Re: [MiNT] New object-format

["Guido Flohr" <gufl0000@stud.uni-sb.de>]

Hi,

On Tue, Jun 29, 1999 at 09:18:40AM +0100, Jo-Even.Skarstein@gjensidige.no wrote:
> Hi!
> 
> Craig Graham once made a source-level debugger for MiNT, but it only
> supported Lattice-binaries. Can something like this be done with the new
> object-format and a debugger like Paral
> (http://www.franken.de/users/asrael/index.html)?

I'm in contact with the Paral developer.

Apart from that: I shouldn't be too difficult to build a gdb port based on
the BFD backend for MiNT.  The tracing stuff could probably be taken from
the current gdb port, the other thing left to do would be to teach gdb
that the program image does not reside at the start of the text segment in
memory but that you have to add the address of the BASEPAGE structure to
it.

This is merely a time problem.  The gdb sources are about 11 MB (in gzip
format) and that smells like some work. ;-)

For other debuggers: The debugging information is now present in the
executable (if compiled with -g and not stripped) and the BFD library has
functions to calculate source line numbers from addresses.

Until we have a full-blown new debugger you can still use the old one (GDB
3.6- (tiemann) (AtariST PL4)).  Things are even simpler now.  You don't
have to relink using sym-ld, you can extract the symbol file at any point
from the executable using the new `symex' command.  It's just that this
old gdb has difficulties with the new debugging symbol types.  But that is
not a problem of the new program format, it would also appear if you were
using sym-ld.  Anyway, debugging still works, it has even become easier.

If your favorite debugger/disassembler prefers the old DR symbol table you
can also link with the option `--traditional-format'.  At least Desert
Drain is happy then, I don't see a reason why other programs shouldn't be.

Ciao

Guido
-- 
http://stud.uni-sb.de/~gufl0000
mailto:gufl0000@stud.uni-sb.de