From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 12194 invoked by alias); 24 Oct 2002 06:56:02 -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 12162 invoked by uid 71); 24 Oct 2002 06:56:01 -0000 Resent-Date: 24 Oct 2002 06:56:01 -0000 Resent-Message-ID: <20021024065601.12161.qmail@sources.redhat.com> Resent-From: gcc-gnats@gcc.gnu.org (GNATS Filer) Resent-Cc: gcc-prs@gcc.gnu.org, gcc-bugs@gcc.gnu.org Resent-Reply-To: gcc-gnats@gcc.gnu.org, davmac@ozonline.com.au Received: (qmail 7401 invoked by uid 61); 24 Oct 2002 06:53:48 -0000 Message-Id: <20021024065348.7400.qmail@sources.redhat.com> Date: Wed, 23 Oct 2002 23:56:00 -0000 From: davmac@ozonline.com.au Reply-To: davmac@ozonline.com.au To: gcc-gnats@gcc.gnu.org X-Send-Pr-Version: gnatsweb-2.9.3 (1.1.1.1.2.31) Subject: other/8335: problems with __builtin_frame_address when -fomit-frame-pointer used X-SW-Source: 2002-10/txt/msg00928.txt.bz2 List-Id: >Number: 8335 >Category: other >Synopsis: problems with __builtin_frame_address when -fomit-frame-pointer used >Confidential: no >Severity: non-critical >Priority: low >Responsible: unassigned >State: open >Class: change-request >Submitter-Id: net >Arrival-Date: Wed Oct 23 23:56:00 PDT 2002 >Closed-Date: >Last-Modified: >Originator: Davin McCall >Release: 3.2 >Organization: >Environment: i686-pc-linux-gnu running linux 2.4.18. gcc version 3.2 (and 3.1.1, others not checked) >Description: When using -fomit-frame-pointer option, the builtin function __builtin_frame_address doesn't always work correctly (at least with a 0 argument). This is something of a no-brainer anyway seeing as one probably shouldn't request the frame address when one has explicitly requested no frame pointers, but I couldn't find any reference in the documentation which mentioned this; I should think, in any case, that the compiler should give an error or at least a warning when an attempt to use __builtin_frame_address with -fomit-frame-pointer. As an alternative, any function which uses the builtin could be forced to have a frame generated, OR at least an attempt to return a meaningful or "near-enough" value (such as %esp) could be made. I assume that this is just an oversight, however if it is actually a bug I can provide a preprocessed source file which reproduces it (if needed). >How-To-Repeat: >Fix: >Release-Note: >Audit-Trail: >Unformatted: