[PATCH weston v5 00/11] drm: port the drm backend to the new init api

[Bryce Harrington]

On Wed, Apr 13, 2016 at 05:21:39PM +0300, Pekka Paalanen wrote:
> On Wed, 13 Apr 2016 03:25:04 -0700
> Bryce Harrington <bryce at osg.samsung.com> wrote:
> 
> > In following up on my earlier update of Giulio's drm backend config
> > patch, I've taken the liberty to try and also integrate a couple of
> > Benoit's other backend configuration patches into this patchset.
> > 
> > Giulio and Benoit took different approaches in their implementations.
> > I've attempted to reconcile them so that they are stylistically
> > consistent, along with incorporating pq's request to have the root
> > config structure incorporate versioning information.  I went through all
> > the review comments against each of the backend config patches and
> > incorporated the suggestions people made.  The resultant changes have
> > ended up a bit more extensive than I had expected, and I apologize for
> > any toe stepping I'm doing but my hope is to move the collective work
> > closer to being landable.
> > 
> > ---
> > todo:
> >   - Use backend-specific header #defines for struct_version values
> > v5:
> >   - Add missing compositor-drm.h
> >   - Integrate wayland, x11, and headless backend patches
> >   - Implement struct versioning for wayland, x11, and headless backend
> >     patches in the weston_backend_config structure, as requested by pq
> >   - Drop bzero usage as suggested in review by Jan Engelhardt
> >   - Don't change 'backend_init' entry point function name, as suggested
> >     in review by Giulio
> >   - Prefer use of load_backend_new() for actual module loading, as
> >     suggested in Giulio's review of the x11 backend patch
> >   - Refactor all backend initialization code paths and code style for
> >   	consistency.
> >   - Squashed drm config struct versioning with drm patch; remainder moved
> >   	to a prerequisite patch.
> > v4:
> >   - Update to current trunk
> >   - Add missing param doc for mode in drm_output_choose_initial_mode
> >   - Rebase to account for code changes by 91880f1e to make vt
> >     switching configurable.
> > 
> > 
> > 
> > Benoit Gschwind (1):
> >   x11: port the x11 backend to the new init api
> > 
> > Bryce Harrington (9):
> >   Revert "main: Remove unused function load_backend_new()"
> >   compositor: Version the backend configuration structures
> >   drm: Fix gcc warning about missing braces.
> >   compositor: Document refs for alternatives/assumptions for backend
> >     configs
> >   headless: port the headless backend to the new init api
> >   drm: Code and comments reformatting for consistency with other backend
> >     configs
> >   drm: Don't hang onto the backend config object post-backend_init
> >   Enforce destruction of all backend config objects after initialization
> >   drm: Drop use of drm_config config wrapper
> > 
> > Giulio Camuffo (1):
> >   drm: port the drm backend to the new init api
> > 
> >  Makefile.am               |   5 +
> >  src/compositor-drm.c      | 216 ++++++++++++++------------------
> >  src/compositor-drm.h      | 125 +++++++++++++++++++
> >  src/compositor-headless.c |  69 +++++------
> >  src/compositor-headless.h |  51 ++++++++
> >  src/compositor-x11.c      | 157 ++++++++----------------
> >  src/compositor-x11.h      |  60 +++++++++
> >  src/compositor.h          |  24 +++-
> >  src/main.c                | 304 +++++++++++++++++++++++++++++++++++++++++++++-
> >  9 files changed, 734 insertions(+), 277 deletions(-)
> >  create mode 100644 src/compositor-drm.h
> >  create mode 100644 src/compositor-headless.h
> >  create mode 100644 src/compositor-x11.h
> > 
> 
> Hi,
> 
> patches: 1, 2, 4, 9, 10
> are Reviewed-by: Pekka Paalanen <pekka.paalanen at collabora.co.uk>
> 
> Patch 3 already has my Acked-by and it is good to go, since the patches
> coming later fix the issues I raised earlier.
> 
> Patch 5 seems to have dropped the email archive links, those would be
> good to keep. With them added back, R-b me.
> 
> I think patches 8-11 should have come before 6 and 7. I needed to
> review 8-11 first to see where design ended up, before I can check that
> 6 and 7 follow it too.
> 
> Seems like the changes in patch 10 should be squashed into the patches
> converting each backend,  but okay anyway.
> 
> Patch 11 is R-b me otherwise, except it will need some changes
> according to the comment to patch 8.
> 
> I am still in the middle of reviewing patches 6 and 7. Patch 6 has some
> compositor-drm change mixed in. More comments for these two patches
> will follow another day.

Thanks for reviewing the patchset (and thanks Giulio and Benoit as
well).  Except for one or two items I'm leaving for followup I think
I've incorporated all the requested changes, but I wouldn't be at all
surprised if I missed something, so apologies ahead of time.  With all
the rebasing I could have also introduced an irregularity or two.  So
even though I've included your R-b's, you may want to take another look
at the patches and doublecheck.

Re-testing of all three backends would also be appreciated; I've verified
make check works but haven't re-checked that everything runs (I wanted
to get this posted before I left for the weekend).

Bryce