Hi all,
Russell found a refcount bug in my fb refcounting conversion, but to also fix i915.ko I've opted for a slightly different approach.
While at it I've thought it would be good to backport some of the semantic changes we've implented in i915's ->set_config callback since we've forked our own modeset infrastructure.
The intel code receive lots more refactorings and cleanups (most of them right after the code fork) which would be simple to backport, but I'm kinda swamped (hint, hint, ...). But these few patches are the important bits.
Cheers, Daniel
Daniel Vetter (6): drm/crtc-helpers: Enforce sane set_config api drm/crtc-helper: no need to check for fb->depth/bpp drm/crtc-helper: explicit DPMS on after modeset drm/crtc-helper: don't disable disconnected outputs drm: check that ->set_config properly updates the fb drm: fix fb leak in setcrtc
drivers/gpu/drm/drm_crtc.c | 25 +++++++++++++++----- drivers/gpu/drm/drm_crtc_helper.c | 47 +++++++++++++++++-------------------- include/drm/drm_crtc.h | 4 ++++ 3 files changed, 45 insertions(+), 31 deletions(-)