Hi,
From the multitexture demo:
void loadTexture( void ) { D3DXCreateTextureFromFile( g_pd3dDevice, "test.bmp", &g_pTexture_0 ); D3DXCreateTextureFromFile( g_pd3dDevice, "checker.bmp", &g_pTexture_1 );
g_pd3dDevice->SetSamplerState(0, D3DSAMP_MINFILTER, D3DTEXF_LINEAR); g_pd3dDevice->SetSamplerState(0, D3DSAMP_MAGFILTER, D3DTEXF_LINEAR); }
Two textures are created
//----------------------------------------------------------------------------- // Name: shutDown() // Desc: //----------------------------------------------------------------------------- void shutDown( void ) { if( g_pTexture_0 != NULL ) g_pTexture_0->Release();
if( g_pVertexBuffer != NULL ) g_pVertexBuffer->Release(); if( g_pd3dDevice != NULL ) g_pd3dDevice->Release(); if( g_pD3D != NULL ) g_pD3D->Release();}
One texture is released
void render( void ) {
<snip> g_pd3dDevice->SetTexture( 0, g_pTexture_0 ); g_pd3dDevice->SetTexture( 1, g_pTexture_1 );
The texture stage is set -> Texure AddRef. The texture isn't unset.
The missing release of the 2nd texture is definitly an application bug. What remains open is if not unsetting the texture stages is a bug too. Can someone check on Windows if (a) creating a surface or texture addrefs the Direct3DDevice, and if (b) assigning a texture to a texture stage addrefs the Texture? If (a) isn't true, we should unset all texture stages when the WineD3DDevice is released, if (b) is false we shouldn't do that too. If (a) and (b) are true, the application is buggy too. I'm sorry that I can't test, because I've no D3D9-Equipped windows machine available ATM.