Am Freitag, 26. Februar 2010 schrieben Sie:
Dear Juergen,
On Fri, Feb 26, 2010 at 12:16:22PM +0100, Juergen Schindele wrote:
Am Freitag, 26. Februar 2010 schrieb Wolfgang Denk:
Dear Juergen Schindele,
In message 201002261056.32502.schindele@nentec.de you wrote:
i need a different bootloader than the one installed. so i tried to develop a private one by loading it to RAM. So i modified the TEXT_BASE = 0x00400000 to not collide with the installed one.
This is not enough. You must also make sure not to try to perform all the low level initializations again that have already been done.
I know that, but SKIP_LOW_LEVEL_INIT is defined because register init is done with kirkwood boot header for booting from SPI/NAND-Flash. I dont see other blocking points. :-( i tried to skip relocate too but it did not help !
I have absolutely no experience concerning kirkwood, but did the same for coldfire for running U-Boot as the flasher application started from the debugger... IIRC there were three places I had to change for coldfire:
- low-level init: do not put the vector table at the beginning of code (probably better: initialize it in the proper place?)
- leave the vector base register as it is (should be set up properly by first loader/debugger, and U-Boot does not use interrupts)
- the cpu speed detection/setting code was basically switched back to detection for CONFIG_MONITOR_IS_IN_RAM because it involved putting the RAM to self-refresh, which is obviously not possible when running from it
(For Coldfire, a special case was that I had to disable monitor protection manually because the standard condition fails due to the memory layout.)
Thanks for your hint. It lead me to comment out "kirkwood_mpp_conf" in function "board_init" and now it works quite great like expected. In my understanding the mpp configuration of rd6281a is buggy !?
we'll see in furter tests.
Here's my patch for Coldfire MCF532x/7x: http://www.mail-archive.com/u-boot@lists.denx.de/msg28779.html
BTW, disabling relocation may lead to bad results for malloc and the initial stack setup. I am not sure but seem to remember I had some problems when I tried it - and, it should not be a problem to do it as usual anyways if everything else is working correctly.
Hope this can give some useful hints, Wolfgang
-------------------------------------------------------------- Jürgen Schindele Software-Entwicklung
NENTEC Netzwerktechnologie GmbH Greschbachstr. 12 76229 Karlsruhe Deutschland Telefon: +49 721 94249-51 Telefax: +49 721 94249-10 E-Mail: schindele@nentec.de WEB: www.nentec.de
Geschäftsführung: Klaus Becker, Roland Knapp Sitz der Gesellschaft: Karlsruhe Handelsregister: Amtsgericht Mannheim HRB 107658 --------------------------------------------------------------