[PATCH] drm.rst: Some notes

[Daniel Vetter]

On Thu, Jan 17, 2019 at 8:21 AM Jani Nikula <jani.nikula at linux.intel.com> wrote:
>
> On Fri, 11 Jan 2019, Daniel Vetter <daniel.vetter at ffwll.ch> wrote:
> > Some early notes from my first few pull requests. Lots of things that
> > need to be discussed/clarified/agreed upon, but should be good for a
> > starting point at least.
> >
> > Not entirely sure how we should split this all up, so all in one file
> > for now.
>
> For drm-misc and drm-intel we (read: I) have split it into three:
>
> 1) general branch description ("what's this for")
> 2) guidelines for committers (obviously not needed for drm)
> 3) guidelines for maintainers (the stuff the others don't care about)
>
> There are some PHB types who'll want to look at 1), so the audience for
> all of these is slightly different. So rather than splitting everything
> by branch, I'd like to split by audience, and try to group common stuff
> across branches together. (e.g. committer-guidelines.rst should pull in
> more stuff from the branch specific guidelines, with some unification,
> and perhaps the various files for 1) should be merged together too.)
>
> I know the content here is still pretty thin, and splitting it up might
> seem like overkill now, but would save from restructuring later.
>
> Does it make sense to you?

Yeah I guess the confusion is that we're in an in-between state where
some of this is done, but some of it isn't, so it's a bit confusing
still. If you want I think it makes sense to move the drm.rst under
the maintainer guide. Right now I don't think there's any need for drm
committers, it's all pull requests anyway.
-Daniel

>
> BR,
> Jani.
>
>
> >
> > v2: Fix typo
> >
> > Cc: Dave Airlie <airlied at gmail.com>
> > Signed-off-by: Daniel Vetter <daniel.vetter at intel.com>
> > ---
> >  drm.rst   | 66 +++++++++++++++++++++++++++++++++++++++++++++++++++++++
> >  index.rst |  1 +
> >  2 files changed, 67 insertions(+)
> >  create mode 100644 drm.rst
> >
> > diff --git a/drm.rst b/drm.rst
> > new file mode 100644
> > index 000000000000..ff2bbe9b8bfe
> > --- /dev/null
> > +++ b/drm.rst
> > @@ -0,0 +1,66 @@
> > +drm
> > +===
> > +
> > +Introduction
> > +------------
> > +
> > +drm is the overall graphics integration tree, and as such works slightly
> > +differently than feature trees managed with dim_:
> > +
> > +- Normally only takes pull requests.
> > +
> > +- Freezes for features from -rc6 to the end of the merge window as kernel
> > +  subsystem trees usually do. There`s no constantly open feature branch. After
> > +  -rc6 `drm-next` only accepts bugfixes and smaller cleanups aimed for the merge
> > +  window.
> > +
> > +- Doesn`t have committers, just maintainers, since the pull request load is
> > +  fairly minimal (for now). To keep it that way small trees are encouraged to
> > +  collaborate together in drm-misc or other groups of drivers.
> > +
> > +.. _dim: dim.html
> > +
> > +Backmerges
> > +----------
> > +
> > +All backmerges for `-next` trees should first land in `drm-next`, with an
> > +explicit merge commit.  This includes pull requests from driver trees based on
> > +newer upstream. In that case first apply the backmerge, then take the pull
> > +request.
> > +
> > +Only exception is right after -rc1 when `drm-next` reopens for features, where a
> > +fast-forward is all that's needed.
> > +
> > +Only backmerge tagged releases.
> > +
> > +Pull Request Review
> > +-------------------
> > +
> > +Special care should be taken to review commits which:
> > +
> > +- Touch files outside of what the maintainer maintains (drm core code, other
> > +  drivers, or even other subsystems).
> > +
> > +- Not reviewed patches - occasionally the lack of review is a process fumble and
> > +  patches never even made it to any list.
> > +
> > +- Changing uapi. Look both for include/uapi and anything adding new KMS
> > +  properties.
> > +
> > +- Check for last-minute rebases (all the patches have roughly the same commit
> > +  timestamp). Especially rebasing onto latest upstream right before sending out
> > +  is discouraged by Linus (since it invalidates the testing that happened).
> > +
> > +FIXME: Script as much as possible of the above.
> > +
> > +Opens
> > +-----
> > +
> > +- Lots of the above needs to be discussed.
> > +
> > +- Hard rule against being both drm and sub-tree maintainer, to prevent glaring
> > +  conflicts of interest? Commit rights for Dave in drm-misc?
> > +
> > +- To rebase or not to rebase (probably no, except the tree is on fire)
> > +
> > +- Recipient list for pulls to Linus (Daniel botched this, should be scripted?)
> > diff --git a/index.rst b/index.rst
> > index 48982cbde5a2..14086bb4ae68 100644
> > --- a/index.rst
> > +++ b/index.rst
> > @@ -24,6 +24,7 @@ Contents:
> >
> >     repositories
> >     drm-tip
> > +   drm
> >     drm-misc
> >     drm-intel
> >     committer-guidelines
>
> --
> Jani Nikula, Intel Open Source Graphics Center



-- 
Daniel Vetter
Software Engineer, Intel Corporation
+41 (0) 79 365 57 48 - http://blog.ffwll.ch