Re: [PATCH v2 0/9] MR6157: d3dx9: More miscellaneous fixes.

[Matteo Bruni (＠Mystral)]

Matteo Bruni (@Mystral) commented about dlls/d3dx9_36/surface.c:
>  if (FAILED(hr = lock_surface(dst_surface, &dst_rect_aligned, &lockrect, &surface, TRUE)))
>      return hr;
>
> -
>
> set_d3dx_pixels(&dst_pixels, lockrect.pBits, lockrect.Pitch, 0, dst_palette,
>            (dst_rect_aligned.right - dst_rect_aligned.left), (dst_rect_aligned.bottom - dst_rect_aligned.top), 1,
>            dst_rect);
>    OffsetRect(&dst_pixels.unaligned_rect, -dst_rect_aligned.left, -dst_rect_aligned.top);
> -
>
> d3dx_load_pixels_from_pixels(&dst_pixels, destformatdesc, &src_pixels, srcformatdesc, filter, color_key);
> - if (FAILED(hr = d3dx_load_pixels_from_pixels(&dst_pixels, destformatdesc, &src_pixels, srcformatdesc, filter, color_key)))
> - {
> -
>    unlock_surface(dst_surface, &dst_rect_aligned, surface, TRUE);

A bit of an academic question, but: should we rather call `unlock_surface(..., FALSE);` in error paths?