Re: [PATCH v1] xen/riscv: Increase XEN_VIRT_SIZE

[Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>]

On 4/4/25 8:56 AM, Jan Beulich wrote:

> On 03.04.2025 18:20, Oleksii Kurochko wrote:
> > On 4/1/25 6:04 PM, Jan Beulich wrote:
> > > On 01.04.2025 17:58, Oleksii Kurochko wrote:
> > > > On 3/31/25 6:14 PM, Jan Beulich wrote:
> > > > > On 31.03.2025 17:20, Oleksii Kurochko wrote:
> > > > > > +        _AC(XEN_VIRT_START, UL) >> vpn1_shift;
> > > > > > +    const unsigned long xen_virt_end_vpn =
> > > > > > +        xen_virt_starn_vpn + ((XEN_VIRT_SIZE >> vpn1_shift) - 1);
> > > > > > +
> > > > > >        if ((va >= DIRECTMAP_VIRT_START) &&
> > > > > >            (va <= DIRECTMAP_VIRT_END))
> > > > > >            return directmapoff_to_maddr(va - directmap_virt_start);
> > > > > >    
> > > > > > -    BUILD_BUG_ON(XEN_VIRT_SIZE != MB(2));
> > > > > > -    ASSERT((va >> (PAGETABLE_ORDER + PAGE_SHIFT)) ==
> > > > > > -           (_AC(XEN_VIRT_START, UL) >> (PAGETABLE_ORDER + PAGE_SHIFT)));
> > > > > > +    BUILD_BUG_ON(XEN_VIRT_SIZE != MB(8));
> > > > > >               
> > > > > Is it necessary to be != ? Won't > suffice?
> > > > >             
> > > > It could be just > MB(2). Or perphaps >=.
> > > > = would make the build fail, wouldn't it?
> > > >           
> > >         
> > I just realized that BUILD_BUG_ON() condition is compared to zero so actually everything what
> > will make the condition true will cause a build fail as inside it used !(condition).
> >       
> ???
>     
BUILD_BUG_ON() forces a compilation error if the given condition is true. Therefore, if the condition
XEN_VIRT_SIZE != MB(2) is changed to XEN_VIRT_SIZE > MB(2), the condition will always evaluate to true
(assuming XEN_VIRT_SIZE is greater than 2 MB), which will result in a compilation error.

~ Oleksii
> > So it seems like we have to check for XEN_VIRT_SIZE != MB(16) and change each time when XEN_VIRT_SIZE
> > is increased.
> >       
> I don't think so, but I need to first understand the point you make above.
>