This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Weird issue with .clone symbol
- From: Paolo Carlini <paolo dot carlini at oracle dot com>
- To: libstdc++ <libstdc++ at gcc dot gnu dot org>
- Cc: Richard Guenther <rguenther at suse dot de>, Jan Hubicka <hubicka at ucw dot cz>
- Date: Tue, 13 Oct 2009 19:07:29 +0200
- Subject: Weird issue with .clone symbol
Hi everyone,
today I tried once more to run check-parallel and in mainline I'm seeing
a curious issue with the following testcase:
27_io/manipulators/standard/wchar_t/2.cc (it's thus -O2
-D_GLIBCXX_PARALLEL -fopenmp -fno-implicit-templates in case you want to
check it outside the testsuite after replacing VERIFY -> assert)
In short, this symbol exists in the .o:
_ZNSbIwSt11char_traitsIwESaIwEEC1EPKwRKS1_.clone.1
instead of
_ZNSbIwSt11char_traitsIwESaIwEEC1EPKwRKS1_
for 4.4, and it wants:
_ZNSbIwSt11char_traitsIwESaIwEE12_S_constructIPKwEEPwT_S7_RKS1_St20forward_iterator_tag
which is not exported and, to be clear, never was in the past.
I think we could well export the latter for 4.5.0, but I want to make
sure we understand this .clone business, is there anything special we
should know, or it's just that this combination of command line options
(in particular the effects of -fno-implicit-templates can be quite
nasty) asks now, together with some changes in inlining, for one more
exported symbol, nothing special?
Thanks in advance,
Paolo.