D24556: [platforms/drm] Fix hardware cursor initialization

ezequielg created this revision.
Herald added a project: KWin.
Herald added a subscriber: kwin.
ezequielg requested review of this revision.

REVISION SUMMARY
  We can't connect cursor until drmOutput::initCursor
  is called, which is done by drmBackend::updateOutputs.
  
  Fix this by splitting the connect part into its own
  drmBackend::connectCursor private function.
  
  This allows to use the DRM legacy cursor API.
  
  Signed-off-by: Ezequiel Garcia <[email protected]>
  
  [platforms/drm] Allow to specify to dumb buffers' format
  
  Signed-off-by: Ezequiel Garcia <[email protected]>
  
  [platforms/drm] RFC/PoC: Support hardware cursors using DRM atomic API
  
  This needs the new DRM AMEND uAPI in order to make sure
  the cursor is committed on the same vblank as content.
  
  Currently, the legacy cursor API is the best choice. Once the
  AMEND uAPI is merged upstream, it'll be possible to switch to it.
  
  Signed-off-by: Ezequiel Garcia <[email protected]>

REPOSITORY
  R108 KWin

BRANCH
  atomic-cursors (branched from master)

REVISION DETAIL
  https://phabricator.kde.org/D24556

AFFECTED FILES
  plugins/platforms/drm/drm_backend.cpp
  plugins/platforms/drm/drm_backend.h
  plugins/platforms/drm/drm_buffer.cpp
  plugins/platforms/drm/drm_buffer.h
  plugins/platforms/drm/drm_output.cpp
  plugins/platforms/drm/drm_output.h

To: ezequielg
Cc: kwin, LeGast00n, The-Feren-OS-Dev, sbergeron, jraleigh, fbampaloukas, GB_2, mkulinski, ragreen, jackyalcine, iodelay, crozbo, bwowk, ZrenBot, ngraham, alexeymin, himcesjf, lesliezhai, ali-mohamed, hardening, romangg, jensreuterberg, abetts, sebas, apol, ahiemstra, mart