Re: [PATCH v2 for 4.5] xen/arm: Add support for GICv3 for domU

[Ian Campbell <Ian.Campbell@citrix.com>]

On Wed, 2014-11-05 at 11:10 +0000, Julien Grall wrote:
> Hi Ian,
> 
> On 05/11/2014 10:24, Ian Campbell wrote:
> >> @@ -30,6 +32,39 @@ long arch_do_domctl(struct xen_domctl *domctl, struct domain *d,
> >>
> >>           return p2m_cache_flush(d, s, e);
> >>       }
> >> +    case XEN_DOMCTL_arm_configure_domain:
> >> +    {
> >> +        uint8_t gic_version;
> >> +
> >> +        /*
> >> +         * Xen 4.5: The vGIC is emulating the same version of the
> >
> > No need to say "Xen 4.5" here, this comment is true until it is removed.
> > You could say "Currently the vGIC is..." or something if you wanted to
> > indicate that this is temporary.
> 
> Just saying temporary doesn't say anything about when this has been 
> added and will be removed.
> 
> "Xen 4.5" gives explicitly the time and show that we want to remove soon.

Neither of which things belong in the code comment IMHO. Either this
happens soon so it is moot, or it doesn't and we have a strange
misleading comment...

> >
> >> +         * hardware GIC. Only the value XEN_DOMCTL_CONFIG_GIC_DEFAULT
> >> +         * is allowed. The DOMCTL will return the actual version of the
> >> +         * GIC.
> >> +         */
> >> +        if ( domctl->u.configuredomain.gic_version != XEN_DOMCTL_CONFIG_GIC_DEFAULT )
> >> +            return -EOPNOTSUPP;
> 
> > I'm also tempted to suggest that we should accept gic_version == the hw
> > value, i.e. by moping the switch below up and including
> >          && domctl->u.configuredomain.gic_version != gic_version
> > in the condition.
> 
> I consider that explicitly asking for a version of the GIC in Xen 4.5 is 
> invalid. The user should only be able to use the default GIC.
> 
> As the DOMCTL is not set in stone, 

I disagree with your reasoning, but given this I'm not inclined to keep
arguing about it.

Ian.