On Wed, 2009-03-25 at 22:40 +0100, John Tytgat wrote:
> In message <1237983394.18642.36.ca...@duiker>
>           John-Mark Bell <[email protected]> wrote:

> > As it stands the current design of GDBServer is very much geared to
> > normal applications. Modules should be possible, assuming you've got a
> > sufficiently modern OS version (likely sit on the pre-init and
> > post-final services).
> 
> Maybe a different approach is more feasible: one of the last steps
> in the module creation process is to copy out the binary object blob
> out of the ELF file created by the linker.  In this process, the
> DWARF data gets dropped.
> 
> I'm wondering if GDBServer can't be teached to process this ELF jacket
> wrapped module himself allowing access to the DWARF data.  And it could
> link and delink the module in the module chain itself too instead of
> letting RISC OS doing this.

I guess that could be made to work. GDBServer itself couldn't care less
about the debugging data -- that's the responsibility of GDB itself
(this is why you can happily debug a stripped binary on a remote host if
the machine running gdb has a copy of the unstripped binary).

How much processing would be required on the ELF-wrapped module to get
it in a format RO will cope with? Is it overly naive to hope that it's
simply a case of calling the "load module from memory" OS_Module reason
code with an appropriate offset into the memory-resident binary?


J.


_______________________________________________
GCCSDK mailing list [email protected]
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK

Reply via email to