[Dwarf-Discuss] question about [0x0..0xffffffff] location list entry

Joel Brobecker brobecker at adacore.com
Thu Jan 20 07:05:25 PST 2011


> Assuming there is no previous base address entry in that location list,
> then that's roughly correct.

I confirm that there was only 1 entry in that list, and that was the
normal location list entry.

> What makes it "rough" is that I don't think it's at all clear what
> overflowing/wrapping address computations are supposed to mean.  It
> seems highly dubious that the producer should ever emit values wherein
> the additions the consumer does would overflow like this.

Indeed.

I've trying to convince the producer that the debugging info they
generated is wrong on the basis that "it doesn't say what you want
it to say", but we also agree that it's peculiar at best. If we can
agree that overflow situation are erroneous in the sense that
consumers are allowed to do whatever they want, that would be good
enough for me.  However, if we decide to define what this means,
then I can adjust GDB to follow that decision...

Thanks for the feedback.
-- 
Joel



More information about the Dwarf-Discuss mailing list