Re: [PATCH v2 for 4.20? 2/3] xen/riscv: update defintion of vmap_to_mfn()

[Oleksii Kurochko <oleksii.kurochko@xxxxxxxxx>]

On 2/4/25 2:56 PM, Jan Beulich wrote:

> On 03.02.2025 14:12, Oleksii Kurochko wrote:
> > @@ -160,6 +158,18 @@ static inline struct page_info *virt_to_page(const void *v)
> >  
> >  pte_t * pt_walk(vaddr_t va, unsigned int *pte_level);
> >  
> > +static inline mfn_t vmap_to_mfn_(vaddr_t va)
> >       
> Btw., for static functions (and variables) a prefixing underscore is
> fine to use. Its identifiers that don't have file scope which shouldn't.
>     
Should it be used a single underscore prefixing or a double one?

> > +{
> > +    pte_t *entry = pt_walk(va, NULL);
> >       
> Oh, noticing the anomaly only here: Why would pt_walk() return a pointer
> to a PTE, rather than the pte_t by value? All this does is encourage
> open-coded accesses (even writes), when especially writes are supposed
> to be going through pt_update().
>     
I tried to play with forward declaration of pte_t to not introduce 
circular dependency in the previous patch. It would be really better to return
pte_t by value, I will update that.

Thanks.

~Oleksii