On Wed, Sep 24, 2025 at 07:43:29PM +0300, Jani Nikula wrote:
>
> Jani Nikula (11):
> drm/{i915,xe}/stolen: rename i915_stolen_fb to intel_stolen_node
> drm/xe/stolen: rename fb to node in stolen compat header
> drm/xe/stolen: convert compat stolen macros to inline functions
> drm/xe/stolen: switch from BUG_ON() to WARN_ON() in compat
> drm/i915/stolen: convert intel_stolen_node into a real struct of its
> own
> drm/xe/stolen: convert compat static inlines to proper functions
> drm/{i915,xe}/stolen: make struct intel_stolen_node opaque
> drm/{i915,xe}/stolen: add device pointer to struct intel_stolen_node
> drm/{i915,xe}/stolen: use the stored i915/xe device pointer
> drm/{i915,xe}/stolen: convert stolen interface to struct drm_device
> drm/xe/stolen: use the same types as i915 interface
Looks fine by me. Series is
Reviewed-by: Ville Syrjälä <[email protected]>
Side note: I have branch somewhere that replaces the raw
drm_mm_node FBC stuff with a real i915_gem_object. I used
that as a way to easily expose the CFB and LLB as files in
debugfs so that I could observe/modify the actual CFB contents.
I should look into making that official to help future FBC
debugging. With the abstraction layer I shouldn't even need
to touch the FBC code itself anymore...
--
Ville Syrjälä
Intel
