From: Ville Syrjälä <ville.syrjala@xxxxxxxxxxxxxxx> GLK blew up in ci in the glk_force_audio_cdclk() path. The reason being some reordering in intel_atomic_check() so we're no longer guaranteed to have a dbuf_state when we commit crtcs. So I changed the approach slightly to keep duplicate pipe ddb allocations in the crtc state as well. A bit annoying, especially for just sanity checking against ddb overlaps, but couldn't think of anything nicer instead. The change trickled over to several patches so resending the lot. Ville Syrjälä (8): drm/i915: Extract intel_crtc_ddb_weight() drm/i915: Pass the crtc to skl_compute_dbuf_slices() drm/i915: Introduce intel_dbuf_slice_size() drm/i915: Introduce skl_ddb_entry_for_slices() drm/i915: Add pipe ddb entries into the dbuf state drm/i915: Extract intel_crtc_dbuf_weights() drm/i915: Encapsulate dbuf state handling harder drm/i915: Do a bit more initial readout for dbuf drivers/gpu/drm/i915/display/intel_display.c | 19 - .../drm/i915/display/intel_display_debugfs.c | 1 - drivers/gpu/drm/i915/i915_drv.h | 9 - drivers/gpu/drm/i915/intel_pm.c | 542 ++++++++---------- drivers/gpu/drm/i915/intel_pm.h | 7 +- 5 files changed, 258 insertions(+), 320 deletions(-) -- 2.26.2 _______________________________________________ Intel-gfx mailing list Intel-gfx@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/intel-gfx
