Backtrace in uClibC

[Carmelo Amoroso]

On 18/01/11 17:21, ANDY KENNEDY wrote:
>> On 12/01/11 17:50, ANDY KENNEDY wrote:
>>> I noticed that almost exactly a year ago Carmelo Amoroso posted
>> stating
>>> that he hoped to do more work on backtrace.  After looking at
>>> libubactrace/sh/backtrace.c, I'm confused as to why this is platform
>>> dependant code.  The code appears to ONLY use library calls that
>> should
>>> be available on _any_ platform.  I have copied the code to a mips
>>> subdirectory and am testing at the moment, but I'm curious as to
> what
>>> potential problems I'll find; due to my wealth of knowledge of
>> backtrace
>>> (that would be 0) I'm sure I don't know why this would not work.
>>>
>>> If you have any information, please enlighten me!
>>>
>>> Thanks,
>>> Andy
>>
>> Hi Andy,
>> likely you are right and the current sh implementation can be widely
>> used.
>> Let me a few time to review this code.
>> In the meanwhile you're feedback with your try n mips would be
> helpful.
> 
> It failed.  The message was that I needed to have mips specific code to
> include backtrace support.

Sure, I've added the #error directive.
Could you try copying the backtrace.c from the sh directory in the root and give it a try ?
Ensure that the backtrace.c is built with -fexception (see Makefile.arch in the sh folder)

> I worked around the issue by rolling back
> the version of the package I was attempting to use.  If I get time I'll
> attempt it again.
> 
> Andy
> 

I know have an ARM target to use, I'll try to test the sh implementation (simply based on dwarf2) to see if it can
be exported to all archs.

Cheers,
Carmelo