> Yes, I'm aware of that and *perhaps* you can do that with the tulip driver. > However, from my own impression and that from previous talks with Hugo and > Travis that this is not recommended. I don't quite recall these talks. I guess this multiple versions of tulip chipset can be supported using the single or several FreeBSD drivers available for it. The only issues i had in the past were regarding interrupt handling. > IMHO the driver covers too many different chipsets and in particular the > admtek chipsets should be a separate driver. Whether it would be "easy" to > use the freebsd code to do so was what I was getting at. I would like a lot of things to be and work differently, but we have to be pragmatic. If there's a possible path that allows us to support the hardware, and re-using the know-how of a group of people that work in the field and already spent a lot of effort to make sure the hardware is supported, it seems the most positive thing to me. I would rather concentrate our efforts on the components that are part of the operating system architecture, such as the network stack itself, and making sure that these work correctly, are full featured and follow the same design rules as the rest of the system. Hugo