[okl4-developer] Debugging tips for a new platform

[Joshua Root]

Geoffrey Lee wrote:
> On Thu, Jun 19, 2008 at 12:53:09PM +1000, Joshua Root wrote:
>> Hi all,
>>
>> I'm porting OKL4 2.1.1-fix7 to the Linksys NSLU2 (aka Slug), which is a 
>> big-endian XScale platform, IXP420-based. I've referenced the ixp4xx 
>> platform code from the old N1 version of L4, and the current pxa code, 
>> and have produced something which compiles and links, but the image 
>> produces no serial output when run.
> 
> Josh - 
> 
> There is a platforms/pxa/tools/machines.py where you put specific
> platform description for the PXA.  This tells it where the memory
> is located, as well as other platform-specific descriptions.  Most
> of these are for Iguana, though.

Right, I actually have a new ixp4xx platform subdir. I'd be interested 
in documentation for the attributes you can set in machines.py actually, 
particularly which ones affect pistachio.

> The head.spp should work unmodified, we utilize the PXA-based
> gumstix extensively.  That runs in little endian though, but 
> I don't think that is the specific issue.

Right, I have endian = "big" so it looks like it should do the right 
thing. I was just wondering because the slug is a little different to 
the gumstix.

> I just took a brief look at the Skyeye source code and it seems there
> is support for running in big endian mode.  If you specify the -e
> parameter it infers it from the ELF file automatically.  You may
> want to take a look whether it is worth trying out Skyeye, as
> it can give you instruction traces, and from there, you can narrow
> down where to look further.

I actually tried that, but it dies with "elf_checkFile failed: -3". Any 
idea what that means? Getting it running in the simulator would 
certainly be the quickest way forward.

I've put up a diff between the 2.1.1-fix7 release and what I have now, 
if you want to have a look: 
<http://www.cse.unsw.edu.au/~jmr/okl4-ixp4xx.diff>

Cheers,
Josh