[U-Boot-Users] bl cpu_init_f jumps to an wrong point.

[U-Boot-Users] bl cpu_init_f jumps to an wrong point.

[Nekto Peresmeshnik]

Hello community,

I've problem with jump to cpu_init_f. I trace the begin of u-boot execution,
i.e. assembler code. It is okay, until it achieves the jump to cpu_init_f.
Then it jumps to a random point. "Random" means that it is same point for
every compilation. After next compilation this point changes randomly.

The picture is next: in debugger I see that cpu_init_f placed on the address
0xfff06924.
I see same when I do disassemble the u-boot image.

Please hint where may I be mistaken?

The board: custom 8247A based, 2 M. flash.
Compilation: configured and compiled for 8260. Neither error nor warnings
during compilation.
Toolchain: ELDK version 3.1.1 Build 2005-06-07
Linux: SuSE 9.3 kernel 2.6.11.4-21.9-default
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.denx.de/pipermail/u-boot/attachments/20060104/3c1d8abc/attachment.htm 

[U-Boot-Users] bl cpu_init_f jumps to an wrong point.

[Jerry Van Baren]

Nekto Peresmeshnik wrote:
> Hello community,
> 
> I've problem with jump to cpu_init_f. I trace the begin of u-boot 
> execution, i.e. assembler code. It is okay, until it achieves the jump 
> to cpu_init_f. Then it jumps to a random point. "Random" means that it 
> is same point for every compilation. After next compilation this point 
> changes randomly.
> 
> The picture is next: in debugger I see that cpu_init_f placed on the 
> address 0xfff06924.
> I see same when I do disassemble the u-boot image.
> 
> Please hint where may I be mistaken?
> 
> The board: custom 8247A based, 2 M. flash.
> Compilation: configured and compiled for 8260. Neither error nor 
> warnings during compilation.
> Toolchain: ELDK version 3.1.1 Build 2005-06-07
> Linux: SuSE 9.3 kernel 2.6.11.4-21.9-default

It undoubtedly is _not_ a random address, it is where the linker placed 
cpu_init_f that u-boot is jumping to.  My guess is that the problem is 
that you didn't program the code where it belongs.  The reset vector is 
getting programmed properly but the code (BR0 flash mapping) is not 
where it is expected by the linker.

Check your configuration, especially where you placed flash in your 
linker control file vs. where flash is on power up vs. where you 
configured BR0/OR0 to point to.

gvb