This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
[Bug c++/16027] Enabling optimisation causes segfault
- From: "s_gccbugzilla at nedprod dot com" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 17 Jun 2004 18:46:37 -0000
- Subject: [Bug c++/16027] Enabling optimisation causes segfault
- References: <20040616212630.16027.s_gccbugzilla@nedprod.com>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Additional Comments From s_gccbugzilla at nedprod dot com 2004-06-17 18:46 -------
(In reply to comment #7)
> Even if it were a larger testcase that would help us.
ICC with -cxxlib-gcc exhibits precisely the same behaviour as GCC (works when
non optimised, segfaults when optimised) - therefore it is not a bug in GCC.
I'll have to keep looking.
> As to why the compiler is using a different function in one library than
> in another: the library prohibits you from redefining a global function to
> be different in two translation units (that includes the source files from
> which libstdc++ is compiled on the one hand, and your application program
> on the other hand). If you violate this, you have to expect almost everything.
> Note in particular, that you can't control whether inlining is used inside
> libstdc++, since this is compiled when the compiler itself is built, not
> when you compile your application.
I've been careful not to override malloc, calloc, realloc or free - my versions
live within a namespace. And global operator new and delete is guaranteed by the
C++ standard to be user replaceable which suggests to me it must be always
treated as a weak symbol by GCC. However perhaps me marking it as inline has
deleterious effects (if so, perhaps GCC should reject inline operator new?).
Cheers,
Niall
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=16027