Re: [PATCH v8 6/7] xen/riscv: page table handling

[oleksii.kurochko@xxxxxxxxx]

On Mon, 2024-09-30 at 17:30 +0200, Jan Beulich wrote:
> > + */
> > +static int pt_next_level(bool alloc_tbl, pte_t **table, unsigned
> > int offset)
> > +{
> > +    pte_t *entry;
> > +    mfn_t mfn;
> > +
> > +    entry = *table + offset;
> > +
> > +    if ( !pte_is_valid(*entry) )
> > +    {
> > +        if ( !alloc_tbl )
> > +            return XEN_TABLE_MAP_NONE;
> > +
> > +        if ( create_table(entry) )
> > +            return XEN_TABLE_MAP_NOMEM;
> 
> At the risk of being overly picky: You still lose -ENOMEM here. I'd
> suggest
> to make create_table() return boolean (success / fail) to eliminate
> that
> concern. Any hypothetical plan for someone to add another error code
> there
> will then hit the return type aspect, pointing out that call sites
> need
> looking at for such a change to be correct.
I agree with these changes. Let's go with this approach.

Thanks.

~ Oleksii