On Monday, January 03, 2011 09:34:34 pm Reinhard Meyer wrote:
Dear Marcel,
On Monday, January 03, 2011 07:57:33 pm Marcel wrote:
Great, I've got it to boot.
I have no idea why it suddenly works, but I guess after reviewing my config and changing a couple of things one of those things must have been the cause.
CONFIG_SYS_TEXT_BASE is 0x7f300000 which is the final jump address as mentioned in AT91bootstrap.
If this address is near end of SDRAM, and since u-boot relocates itself to end of SDRAM, you might risk an overlap, depending on malloc and stack sizes.
I'd advise to have AT91Bootstrap load u-boot to the very SDRAM start (so I did for our board).
OK, thanks for the advice. If seems the SDRAM end address is 0x80000000
I guess the issue is solved now and I'll certainly check out at91bootstrap to create my own low level init in u-boot. It eventually saves time at production level I think and I generally like to keep things as simple as possible anyways. Since all initialisation code is in at91bootstrap I guess I can just move it from there to u-boot (sounds easy, but I guess there's some more to it).
It might not be THAT easy. Also be aware that at91bootstrap is not GPL-ed...
Does that really make a difference if all these values basically can be read from specs ? I know what you mean however and I respect code if it is not OK to copy. I didn't realise at91bootstrap wasn't GPL, so thanks for letting me know.
I did check the code already today and it does seem to be not all that complex. However, I also have no real need to bypass at91bootstrap. As long as it works for me, basically the only need would be to simplify my production procedure and that's just a very minor step. I don't expect I'll focus on this for some time.
Best regards, Marcel