Re: [PATCH] xen/arm: Prevent Dom0 to be loaded when using dom0less

[Stefano Stabellini <sstabellini@xxxxxxxxxx>]

On Mon, 8 Mar 2021, Jan Beulich wrote:
> On 08.03.2021 15:12, Julien Grall wrote:
> > On 08/03/2021 11:56, Luca Fancellu wrote:
> >> This patch prevents the dom0 to be loaded skipping its
> >> building and going forward to build domUs when the dom0
> >> kernel is not found and at least one domU is present.
> > 
> > As you are skipping dom0, the domid 0 will not be usable for another 
> > domain. I can see a few issues:
> >     1) The first domU created will now be considered as the hardware 
> > domain (see domain_create()).
> >     2)  There are still a few hardcoded use of d->domain_id == 0 in the 
> > codebase (I could spot at least on in the RTDS code).
> >     3) Not all the code seems to be able to cope with hardware_domain is 
> > NULL (although most of it looks to be only reachable by x86)?
> >     4) is_hardware_domain() will return true when passing NULL. It is 
> > not clear whether one may pass NULL here.
> > 
> > For 2), ideally this needs to be fixed. But we may also want to reserve 
> > domid 0 just for sanity.
> 
> +1 to reserving ID zero in such a case.

I agree too