Can you provide some more information? What has been fixed here? Can you provide a use case? Could you provide some screenshots that illustrate the problem?

Some background information can be found in this issue: https://bugs.scummvm.org/ticket/14024

This isn't a problem with SDL2 since the list of dirty rectangles is ignored by the SDL_UpdateRects replacement, but I'm not sure why it was only visible with 32-bit displays. This PR attempts to fix the issue by moving the addition of both mouse dirty rects to before the game screen or overlay is blitted.

Thank you!

Unfortunately, it seems that this commit introduced (or uncovered) a regression either in SURFACESDL itself or the KYRA engine: https://bugs.scummvm.org/ticket/14147

@ccawley2011 could you give an example of how to reproduce the issue with a 32-bit game? The bit making internUpdateScreen() call undrawMouse() every time is causing performance issues in Director (there's a lot of ad-hoc places which can call updateScreen, relying a bit on the assumption that it's near-zero cost when there's no dirty rects).

@ccawley2011 could you give an example of how to reproduce the issue with a 32-bit game? The bit making internUpdateScreen() call undrawMouse() every time is causing performance issues in Director (there's a lot of ad-hoc places which can call updateScreen, relying a bit on the assumption that it's near-zero cost when there's no dirty rects).

It can be replicated in the launcher when compiling on Linux with SDL 1.2 and changing the call to SDL_SetVideoMode to use 32bpp instead of 16bpp or 8bpp - without this PR, you should be able to notice the cursor graphic getting clipped when moving the mouse from side to side.