From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 17026 invoked by alias); 28 May 2003 08:55:18 -0000 Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-owner@gcc.gnu.org Received: (qmail 16941 invoked from network); 28 May 2003 08:55:15 -0000 Received: from unknown (HELO motgate2.mot.com) (136.182.1.10) by sources.redhat.com with SMTP; 28 May 2003 08:55:15 -0000 Received: from az33exr03.mot.com (pobox3.mot.com [10.64.251.242]) by motgate2.mot.com (Motorola/Motgate2) with ESMTP id h4S8tEAU016993 for ; Wed, 28 May 2003 01:55:14 -0700 (MST) Received: from latour.rsch.comm.mot.com (latour.rsch.comm.mot.com [145.1.80.116]) by az33exr03.mot.com (Motorola/az33exr03) with ESMTP id h4S8tC9g007794 for ; Wed, 28 May 2003 03:55:12 -0500 Received: from latour.rsch.comm.mot.com (localhost.rsch.comm.mot.com [127.0.0.1]) by latour.rsch.comm.mot.com (8.12.9/8.12.9) with ESMTP id h4S8tC1o014114 for ; Wed, 28 May 2003 03:55:12 -0500 (CDT) (envelope-from rittle@latour.rsch.comm.mot.com) Received: (from rittle@localhost) by latour.rsch.comm.mot.com (8.12.9/8.12.9/Submit) id h4S8tBv7014113; Wed, 28 May 2003 03:55:11 -0500 (CDT) Date: Wed, 28 May 2003 12:51:00 -0000 From: Loren James Rittle Message-Id: <200305280855.h4S8tBv7014113@latour.rsch.comm.mot.com> To: gcc@gcc.gnu.org Subject: Re: Failure to run gcc 3.4 cvs HEAD libstdc++ testsuite on i686-linux with -fPIC: ICEs In-Reply-To: <20030526101028.0325f1bc.bkoz@redhat.com> Organization: Networks and Infrastructure Lab (IL02/2240), Motorola Labs Cc: X-SW-Source: 2003-05/txt/msg02330.txt.bz2 Benjamin Kosnik wrote: > In the meantime you have two options, either add another PCH multilib to > the libstdc++-v3/include/Makefile.am, so that -fPIC varients are built > in addition to -O0 and -O2, or disable the use of pch in the libstdc++ > testsuite via manipulation/removal of PCHFLAGS in testsuite_flags. I investigated adding additional variants. It doesn't work since the order that PCH files are considered is psuedo-randomized in a manner that the crash on -fpic/no -fpic incompatibility will still hit. (However, now you would see a psuedo-randomized pattern of test case failures instead of 100% ;-) - Loren