Hi Jan,
> On 11 Aug 2022, at 8:02 am, Jan Beulich <jbeul...@suse.com> wrote:
>
> On 10.08.2022 14:13, Andrew Cooper wrote:
>> On 09/08/2022 16:50, Jan Beulich wrote:
>>> When passed -1, the function (taking a u16) will look for segment
>>> 0xffff, which might exist. If it exists, we may find (return) the wrong
>>> device.
>>>
>>> Signed-off-by: Jan Beulich <jbeul...@suse.com>
>>> ---
>>> An alternative would be to declare that both functions cannot be called
>>> with "wildcards" anymore. The last such use went away with f591755823a7
>>> ("IOMMU/PCI: don't let domain cleanup continue when device de-assignment
>>> failed") afaict.
>>
>> The way wildcards were used before were always bogus IMO.
>>
>> I suggest we take this opportunity to remove the ability to re-introduce
>> that anti-pattern.
>
> Okay, will do that in v2. Rahul - this means there's no point anymore
> sending a v2 of your fix, as the bug will disappear as a side effect.
> I'll add you as the reporter of that bug.
Ok. I will test the patch once you sent it..
Regards,
Rahul
