Re: [PATCH v6 4/7] xen/arm: configure dom0less domain for enabling xenstore after boot

[Julien Grall <julien@xxxxxxx>]

Hi Stefano,

On 13/05/2022 21:52, Stefano Stabellini wrote:
> On Fri, 13 May 2022, Julien Grall wrote:
> > On 13/05/2022 02:23, Stefano Stabellini wrote:
> > > On Wed, 11 May 2022, Julien Grall wrote:
> > > > > If dom0 is mandatory yes, we should still make sure that this code
> > > > > cannot be
> > > > > reached so an ASSERT would be nice here at least in case someone tries
> > > > > to
> > > > > activate this code without dom0 (which might happen when we will push
> > > > > the
> > > > > serie for static event channels).
> > > > I am fine with an ASSERT().
> > > I added an ASSERT(hardware_domain).
> > Just to clarify and avoid a round trip. The ASSERT() is not sufficient here.
> > We also need to forbid the user to set xen,enhanced when the HW domain is not
> > NULL, at least until Rahul's pre-allocated event channel series.
> >
> > This check would have to be done when parsing the configuration. The ASSERT()
> > would be here to ensure that if someone is reworking the parsing, it would get
> > caught nicely rather than a NULL dereference.
> Thanks for avoiding a roundtrip. I added a check when parsing device
> tree if xen,enhanced is specified but dom0 is missing. Initially I wrote
> it as a "panic" but then I changed it as a regular printk. I am OK
> either way in case you prefer otherwise.
This is a configuration issue from the user. So I think we shouldn't 
continue (i.e panic()) if we can't honor what the user requested. This 
make the problem a lot more obvious to the user (printk() can be easily 
overlooked).
Cheers,

--
Julien Grall