While converting page flip to atomic I've noticed it was easier to kill off intel_crtc->atomic first. This can be done by adding fb_bits, visible_changed, wm_changed to the crtc_state and deriving some primary state during pre/post_plane_update. This will probably conflict with the atomic wm work, and may conflict slightly with the fbc rework, but those conflicts shouldn't be too hard to fix up. After fixing that I've ran some tests and noticed a few SKL bugs, related to WM's and scaling, so I've fixed those up at the end. Maarten Lankhorst (11): drm/i915: Use passed plane state for sprite planes. drm/i915: Do not acquire crtc state to check clock during modeset. drm/i915: Kill off intel_crtc->atomic.wait_vblank. drm/i915: Update watermark related members in the crtc_state. drm/i915: Remove intel_crtc->atomic.disable_ips. drm/i915: Remove atomic.pre_disable_primary. drm/i915: Remove some post-commit members from intel_crtc->atomic. drm/i915: Nuke fbc members from intel_crtc->atomic. drm/i915/skl: Prevent unclaimed register writes on skylake. drm/i915/skl: Update watermarks before the crtc is disabled. drm/i915/skl: Do not allow scaling when crtc is disabled. drivers/gpu/drm/i915/intel_atomic.c | 3 + drivers/gpu/drm/i915/intel_display.c | 319 +++++++++++++++++------------------ drivers/gpu/drm/i915/intel_drv.h | 46 ++--- drivers/gpu/drm/i915/intel_pm.c | 5 + drivers/gpu/drm/i915/intel_sprite.c | 67 +++++--- 5 files changed, 224 insertions(+), 216 deletions(-) -- 2.1.0 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx http://lists.freedesktop.org/mailman/listinfo/intel-gfx
