https://bugs.winehq.org/show_bug.cgi?id=9378
--- Comment #24 from Kevin2k gbroers@quicknet.nl --- joaopa,
I just looked at your Diff attachment from 2020-09-23. Yes, that makes much more sense then the previous workaround from Alexander from 2017. According to what I see in your diff, 'shift' is just the bit-offset of the pixel's r,g,b,a component data.
I do find it odd that in my hardware tests cAlphaShift is 0 for many pixelformats that do have an alpha channel. Does that suggest the alpha channel is the first byte for these modes?