Re: [coreboot] AGESA fam15 code removals

[Stefan Reinauer]

* Martin Roth via coreboot  [171005 19:00]:
> I've got very mixed feelings about pushing the changes that we know aren't
> going to work, especially right before we abandon the boards to a branch.

We are not abandoning any boards. The whole reason we are doing branches
and not tags is that boards can be fixed on older branches after they
are no longer part of ToT.

Part of keeping coreboot flexible and fresh is to reduce the amount of
boards one has to consider for any given change. Trying to keep boards
forever was great when we had 30 boards in the tree, and Ron had all of
them. These days we add 30 boards per year or more, and 10yr old silicon
does not bring a whole lot to the table for making the latest and
greatest coreboot better. If it makes the code much worse and nobody
maintains it, it should live in a branch. Where it also does not further
decay without testing, so it really is a win win situation both for old
and new boards.

Stefan



-- 
coreboot mailing list: coreboot@coreboot.org
https://mail.coreboot.org/mailman/listinfo/coreboot

Re: [coreboot] AGESA fam15 code removals

[Kyösti Mälkki]

On Thu, Oct 5, 2017 at 11:09 PM, taii...@gmx.com  wrote:
> I have an H8SCM, but have never successfully booted it with coreboot due to
> being lazy.
>
> I have a test clip and would be willing to test any changes.

Great!

We have this thing called git repository, you know. Please pick a
commit. ANY commit. Some people also call those "changes". Happy
bisecting, please report last known good commit back to us.

Kyösti

-- 
coreboot mailing list: coreboot@coreboot.org
https://mail.coreboot.org/mailman/listinfo/coreboot

Re: [coreboot] AGESA fam15 code removals

[taii...@gmx.com]

I have an H8SCM, but have never successfully booted it with coreboot due 
to being lazy.


I have a test clip and would be willing to test any changes.

--
coreboot mailing list: coreboot@coreboot.org
https://mail.coreboot.org/mailman/listinfo/coreboot

Re: [coreboot] AGESA fam15 code removals

[Martin Roth via coreboot]

I've got a supermicro H8SCM board, although I don't think I've ever tried
to boot it, so I don't know its state.  I also suspect that it's got a
family 10h processor, not a fam15.

I can try to get it set up this weekend, and I can bring it to the
conference with me if that would help.
Here's the list of boards I've got for future reference:
https://www.coreboot.org/User:MartinRoth#Platforms_supported_by_coreboot

I've got very mixed feelings about pushing the changes that we know aren't
going to work, especially right before we abandon the boards to a branch.
The idea is that the branch should be in a working state, though as
mentioned, we have no idea if these boards are currently working. On the
other hand, I don't want to get in the way of the progress on the AGESA
changes.  I guess if we do go ahead and push the changes, document them as
well as we can and list what would need to be done to get the boards to a
working state (for these changes).

Martin
-- 
coreboot mailing list: coreboot@coreboot.org
https://mail.coreboot.org/mailman/listinfo/coreboot

[coreboot] AGESA fam15 code removals

[Kyösti Mälkki]

Hi

I can remember exactly one user of AGESA fam15 code from the past 5
years or so. He has informed me of having no possibility to test with
the H8QGI he once had access to and as the little discussion on this
list about H8SCM also appears seized now, I have to inform none of the
AGESA fam15 boards in the tree have reached EARLY_CBMEM_INIT and
nobody is actively working on the topic.

So after release of 4.7, I would place the following boards on the
immediate LATE_CBMEM_INIT kill-list:
  amd/dinar, supermicro/h8scm, supermicro/h8qgi, tyan/s8226.

After removal of the boards, related platform code and vendorcode
subtree will be removed as well.

Actually, I would like to commit the incomplete work on these fam15
boards (that do pass abuild) even before 4.7 release to proceed with
other related cleanups on AGESA. Work on the boards has been available
for review/improvement since Aug 01 but there has been no takers, and
really nobody with hardware to even ask for testing.

Here's the top-of-tree cleanup commit, with fam15 as dependency, that
I'd like to get in before 4.7 release:
https://review.coreboot.org/c/21695/3

Now this would be against our normal no-regressions policy, but we
really do not have any feedback for these fam15 boards for so long
time now... So would it matter? If I just tagged them with
[INCOMPLETE] or something like that?

Kyösti

-- 
coreboot mailing list: coreboot@coreboot.org
https://mail.coreboot.org/mailman/listinfo/coreboot