user.
> +             */
> +            env->features[w] |=
> +                x86_cpu_get_supported_feature_word(w, cpu->migratable) &
> +                ~env->user_features[w];
>          }
>      }
>  
> @@ -3731,6 +3735,7 @@ static void x86_cpu_set_bit_prop(Object *obj, Visitor *v, const char *name,
>      } else {
>          cpu->env.features[fp->w] &= ~fp->mask;
>      }
> +    cpu->env.user_features[fp->w] |= fp->mask;
>  }
>  
>  static void x86_cpu_release_bit_prop(Object *obj, const char *name,