Re: [RFC PATCH v2 1/3] drivers/accel: define kconfig and register a new major

[Dave Airlie]

On Tue, 8 Nov 2022 at 22:28, Jason Gunthorpe <jgg@xxxxxxxxxx> wrote:
>
> On Tue, Nov 08, 2022 at 06:33:23AM +1000, Dave Airlie wrote:
>
> > At plumbers we decided a direction, I think the direction is good, if
> > there is refactoring to be done, I'd rather it was done in tree with a
> > clear direction.
> >
> > Coming in now and saying we should go down a different path isn't
> > really helpful. We need to get rolling on this, we have drivers that
> > want to land somewhere now, which means we need to just get a
> > framework in place, leveraging drm code is the way to do it.
>
> It is not a different path, at plumbers we decided accel should try to
> re-use parts of DRM that make sense. I think that should be done by
> making those DRM parts into libraries that can be re-used, not by
> trying to twist DRM into something weird.

There isn't much twisting here, the thing is this is just the code for sharing,
there isn't going to be mountains more. This code gives accel drivers access
to a lot of things. Refactoring it out will take a year or so, and I don't think
buys us anything.

>
> If this thing needs special major/minor numbers, it's own class, its
> own debufs, sysfs, etc, then it should not be abusing the DRM struct
> device infrastructure to create that very basic kernel infrastructure.
>
> Somehow we ended up with the worst of both worlds. If you want to to
> be DRM then it should just be DRM and we shouldn't see all this core
> infrastructue code for debugfs/sysfs/cdevs/etc in thes patches at all.

We can refactor this out even clearer in the long run if it needs to,
but you are overly focusing on the small picture of these patches and
not the larger sharing this enables.

At this point I'm going to be merging close to what we have here, so
we can move forward with getting some drivers lined up.

Dave.