More gcc/glibc weirdness

Greg Schafer gschafer at zip.com.au
Sat Feb 8 18:51:55 PST 2003


On Sat, Feb 08, 2003 at 05:30:01PM +0100, Matthias Benkmann wrote:
> Have you checked whether HAVE_DWARF2_UNWIND_INFO is or is not defined in
> the second case. Notice that the #undef is commented out, so it's not
> actually being #undef'd. It could be that on a normal system the
> HAVE_DWARF2_UNWIND_INFO is already imported from somewhere else and
> doesn't need to be #defined.

Yes. AFAICT, HAVE_DWARF2_UNWIND_INFO is meant to remain commented out on a
"sane" system with current tools. This equates with:-

checking for DWARF2 unwind info support... no_registry_needed

I found a thread on the glibc lists which went into some detail on the
matter. http://sources.redhat.com/ml/libc-hacker/2002-03/msg00076.html

So I am convinced with our current setup, we don't get the expected result
because of our Ch 5 gcc.

I have done the basic tests to confirm changing the switch to
--enabled-shared results in the correct result. But what I haven't confirmed
is building a full Ch 6 system to confirm everthing is still sane. It would
be even better to go even further and build some c++ libs (say KDE) to
confirm all is well.

I honestly think is not really a problem as most people do not appear to
have broken systems because of it. I would just feel better if we get rid of
any doubts.

Greg
-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-dev' in the subject header of the message



More information about the lfs-dev mailing list