Re: [PATCH v18 08/18] s390/vfio-ap: allow assignment of unavailable AP queues to mdev device

[Tony Krowiak <akrowiak@linux.ibm.com>]

On 3/8/22 05:06, Halil Pasic wrote:
> On Mon, 7 Mar 2022 18:45:45 -0500
> Tony Krowiak <akrowiak@linux.ibm.com> wrote:
>
> [..]
>>>>> s/belong to the matrix_mdev's matrix/are fields of the matrix_mdev
>>>>> object/
>>>> This is the comment I wrote:
>>>>
>>>>            /*
>>>>             * Comparing an mdev's newly updated apm/aqm with itself would
>>>>             * result in a false positive when verifying whether any APQNs
>>>>             * are shared; so, if the input apm and aqm belong to the
>>>>             * matrix_mdev's matrix, then move on to the next one.
>>>>             */
>>>>
>>>> However, I'd be happy to change it to whatever either of you want.
>>> What ain't obvious for the comment is that "belong to" actually means
>>> composition and not association. In other words, there there is no
>>> pointer/indirection involved, a pointer that would tell us what matrix
>>> does belong to what matrix_mdev, but rather the matrix is just a part
>>> of the matrix_mdev object.
>>>
>>> I don't like 'false positive' either, and whether the apm/aqm is
>>> newly updated or not is also redundant and confusing in my opinion. When
>>> we check because of inuse there is not updated whatever. IMHO the old
>>> message was better than this one.
>>>
>>> Just my opinion, if you two agree, that this is the way to go, I'm fine
>>> with that.
>>>
>>> Regards,
>>> Halil
>> Feel free to recommend the verbiage for this comment. I'm not married
>> to my comments and am open to anything that helps others to
>> understand what is going on here. It seems obvious to me, but I wrote
>> the code. Obviously, it is not so obvious based on Jason's comments,
>> so maybe someone else can compose a better comment.
> /*
> * If the input apm and aqm are fields of the matrix_mdev object,
> * then move on to the next matrix_mdev.
> */

Perfect, you write better English than me!

>
> Regards,
> Halil