Hello Wolfgang,
On 01/15/2014 12:04 PM, Wolfgang Denk wrote:
Dear Holger,
In message 52D64089.6070207@keymile.com you wrote:
This commit removes support for the Freescale MPC82xx Power Architecture processors, i. e. MPC8240, MPC8245, MPC8247, MPC8248, MPC8250, MPC8255, MPC8260, MPC8265, MPC8266, MPC8272, MPC8280, ...
They have been out of production for years, and no active users left here. As some boards start causing problems, let's drop the obsolete and now dead code.
thats not valid for us. Our mgcoge3ne target which comes with a MPC8247 is still in production and maintained. If you look at the git log of
Argh... Can you foresee how much longer this hardware is likely to be maintained?
uhm. There is currently no plan to stop the production of this board. So for the next two years at least I would expect that they were still produced.
And as a sidenode I still have the request on my desk to integrate the POST tests for this board, which we already have for our PPC83xx and kirkwood boards.
So isn't it possible to remove only the broken boards and keep the generic parts?
Yes, this would be possible, too. But then, it appears you are the only remaining active user of MPC82xx. OK, MPC8247 is actually still marked as "active" at Freescale, soory I missed that - the MPC824x types I checked were in "No Longer Manufactured" state.
The thing is that there are tons of interdependencies an #defines that need to be checked so we don't leave too many unused #defines and such around.
yes I understand the desire to remove as much as unneeded code as possible.
I see several options now:
- We apply the patch as is, and if you really have to modify your code you would do this out-of-tree based on the last frozen version.
yes we could do that and keep a seperate branch for this board, but I don't like this. I guess I don't need to explain why I would like to avoid an additional branch on our site.
- I rework the patch to remove only the MPC826x / MPC828x code.
honestly this would be my favorite approach.
So if keeping 82xx support would't generate to much overload for u-boot I would appreciate to keep it. But if it interferes with future u-boot development we could also move it to a keymile specific branch.
And just out of curiosity. Why do you keep still 8xx board support? Is this more in use then 82xx? This is suprising to me.
Regards Holger