-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
On 10/27/2015 11:14 AM, Martin Roth wrote:
The current thought is to create a new branch for 4.2, *THEN* remove the decided upon code from the master branch.
This removes the maintenance work on master going forward while providing a known point if someone wants to pick up the boards going forward.
I don't understand the question about creating a branch for other boards as well. The branch would contain all current boards.
The question is really about what boards/code are inactive. The code that's being updated right now for the ASUS/KGPE-D16 board looked inactive for a significant time.
Personally, I think it's a bad plan to remove the AGESA code. Throwing away the code that AMD has given us is telling them that they were right, and they shouldn't bother opening the source. This is a total contradiction of what we ACTUALLY want, which is the source code for the binary PI releases.
If it's still decided to remove the AGESA family 10/15 codebases, my thought would be to wait for at least the 4.3 release. That's going to be 3 months from now, giving us some time to finish getting in the code that's supposed to replace it and giving it time to stabilize.
Martin
If I may add to this discussion, much of the rationale for dropping AGESA has to do with the largely incompatible design (and coding style) of AGESA versus the rest of coreboot. By encouraging the use of AGESA over native intialization, not only does the native initialization code receive less attention, but development of new features (such as timestamping and early CBMEM) is also hampered.
- From what I understand, what coreboot actually wants is access the source code, but not necessarily to just use the source code as-is. Would it make more sense to keep the vendorcode available, but put a large warning on it that boards using said vendorcode are not supported (and remove them from Jenkins builds, etc.)? Maybe even moving those boards to a different subdirectory (vendor_shim_mainboards) would help reinforce this idea.
- -- Timothy Pearson Raptor Engineering +1 (415) 727-8645 (direct line) +1 (512) 690-0200 (switchboard) http://www.raptorengineeringinc.com
The AGESA code was always an awkward fit into coreboot. It was more like a badly designed artificial limb than a real part of the code base. I understand the idea of encouraging vendors to commit source but, at this point, the AMD ship has sailed off to Port Binary Blob. AGESA was helpful in its time but I think I'm with tpearson on this point.
I believe we should drop AGESA on any boards that have native support, and the sooner the better.
ron
-
ron minnich -
Timothy Pearson