Re: [PATCH v2 3/5] vtd: Collapse x86 subdirectory

[Jan Beulich <jbeulich@xxxxxxxx>]

On 23.10.2025 14:23, Teddy Astie wrote:
> Le 23/10/2025 à 12:20, Jan Beulich a écrit :
>> On 23.10.2025 11:12, Teddy Astie wrote:
>>> The x86 subdirectory was originally meant to split x86-specific logic from
>>> general VT-d code. Today, Xen only supports VT-d on x86 and the code 
>>> actually
>>> implemented in this subdirectory isn't x86-specific (e.g ATS code is not x86
>>> specific).
>>>
>>> As ats.c is the only remaining file in this directory, and isn't x86 
>>> specific,
>>> move it up to vtd directory and remove the now empty x86 subdirectory.
>>>
>>> Signed-off-by: Teddy Astie <teddy.astie@xxxxxxxxxx>
>>> ---
>>> v2:
>>>   - reviewed commit description
>>>
>>>   xen/drivers/passthrough/vtd/Makefile        |  3 +--
>>>   xen/drivers/passthrough/vtd/{x86 => }/ats.c | 10 +++++-----
>>>   xen/drivers/passthrough/vtd/x86/Makefile    |  1 -
>>>   3 files changed, 6 insertions(+), 8 deletions(-)
>>>   rename xen/drivers/passthrough/vtd/{x86 => }/ats.c (97%)
>>>   delete mode 100644 xen/drivers/passthrough/vtd/x86/Makefile
>>
>> I should have remembered this when looking at v1 already: While some of your
>> other changes go farther (and hence are likely preferable), this one I'd 
>> rather
>> not see go in. I'd favor respective pieces of what's still pending from
>> https://lists.xen.org/archives/html/xen-devel/2024-02/msg00983.html. First 
>> and
>> foremost because ATS really is not only not x86-specific - it's not VT-d-
>> specific either.
> 
> CC-ing Jason Andryuk now reviewer of AMD-Vi code and likely interested 
> with https://lists.xen.org/archives/html/xen-devel/2024-02/msg00983.html 
> at least for the AMD part.
> 
> I wasn't aware of this pending series. And some of my patches overlap 
> with some of these ones, so I guess we need to complete it and then I 
> rebase mine on top of it.

It's going to be an intermix, I think. Your map/unmap changes are preferable
to mine, for example.

Jan