Re: [PATCH] iommu/arm-smmu-v3: Make the kunit into a module

[Jason Gunthorpe <jgg@nvidia.com>]

On Thu, May 09, 2024 at 04:23:22PM +0100, Will Deacon wrote:
> On Wed, May 08, 2024 at 03:04:23PM -0300, Jason Gunthorpe wrote:
> > On Wed, May 08, 2024 at 05:53:33PM +0100, Will Deacon wrote:
> > > On Tue, May 07, 2024 at 11:33:21AM -0300, Jason Gunthorpe wrote:
> > > > On Tue, May 07, 2024 at 03:22:48PM +0100, Will Deacon wrote:
> > > > > On Tue, May 07, 2024 at 11:09:46AM -0300, Jason Gunthorpe wrote:
> > > > > > On Tue, May 07, 2024 at 02:58:17PM +0100, Will Deacon wrote:
> > > The DRM_XE one is tristate and has this interesting variant:
> > > 
> > > 	depends on ... && (m || (y && KUNIT=y))
> > 
> > Yeah, that is one of the novel ways to write the 'compatible
> > modularity but not force enabled' check
> 
> Bah, and it looks like this has its own set of issues anyway:
> 
> https://lore.kernel.org/all/20240224121528.1972719-1-arnd@kernel.org/

:\

> > I suspect alot of these just predate the EXPORT_SYMBOL_IF_KUNIT
> > infrastructure and should probably just be moved into
> > modules.. modules clearly work better with kunit's ecosystem.
> 
> Fair enough. I'll suck it up given that this seems to be the new and
> improved way of dealing with this problem. I'm just a bit paranoid about
> exporting symbols (even within a namespace), probably thanks to my time
> working on Android.

Indeed, this is the primary reason I kept Michael's original
integration as non-modular and didn't want to do this when Mostafa
first suggested it.. kunit is new to me as well.

Thanks,
Jason