Re: [PATCH v3 2/4] net: qrtr: Assign unique node id for MHI endpoints

[Marcel Holtmann <marcel@holtmann.org>]

Hi Juha-Matti,

>>>> I'm still willing to submit a new version of that patchset originally by
>>>> Denis, if the kernel devs so decide that it's the way. For example, the
>>>> "radix tree of radix trees" approach could be replaced with a hash
>>>> table.
>>>> 
>>>> There was also some criticism about a layering violation in storing the
>>>> endpoint id to mhi controller. However, maybe that could be reduced by
>>>> providing a correct abstraction around it, so that other code doesn't
>>>> touch the mhi controller structure directly.
>>>> 
>>>> Other criticism for that patchset by Denis was its overt complexity.
>>>> Yes, it has more lines of code changed.
>>>> 
>>>> This dual-card setup needs to be handled somehow, either by Denis's
>>>> patchset or by Manivannan's patchset.
>>>> 
>>> 
>>> I wish to go ahead with my series as it is simple and tested across the
>>> multi-WLAN setups. For concern around exposing the nid to userspace, I'll
>>> implement it in next version, thanks!
>> 
>> actually no. Seems you have no plan on how to address the userspace problem.
>> 
>> It would be useful to actually review the endpoint_id concept introduced by Denis since it has a userspace facing API. What is your plan? Are you planning to break all existing userspace code?
> 
> Well, in defense of Manivannan's patchset it currently works for PCI+AHB
> setups better than the last version of that by Denis/Mihai/me (although 
> I could easily fix that), and it may be simpler for userspace to
> continue using nid than having to bind to an endpoint, which requires
> more changes in userspace code.
> 
> The userspace just needs to know the nid.

and existing userspace? You can _NOT_ break existing userspace.

Don’t you need to re-write all existing QRTRIPC packets to actually even make this work?

Regards

Marcel