From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 21544 invoked by alias); 2 Sep 2003 20:14:03 -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 21532 invoked by uid 48); 2 Sep 2003 20:14:02 -0000 Date: Tue, 02 Sep 2003 20:14:00 -0000 Message-ID: <20030902201402.21531.qmail@sources.redhat.com> From: "bgreen at nas dot nasa dot gov" To: gcc-bugs@gcc.gnu.org In-Reply-To: <20030618234434.11247.bgreen@nas.nasa.gov> References: <20030618234434.11247.bgreen@nas.nasa.gov> Reply-To: gcc-bugzilla@gcc.gnu.org Subject: [Bug c++/11247] -frepo fails to instantiate typeinfo X-Bugzilla-Reason: CC X-SW-Source: 2003-09/txt/msg00131.txt.bz2 List-Id: PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org. http://gcc.gnu.org/bugzilla/show_bug.cgi?id=11247 ------- Additional Comments From bgreen at nas dot nasa dot gov 2003-09-02 20:14 ------- I have uncovered what appears to be the cause for this bug. I hope this is helpful in some way. When the compiler processes the dynamic_cast, the repo code emits the following into the .rpo file: O _ZTV5ITypeIiE running c++filt on this shows that this refers to the class's vtable: O vtable for IType However, the linker (run via collect2) produces an error indicating a missing 'typeinfo', but not a missing 'vtable': GNU ld version 2.14 20030612 test.o(.text+0x32): In function `main': /home/bgreen/code/tests/comcasts/tex3/test.C:23: undefined reference to `typeinfo for IType' collect2 is then unable to match the unresolved typeinfo to the uninstantiated vtable. It seems that the problem boils down to the compiler repo code emitting a 'vtable' where it should emit a 'typeinfo' in response to a dynamic cast.