This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: [BUG] Regression in 2.14.90 (relative to 2.13.90)
On Wed, Nov 26, 2003 at 04:35:50PM +0000, Nick Clifton wrote:
> Hi Daniel, Hi HJ
>
> >> Is there a way to tell GDB that a particular line table entry is
> >> invalid ? Or is the linker supposed to delete such entries entirely
> >> from the line table ?
> >
> > I believe the latter. Can anyone see a use of line table entries which
> > do not correspond to used code? I can't think of an example.
>
> HJ - what is the reason behind your patch to preserve as much debug
> info as possible ? It seems that including this information is what
> it causing the problem, and ideally what we ought to be doing is
> discarding .gnu.linkonce sections containing debug info, rather than
> retaining them.
The root problem is ld didn't discard it. It put 0 there instead, which
doesn't work too well. My testcase shows it. It is the problem which
can be fixed with SHT_GROUP. My change solves most, but not all, of the
problem.
H.J.