Re: [PATCH v4 0/4] new subsystem for compute accelerator devices

[Daniel Vetter <daniel@ffwll.ch>]

On Tue, Nov 22, 2022 at 03:46:25PM +1000, Dave Airlie wrote:
> On Tue, 22 Nov 2022 at 09:06, Sonal Santan <sonal.santan@amd.com> wrote:
> >
> > On 11/19/22 12:44, Oded Gabbay wrote:
> > > This is the fourth (and hopefully last) version of the patch-set to add the
> > > new subsystem for compute accelerators. I removed the RFC headline as
> > > I believe it is now ready for merging.
> > >
> > > Compare to v3, this patch-set contains one additional patch that adds
> > > documentation regarding the accel subsystem. I hope it's good enough for
> > > this stage. In addition, there were few very minor fixes according to
> > > comments received on v3.
> > >
> > > The patches are in the following repo:
> > > https://git.kernel.org/pub/scm/linux/kernel/git/ogabbay/accel.git/log/?h=accel_v4
> > >
> > > As in v3, The HEAD of that branch is a commit adding a dummy driver that
> > > registers an accel device using the new framework. This can be served
> > > as a simple reference.
> > >
> > > v1 cover letter:
> > > https://lkml.org/lkml/2022/10/22/544
> > >
> > > v2 cover letter:
> > > https://lore.kernel.org/lkml/20221102203405.1797491-1-ogabbay@kernel.org/T/
> > >
> > > v3 cover letter:
> > > https://lore.kernel.org/lkml/20221106210225.2065371-1-ogabbay@kernel.org/T/
> >
> > Thanks for defining the new accel subsystem. We are currently working on
> > DRM based drivers for unannounced acceleration devices. I am fine with
> > these changes with the assumption that the choice of using classic DRM
> > or accel is left up to the individual driver.
> 
> I don't think that decision should be up to any individual driver
> author. It will have to be consensus with me/Daniel/Oded and the
> driver authors.

Plus the entire point of this is that it's _still_ a drm based driver. So
aside from changing a flag in the kernel driver and adjusting userspace to
find the right chardev, there should be zero changes need for an existing
drm based driver stack that gets ported to drivers/accel.

And of course if we realize there's issues as we add drivers, we can fix
things up. This is just to kick things off, not something that's going to
be cast in stone for all eternity.

Sonal, with that clarification/explanation, is this entire thing
reasonable in principal and you can drop an Ack onto the series?

Thanks, Daniel
-- 
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch