On Wednesday 19 Oct 2022 at 12:57:24 (+0100), Will Deacon wrote:
> On Tue, Oct 18, 2022 at 04:33:45PM +0000, Quentin Perret wrote:
> > On Monday 17 Oct 2022 at 12:51:56 (+0100), Will Deacon wrote:
> > > +void pkvm_hyp_vm_table_init(void *tbl)
> > > +{
> > > + WARN_ON(vm_table);
> > > + vm_table = tbl;
> > > +}
> >
> > Uh, why does this one need to be exposed outside pkvm.c ?
>
> We need to initialise the table using the memory donated by the host
> on the __pkvm_init path. That's all private to nvhe/setup.c, so rather
> than expose the raw pointers (of either the table or the donated memory),
> we've got this initialisation function instead which is invoked by
> __pkvm_init_finalise() on the deprivilege path.
>
> Happy to repaint it if you have a patch?
I don't, I just got confused, maybe because in an older version of this
(possibly quite old) the table was statically allocated? Anyways, it's
all fine as-is, thanks for the reply.
