wined3d: Regression in DX8 SDK samples with VBO patch

[Jason Green]

http://source.winehq.org/git/?p=wine.git;a=commit;h=2122026713baa31809318160...

This patch breaks most of the DirectX 8 SDK Direct3D samples
(Billboard, ClipMirror, Cull, etc.).  It seems like most of the
samples that use Shaders work fine, but those that don't are broken
(misplaced vertices, etc.).

[Stefan Dösinger]

Am Dienstag 27 Juni 2006 04:08 schrieb Jason Green:
> http://source.winehq.org/git/?p=wine.git;a=commit;h=2122026713baa3180931816
> 07db79d7577f728e9
>
> This patch breaks most of the DirectX 8 SDK Direct3D samples
> (Billboard, ClipMirror, Cull, etc.).  It seems like most of the
> samples that use Shaders work fine, but those that don't are broken
> (misplaced vertices, etc.).

I have done a quick test with the Billboard demo, and I can confirm that it is 
broken :-( Apparently the terrain is broken, and the vertex buffer containing 
it is created by native d3dxof.dll. Looks like this dll is doing something I 
didn't expect :-\

[Stefan Dösinger]

Am Dienstag 27 Juni 2006 04:08 schrieb Jason Green:
> http://source.winehq.org/git/?p=wine.git;a=commit;h=2122026713baa3180931816
> 07db79d7577f728e9
>
> This patch breaks most of the DirectX 8 SDK Direct3D samples
> (Billboard, ClipMirror, Cull, etc.).  It seems like most of the
> samples that use Shaders work fine, but those that don't are broken
> (misplaced vertices, etc.).

Apparently(and according to the msdn) vertex buffers can be locked multiple 
times( lock->lock->...->lock ->unlock->...->unlock->unlock). The some d3d 
helper lib does that and it fools the vbo loading code. I'll send a fix+test 
for that.