Re: [PATCH 3/4] powerpc/powernv: remove dead NPU DMA code

From: Alexey Kardashevskiy
Date: Thu Jun 20 2019 - 02:25:28 EST

On 20/06/2019 16:03, Christoph Hellwig wrote:
> Hi Linus,
> this goes back to the discussion at last years kernel summit, where
> we had the discussion on removing code never used by any in-kernel
> user an no prospects of one. The IBM folks are unfortunately still
> dragging their feet on the powerpc side. Can we revise this discussion?
> The use case here is a IBM specific bus for which they only have an
> out of tree driver that their partner doesn't want to submit for mainline,
> but keep insisting on keeping the code around (which is also built
> uncondÑtionally for the platform).

I personally keep insisting on correct commit logs, i.e. not calling
working code dead and providing actual reasons for the change. Thanks,

> I hope we had settled that argument back then, but it seems like Big
> Blue insists they are special.
> On Thu, Jun 20, 2019 at 11:45:42AM +1000, Alexey Kardashevskiy wrote:
>> On 19/06/2019 17:28, Christoph Hellwig wrote:
>>> On Wed, Jun 19, 2019 at 10:34:54AM +1000, Alexey Kardashevskiy wrote:
>>>> On 23/05/2019 17:49, Christoph Hellwig wrote:
>>>>> None of these routines were ever used since they were added to the
>>>>> kernel.
>>>> It is still being used exactly in the way as it was explained before in
>>>> previous respins. Thanks.
>>> Please point to the in-kernel user, because that is the only relevant
>>> one. This is not just my opinion but we had a clear discussion on that
>>> at least years kernel summit.
>> There is no in-kernel user which still does not mean that the code is
>> dead. If it is irrelevant - put this to the commit log instead of saying
>> it is dead; also if there was a clear outcome from that discussion, then
>> please point me to that, I do not get to attend these discussions. Thanks,