Re: [PATCH v3 for-next 0/4] xen: Convert __page_to_mfn and _mfn_to_page to use typesafe MFN

[Julien Grall <julien.grall@linaro.org>]

Hi Jan,

On 09/11/17 15:47, Jan Beulich wrote:
>>>> On 09.11.17 at 16:39, <julien.grall@linaro.org> wrote:
>> On 09/11/17 15:36, Jan Beulich wrote:
>>>>>> On 09.11.17 at 16:20, <julien.grall@linaro.org> wrote:
>>>> I had a look at the files that needs to convert. It seems there are few
>>>> files with page_to_mfn/mfn_to_page re-defined but no callers:
>>>> 	- arch/x86/mm/hap/nested_hap.c
>>>> 	- arch/x86/mm/p2m-pt.c
>>>> 	- arch/x86/pv/traps.c
>>>> 	- arch/x86/pv/mm.c
>>>> 	- arch/x86/pv/iret.c
>>>>
>>>> Those can be fixed now. That leave the following files:
>>>> 	- arch/x86/mm/p2m-ept.c
>>>> 		In that file, the override prevents all the caller to use the
>>>> construction mfn_to_page(_mfn(...)) as it mostly deals with hardware.
>>>
>>> I'm not clear what you're trying to tell me here. The file has a total
>>> for four mfn_to_page() uses - if overrides don't help (and perhaps
>>> regardless of if they do), I think it wouldn't be very difficult to simply
>>> change the four places. And note that plain staging has no override
>>> there right now.
>> Because the plain staging still has page_to_mfn() not using typesafe...
>> You would need to override it even I follow your suggestion.
>>
>> What I meant is you would replace the 4 occurrences by
>> mfn_to_page(_mfn(...)). If you are happy with that, then fine.
> 
> Oh, sure, that's a fine intermediate state, which we have all over
> the place right now. It's clear that it'll take a while to fully carry
> through typesafeness to everywhere.

Would you be fine with other part of Xen too? If so, I can have a go at 
removing completely __page_to_mfn/__mfn_to_page.

Cheers,

-- 
Julien Grall

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel