u-boot relocate_code routine...

[atul.sabharwal at exgate.tek.com]

>>You can reinvent the wheel, or  read  the  documented  experience  of
>>others who went the same way long time ago.

I don't prefer re-inventing unless it adds some real value.  I had
mentioned the same thing to the design team and changing the memory map
solved the problem.

My hypothesis is that there still is a bug in relocate_code routine as
if I comment out trap_init routine which relocates the exception vectors
from flash to ram (in this case 0x10000000), the processor dies.  With
I-cache Off, and trap_init/interrupt_enable commented, it runs to u-boot
prompt.
With I-cache On, it dies 100% in this code.

Putting memory to 0x0 solves the problem and I am able to enable
trap_init & interrupt enable. Also, can turn I-cache on but seems to be
some cache problem as without BDI, it dies 100% but with BDI connected,
it runs with
Some fpga errors.  

If you get a chance, you could review the relocate_code routine if it
handles all input data values. I am new to PPC assembly, so if this was
written in C/as, it would have been lot simpler.  That is not important
but handling all data values seems to be a bug.

Best Regards,

Atul