Re: [PATCH v3 8/9] xen/riscv: page table handling

[Jan Beulich <jbeulich@xxxxxxxx>]

On 01.08.2024 13:55, oleksii.kurochko@xxxxxxxxx wrote:
> On Thu, 2024-08-01 at 12:43 +0200, Jan Beulich wrote:
>>>>> +int map_pages_to_xen(unsigned long virt,
>>>>> +                     mfn_t mfn,
>>>>> +                     unsigned long nr_mfns,
>>>>> +                     unsigned int flags)
>>>>> +{
>>>>> +    return xen_pt_update(virt, mfn, nr_mfns, flags);
>>>>> +}
>>>>
>>>> Why this wrapping of two functions taking identical arguments?
>>> map_pages_to_xen() sounds more clear regarding the way how it
>>> should be
>>> used.
>>>
>>> xen_pt_update() will be also used inside other functions. Look at
>>> the
>>> example above.
>>
>> They could as well use map_pages_to_xen() then? Or else the wrapper
>> may
>> want to check (assert) that it is _not_ called with one of the
>> special
>> case arguments that xen_pt_update() knows how to deal with?
> Yes, map_pages_to_xen() will be used in other functions/wrappers.
> At the momemnt, I don't see what should be checked additionally in
> map_pages_to_xen(). It seems to me that xen_pt_update() covers all the
> checks at the start it needs for now. ( i will double-check that ).

I was referring to cases that xen_pt_update() can handle, but that
map_pages_to_xen() isn't supposed to be handling (if already you
want to separate both).

Jan