Re: [PATCH v6 05/25] iommu/arm-smmu-v3: Move IDR parsing to common functions

[Jason Gunthorpe <jgg@ziepe.ca>]

> > Copying a bunch of functions into a shared .c file exactly as it is,
> > then compiling the shared file with some #ifdef'ery at is going to be
> > long term better than trying to mangle the whole thing to avoid using
> > any of the core types and not directly share the code, IMHO.
> >
> 
> There isn't #ifdef'ery at the moment, that's what I was trying to
> avoid by introducing shared code.

Yeah, I think that may be a bad direction. Ultimately it feels like it
will be more burden to maintain this careful split, while some small
list of carefully selected #defines will let you reuse alot more with
no code changes and I think that is ultimately going to be better.

> What concerns me is how fragile that is, any change in the main struct
> can easily break the hypervisor, unlike if we have a clear shared code
> and defined API that is used by 2 entities.
> I will think more about this before v7 and see how intrusive it is.

IMHO so long as it is easy to include pkvm in the compilation I see no
issue with build testing the pkvm driver when working on smmuv3
driver. So I'm not worried about this, any breaks will be compile
breaks and can be delt with.

What I'd like is to minimize logic changes and maximimize re-use so
you don't have to make bad re-implementations. Like pkvm shouldn't be
building a weaker tlbi, it shouldn't have different logic for errata
and FEAT, it shouldn't be doing STE changes without the
hitless logic, etc, etc. All these things are easier to solve with
greater direct code re-use..

Thus I feel the trade of off 'use the code with no changed via
#define' is better than 'try to carefully cut away and avoid #define'

Jason