From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 4720 invoked by alias); 4 Apr 2005 01:30:26 -0000 Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org Received: (qmail 4691 invoked by uid 48); 4 Apr 2005 01:30:24 -0000 Date: Mon, 04 Apr 2005 01:30:00 -0000 Message-ID: <20050404013024.4689.qmail@sourceware.org> From: "mmitchel at gcc dot gnu dot org" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20041226162616.19159.danglin@gcc.gnu.org> References: <20041226162616.19159.danglin@gcc.gnu.org> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug c++/19159] [4.0/4.1 Regression] Undefined symbol: vtable for __cxxabiv1::__vmi_class_type_info X-Bugzilla-Reason: CC X-SW-Source: 2005-04/txt/msg00263.txt.bz2 List-Id: ------- Additional Comments From mmitchel at gcc dot gnu dot org 2005-04-04 01:30 ------- Dave, I think that the problem with _S_do_it is an entirely separate bug. In particular, V3 builds with -fno-inline-templates, but does not use -fno-implicit-inline-templates. The _S_do_it function is inline. Since AIX does not have weak symbols, we probably make that function have internal linkage in the library, which causes the link failures you're seeing. To confirm that, would you please send me the preprocessed version of both valarray-inst.cc and header_cassert.cc, together with the cc1plus command-line used on AIX to compile each? Thanks, -- Mark -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19159