From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 23173 invoked by alias); 8 Oct 2002 18:46:02 -0000 Mailing-List: contact gcc-prs-help@gcc.gnu.org; run by ezmlm Precedence: bulk List-Archive: List-Post: List-Help: Sender: gcc-prs-owner@gcc.gnu.org Received: (qmail 23150 invoked by uid 71); 8 Oct 2002 18:46:02 -0000 Date: Tue, 08 Oct 2002 11:46:00 -0000 Message-ID: <20021008184602.23149.qmail@sources.redhat.com> To: jakub@gcc.gnu.org Cc: gcc-prs@gcc.gnu.org, From: Janis Johnson Subject: Re: optimization/6631: Miscompiled structure access Reply-To: Janis Johnson X-SW-Source: 2002-10/txt/msg00298.txt.bz2 List-Id: The following reply was made to PR optimization/6631; it has been noted by GNATS. From: Janis Johnson To: gcc-gnats@gcc.gnu.org Cc: Subject: Re: optimization/6631: Miscompiled structure access Date: Tue, 8 Oct 2002 11:39:50 -0700 This is a regression from GCC 3.0 that was exposed by the following cp patch: 2002-04-30 Mark Mitchell ABI change, returning simple classes from functions. * class.c (finish_struct_bits): Only mark TREE_ADDRESSABLE if TYPE_HAS_TRIVIAL_INIT_REF is false or TYPE_HAS_NONTRIVIAL_DESTRUCTOR is true. The test passes with optimization on if it is altered by adding a destructor or a copy constructor to class QSize, or by removing "const" from "inline const QSize operator+". I'm not planning to investigate this further, but hope this additional information will be useful. Janis