DRM backend file split, part two
The following discussion from !213 (merged) should be addressed:
-
@pq started a discussion: I have gone ahead and merged this, because it looks good enough, and I don't want to review this for another round. ;-)
I did spot the following, but I think they are minor enough to not care about, or can be better fixed as follow-up as time permits:
-
drm_wait_vblank_pipe()
has shifted to a different position indrm.c
.
-
Oops. I think I'd moved that and then had to reinstate it. I didn't spot it when going through the diffs.
> - `choose_mode()` is renamed to `drm_output_choose_mode()` which would be preferable in a separate commit.
True. Apologies.
> - The header still has `drm_view_transform_supported()` as a static inline.
Eek. I did mean to fix that. Will take care of it in a follow-up MR.
> - It might be nice to reorganize things so that these declarations are not necessary:
> ```
> +extern struct drm_property_enum_info plane_type_enums[];
> +extern const struct drm_property_info plane_props[];
> +extern struct drm_property_enum_info dpms_state_enums[];
> +extern const struct drm_property_info connector_props[];
> +extern const struct drm_property_info crtc_props[];
> ```
Yes, that really bothered me as well. At the moment it's somewhat split between the 'pure KMS' side of kms.c and the output/head/plane handling inside drm.c. The split is a bit awkward, and I was planning to have a longer look at a more sensible split when I came to moving the head/output management into a separate file.
Thanks for the review!