Re: [PATCH] device-dax: don't set kobj parent during cdev init

[Dan Williams]

On Sat, Feb 11, 2017 at 10:55 AM, Dan Williams <dan.j.williams@xxxxxxxxx> wrote:
> On Sat, Feb 11, 2017 at 10:43 AM, Logan Gunthorpe <logang@xxxxxxxxxxxx> wrote:
>>
>> On 11/02/17 11:27 AM, Dan Williams wrote:
>>> Why, when the lifetime of the cdev is already correct?
>>
>> Well, it's only correct if you use the kobj parent trick which Greg is
>> arguing against. As someone reviewing/copying code that trick is
>> unclear, undocumented and it looks rather odd messing with internal
>> kobjects. Taking the explicit reference would be very clear, very
>> standard and only net one additional line.
>>
>>> See commit ba09c01d2fa8 "dax: convert to the cdev api". I used to take
>>> explicit references like you suggest, but cdev made it cleaner.
>>
>> I agree that, on the whole, that patch makes things a good deal cleaner.
>> I'm not so sure that this one small aspect is an improvement.
>>
>> In any case, it's up to you. If you'd like I can certainly submit a v2
>> patch that adds the get/put.
>
> Can we meet in the middle and just add some comments about what is going on?
>
> It's a shame to add reference counts for something that is already
> properly reference counted.

Also when using an embedded cdev how would you recommend avoiding this problem?

https://lists.01.org/pipermail/linux-nvdimm/2016-August/006562.html