Hi Dmitry,

Thanks for the fast response.

>
> Well, windoze build is a bit outdated, cause recently we have fixed some
bugs
> which might be vital. So, try to build it yourself or ask the community to
build
> a recent snap under win.

Well, so far I've only tried once, and that was on my home Win98 machine
(./configure takes forever under Win98).  I'll have a shot later on my work
(W2K) machine, which is generally a lot more reliable for such builds, and
I'll use the latest cvs snapshots.

> >
> BUt you can run built-in gdb simulator.
>

I'll have a look at that.  Presumably that needs gdb patches that are also
on the cvs?

>
> > 2) Generate iar-type debugging information, probably as the target of an
> > msp430-objcopy.  I don't know if enough information is available on
IAR's
> > format to make this possible.
>
> Then you'll have to debug this in assembler mode I think.
>
> >
> > 3) Generate iar assembler files from gcc instead of gas-compatible
assembly
> > files, with the original C code as comments, and use iar's assembler and
> > linker.  This will at least give far better assembly-level debugging
using
> > C-Spy, with at least basic variable watch capability.
>
> gcc does not generate IARs code. It does not include C lines as comments.

That's what I feared.  I have always felt that gcc was missing a good mixed
C/assembly listing output - when doing avr programming using ImageCraft, I
often use its C/assembly listing to be able to quickly see what sort of code
is being produced.  With gcc for the m68k, I sometimes resort to using gdb
and the "disass" command to get the best mix.

>
> >
> > 4) Make a gdb back-end that uses the kickstart jtag adaptor.  This is,
of
> > course, the best solution in the long run, since it is cross-platform
and
> > gives a complete indepentant developement tool set.  This is the route
taken
> > for m68k debugging - there is a gdb back-end that for the m68k using bdm
> > (which is roughly like jtag) adaptors.  I combine it with gvd to give a
nice
> > gui under windows and *nix.
> >
>
> TIs jtag uses propretary interface which is not published yet. But gdb is
prepared
> for it.
>
>
> >
> > Has anyone worked towards any of these solutions?  All I have found out
so
> > far is that there is a work in progress on jtag debugging (presumably
that
> > means gdb), but I'd love to know how things are going - we are going to
have
> > to see whether we can use msp430 gcc for development, or whether we
should
> > wait for other compilers (like Imagecraft - we have been very happy with
> > their avr compiler) which are easier for debugging.
>
> We do... Thing are going well - We do use gcc for development :)
>

What do you use for your debugging?  Burn-and-test debugging has a lot of
uses and in later stages of projects it is often the best method, but
especially at the start of projects it is extremly useful to be able to use
single-stepping, breakpoints, variable watching, and the like.

>
> The only thing I wait is jtag interface, cause the rest works.
>

I understood that the jtag interface specification was available, but only
under an NDA.  Is that correct?  Or if not, have you any idea about when you
might be expecting to get the information?  Would it help to do some
lobbying (our ti distributers here in Norway are very good at passing
complaints and requests up through the chain of command)?

Best regards,

David



Reply via email to