[PATCH 01/10] wined3d: Introduce struct wined3d_dirty_regions.

[Henri Verbeet]

On 17 April 2017 at 22:44, Matteo Bruni <matteo.mystral at gmail.com> wrote:
> 2017-04-17 18:14 GMT+02:00 Stefan Dösinger <stefandoesinger at gmail.com>:
>> A standard wine list would be better IMHO.
>
> Yeah. Have a look at include/wine/list.h, also you have many examples
> of its use in wined3d already.
>
Why not an array? Stefan already mentioned wined3d_buffer.maps, in a fashion.

> I haven't really looked at the patches in detail but the general
> approach seems fine. Thank you for your work!
>
There are certainly some details to work out, like e.g. the first four
patches adding dead code, but in terms of general approach, these are
the things I think need resolving first:
  - How does this interact with the location management? (I.e.,
wined3d_texture_invalidate_location(), etc.) For example, in 132746,
~resource->map_binding is invalidated and dirty regions are added. How
should wined3d_texture_load_location() interpret that when loading for
example WINED3D_LOCATION_DRAWABLE?
  - Does it really make sense to track regions on the level of struct
wined3d_resource instead of e.g. strut wined3d_texture? If it does,
that would make wined3d_buffer.maps redundant.