This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug libstdc++/69310] [6 Regression] Revision r232454 breaks bootstrap on x86_64-apple-darwin15
- From: "torvald at gcc dot gnu.org" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: Wed, 20 Jan 2016 11:45:49 +0000
- Subject: [Bug libstdc++/69310] [6 Regression] Revision r232454 breaks bootstrap on x86_64-apple-darwin15
- Auto-submitted: auto-generated
- References: <bug-69310-4 at http dot gcc dot gnu dot org/bugzilla/>
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69310
--- Comment #11 from torvald at gcc dot gnu.org ---
(In reply to Jack Howarth from comment #10)
> It is unclear if the changes in r232454, to avoid the explicit linkage on
> libitm, can ever be made darwin-friendly. On darwin, every single executable
> linked against libstdc++ would require -Wl,-undefined,dynamic_lookup on the
> linkage to avoid linkage errors from the undefined symbols from libitm
> within libstdc++.
Is Darwin declaring __GXX_WEAK__? I suppose that it does because after r232539
the failing code isn't compiled unless __GXX_WEAK__ is supported. But if it
is, weak references must be supported. Or are you saying that specifically the
alias attribute is a problem?
If indeed weak references to symbols not declared are a problem, then we
probably should disable the transactional clones the same way we did on AIX,
which doesn't support weak references without definitions either.
What about putting this:
// No support for referencing weak symbols without a definition.
#define _GLIBCXX_USE_WEAK_REF 0
into ./config/os/bsd/darwin/os_defines.h? This would disable the transactional
clones altogether. Could you give this a try?