Here are my thoughts on FSP.
FSP is how we deal with a bad situation: the fact that Intel chipsets are no longer open (the way they were until 2000 or so) and require a binary blob from Intel to get them working.
We have, as we can see, used several paths over the years to deal with the binary blob problem.
In 2001 Intel and Linux labs produced a linuxbios for a board in which intel provided a binary blob for the top 64k and linux labs did the rest.
That was all until the chromebooks. In the Intel-based chromebooks we've had to contend with lots of binary blobs: - ME binary - VBIOS (solved now) - memory ref. code - probably more to come if history is a guide
You can find the blobs for 2 of the Intel-based chrome systems at coreboot.org, no others. That's not an encouraging situation for FSP, because from a high level, FSP is another proprietary binary blob. Until it an be part of the coreboot repo it's not a viable solution.
The best approach was and remains a full source-based coreboot image, as is still possible for a few intel chipsets, all AMD chipsets, and many ARM chipsets.
FSP is an interesting idea, but again, until we can put it at coreboot.org, it's not a viable approach. It's also a way to address the LACK of full source-code base for a chipset. It's a way to solve a problem, but it is not the most desirable way.
So, the status of FSP is this: very nice approach to solving the binary blob problem, better than what we have (nothing), not practical until the FSP blob can be redistributed from coreboot.org. It will never unify all x86 chipsets because people will continue to work on source-code-based approaches.
thanks for your help and hard work!
ron
-
ron minnich