How to resume drm-openchrome development
Daniel Vetter
daniel at ffwll.ch
Sun Aug 28 16:33:05 UTC 2016
On Fri, Aug 26, 2016 at 07:39:34AM +0200, Kevin Brace wrote:
> Hi,
>
> I am sending this posting to dri-devel mailing list and Dave Airlie.
> This is sort of a summary of where OpenChrome and drm-openchrome is now.
>
> https://lists.freedesktop.org/archives/dri-devel/2016-August/116736.html
>
> I know that drm-openchrome is still not mainlined, and I am not the
> original developer, but since there was a talk of removing or putting
> DRI1 into probationary status (that's the way I see it), I will like to
> start putting more time into working on drm-openchrome. Last time I
> tried, I tried to merge the drm-openchrome's VIA DRM with Linux 4.2
> kernel on my development computer, but I ran into so many compilation
> problems that I stopped working on it. How can I resolve these
> compilation problems?
> Back then I was surprised to see so many compilation problem even though
> there was only 3 minor version difference between 3.19 and 4.2. Also, I
> recall seeing a posting from Dave Airlie to James Simmons (on
> Openchrome-devel mailing list) some years ago that appear to look like
> Dave was not happy with certain aspects of drm-openchrome at the time.
> What were they, and how can I tackle the problem?
I think Rob's suggestion is probably best - submit a smaller part of the
entire thing (starting with basic kms is probably best) and get that
landed upstream. Then extend from there. From what I can see the worst way
to upstream is trying to get things perfect first out-of-tree, since
long-term you will simply not be able to keep up with upstream changes.
Also 4.2 is already really, really old, current drm work is all aiming at
4.9 already. That's a _much_ bigger difference than between 3.19 and 4.2.
Upstream drm moves at a rather brisk pace, 3 kernel releases is thousands
of commits and not suprise at all that the driver doesn't compile (or
work) any more. Being in upstream is much better, since then you're driver
will co-evolve with everything else.
-Daniel
--
Daniel Vetter
Software Engineer, Intel Corporation
http://blog.ffwll.ch
More information about the dri-devel
mailing list