From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 30017 invoked by alias); 15 Jun 2009 16:52:07 -0000 Received: (qmail 29967 invoked by alias); 15 Jun 2009 16:51:55 -0000 Date: Mon, 15 Jun 2009 16:52:00 -0000 Message-ID: <20090615165155.29966.qmail@sourceware.org> X-Bugzilla-Reason: CC References: Subject: [Bug libfortran/40344] O32 libgfortran.so fails to link on IRIX 6.5 In-Reply-To: Reply-To: gcc-bugzilla@gcc.gnu.org To: gcc-bugs@gcc.gnu.org From: "ro at techfak dot uni-bielefeld dot de" Mailing-List: contact gcc-bugs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Id: List-Archive: List-Post: List-Help: Sender: gcc-bugs-owner@gcc.gnu.org X-SW-Source: 2009-06/txt/msg00966.txt.bz2 ------- Comment #3 from ro at techfak dot uni-bielefeld dot de 2009-06-15 16:51 ------- Subject: Re: O32 libgfortran.so fails to link on IRIX 6.5 rwild at gcc dot gnu dot org writes: > Would it help to work around this issue by collecting the objects in one or > more > convenience archives and linking those (plus at least one plain object) > together > to form libgfortran.la? The intermediate partially linked objects could hit a > different code path in the linker (but could also cause for some less-optimal > linking, at least with LTO?). Of course 'ld -r' has been known to expose other > linker bugs so this might be trading one issue for an unknown other set... A quick test indicates that this doesn't help. I'm inclined to just drop O32 ABI support given SGIs inability to fix (or even explain) the issue. The N32/N64 linker is completely different and not hampered by such problems. Rainer -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40344