Haavard Skinnemoen wrote:
Mark Jackson mpfj-list@mimc.co.uk wrote:
Haavard Skinnemoen wrote:
Possibly, but NGW100 is the only one which I've seen reports about. STK1000 is safe since it doesn't use the CFI driver.
I did kinda report this in the thread "JFFS2 scanning bug", and the "triple-revert" patch you posted on 26/05/09 16:58 appeared to fix it.
Ah...so it breaks JFFS2 as well? I doubt that changing the environment address fixes that...
Since this didn't change any board files (only the core CFI files) I guess I assumed this "revert" would work its way upstream and I wouldn't have to change anything.
Hmm, yeah, maybe I should post the revert again.
I have to admit I'm completely confused about how u-boot deals with virtual and physical addresses these days. It used to expose only physical addresses through external interfaces, but now it looks like it's a bit of both, and it's impossible to tell which goes where.
Shall I just submit a patch to fix the mimc200 board in the same way as your NGW100 patch ?
Yes, that will probably be a good idea if it has the same problem with saveenv.
Okay ... looks like there are 2 problems revolving round CFI.
(1) saveenv (2) jffs2
The CONFIG_ENV_ADDR patch fixes (1) but *not* (2).
The "triple-revert" patch fixes both (1) and (2).
Not quite sure how to proceed from here. For the time being, I'll go down the "triple-revert" patch route until something better pops up !!
Regards Mark