DXTn stride is to the next block, which is the equivalent to 4 rows

dlls/d3d8/drawprim.c

@@ -1487,7 +1487,7 @@ void drawPrimitive(LPDIRECT3DDEVICE8 iface,
                     TRACE("Saving texture %s (Format:%s)\n", buffer, debug_d3dformat(((IDirect3DBaseTexture8Impl *)This->StateBlock->textures[textureNo])->format));
                     IDirect3DTexture8Impl_GetSurfaceLevel((LPDIRECT3DTEXTURE8) This->StateBlock->textures[textureNo], 0, &pSur);
                     IDirect3DSurface8Impl_SaveSnapshot(pSur, buffer);
-                    IDirect3DSurface8Impl_ReleaseRef(pSur);
+                    IDirect3DSurface8Impl_Release(pSur);
                 }
             }
            }

dlls/d3d8/surface.c

@@ -152,10 +152,16 @@ HRESULT WINAPI IDirect3DSurface8Impl_LockRect(LPDIRECT3DSURFACE8 iface, D3DLOCKE
       TRACE("(%p) : rect@%p flags(%08lx), output lockedRect@%p, memory@%p\n", This, pRect, Flags, pLockedRect, This->allocatedMemory);
     }
 
-    pLockedRect->Pitch = This->bytesPerPixel * This->myDesc.Width;  /* Bytes / row */    
-    if (This->myDesc.Format == D3DFMT_DXT1) /* DXT1 is half byte per pixel */
-        pLockedRect->Pitch = pLockedRect->Pitch/2;
-    
+    /* DXTn formats dont have exact pitches as they are to the newt row of blocks, 
+         where each block is 4x4 pixels, 8 bytes (dxt1) and 16 bytes (dxt3/5)      
+          ie pitch = (width/4) * bytes per block                                  */
+    if (This->myDesc.Format == D3DFMT_DXT1) /* DXT1 is 8 bytes per block */
+        pLockedRect->Pitch = (This->myDesc.Width/4) * 8;
+    else if (This->myDesc.Format == D3DFMT_DXT3 || This->myDesc.Format == D3DFMT_DXT5) /* DXT3/5 is 16 bytes per block */
+        pLockedRect->Pitch = (This->myDesc.Width/4) * 16;
+    else
+        pLockedRect->Pitch = This->bytesPerPixel * This->myDesc.Width;  /* Bytes / row */    
+
     if (NULL == pRect) {
       pLockedRect->pBits = This->allocatedMemory;
       This->lockedRect.left   = 0;