Bernard,
Nigel Hathaway wrote:
Hoping to avoid re-inventing the wheel....
It would seem that there are a number of people who are
independently
porting UBI to U-Boot.
I expressed interest in doing so at the time but nothing happened
about
it unfortunately. I can't speak for anyone else though.
Bernard.
In the short term we've found a way to get round having to UBI and hence porting it to U-Boot (we have a NOR/NAND combination, and we've gone for the NAND as entirely jffs2).
However, the next spin of our product will use entirely NAND flash: it will use an SoC which has an internal ROMboot and can load U-Boot directly from NAND flash. We will have to go to UBI then (starting development early 2009).
I have been playing around with a board from Olimex: the SAM9-L9260 (it has an Atmel AT91SAM9260 on it). This has a ROMboot with U-Boot in NAND flash. The board I am using has an interesting feature: the block which U-Boot is configured to use for storing its environment just happens to be a bad block, so I can't change and store the U-Boot environment parameters on it.
Either the U-Boot that comes pre-flashed into the board (v1.2.0) is misconfigured, or that version of U-Boot has failed to take into account the possibility of the scenario described. Either way, there would appear to be some work to do to get U-boot up-to-date with Linux MTD state-of-the-art.
Nigel