From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 88714 invoked by alias); 26 Jan 2020 16:15:47 -0000 Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org Received: (qmail 88682 invoked by uid 89); 26 Jan 2020 16:15:43 -0000 Authentication-Results: sourceware.org; auth=none X-Spam-SWARE-Status: No, score=-8.4 required=5.0 tests=AWL,BAYES_00,RCVD_IN_DNSWL_NONE,SPF_HELO_PASS autolearn=ham version=3.3.1 spammy= X-HELO: gateway32.websitewelcome.com Received: from gateway32.websitewelcome.com (HELO gateway32.websitewelcome.com) (192.185.145.115) by sourceware.org (qpsmtpd/0.93/v0.84-503-g423c35a) with ESMTP; Sun, 26 Jan 2020 16:15:41 +0000 Received: from cm17.websitewelcome.com (cm17.websitewelcome.com [100.42.49.20]) by gateway32.websitewelcome.com (Postfix) with ESMTP id 74D7BF52FA for ; Sun, 26 Jan 2020 10:15:39 -0600 (CST) Received: from box5379.bluehost.com ([162.241.216.53]) by cmsmtp with SMTP id vkZXiTCtGmgNjvkZXiZOXh; Sun, 26 Jan 2020 10:15:39 -0600 DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=tromey.com; s=default; h=Content-Type:MIME-Version:Message-ID:In-Reply-To:Date: References:Subject:Cc:To:From:Sender:Reply-To:Content-Transfer-Encoding: Content-ID:Content-Description:Resent-Date:Resent-From:Resent-Sender: Resent-To:Resent-Cc:Resent-Message-ID:List-Id:List-Help:List-Unsubscribe: List-Subscribe:List-Post:List-Owner:List-Archive; bh=z+7JDiMzVu7IOso1MVlfAp9X/z2xFAWPJsPlpDFBKmI=; b=Mcm1jLCITX5cZH1eep+lUUoeYy 0b1QcgB+bvHvIzSVCrIvJa2tFBLdVa7pBEKf5kHgUo1/aTswS8IRuUXsTw+2inPyvnBReHZU26/Ls zJRAKcrt+K+aFY4blrhAJlRL/; Received: from 75-166-123-50.hlrn.qwest.net ([75.166.123.50]:57742 helo=bapiya) by box5379.bluehost.com with esmtpsa (TLSv1.2:ECDHE-RSA-AES256-GCM-SHA384:256) (Exim 4.92) (envelope-from ) id 1ivkZX-003TKx-80; Sun, 26 Jan 2020 09:15:39 -0700 From: Tom Tromey To: Andrew Burgess Cc: gdb-patches@sourceware.org Subject: Re: [PATCH] gdb: Reinitialize objfile::section_offsets during objfile reload References: <20200125225555.16846-1-andrew.burgess@embecosm.com> Date: Sun, 26 Jan 2020 16:33:00 -0000 In-Reply-To: <20200125225555.16846-1-andrew.burgess@embecosm.com> (Andrew Burgess's message of "Sat, 25 Jan 2020 22:55:55 +0000") Message-ID: <875zgy6vo5.fsf@tromey.com> User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/28.0.50 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain X-SW-Source: 2020-01/txt/msg00851.txt.bz2 >>>>> "Andrew" == Andrew Burgess writes: Andrew> When building and testing with '-D_GLIBCXX_DEBUG=1' I noticed that the Andrew> test gdb.base/reload.exp was failing. This turns out to be because Andrew> the objfile::section_offsets vector is not reinitialilzed during the Andrew> objfile reload process, and in this particular test, GDB ends up Andrew> indexing outside the bounds of the vector. Thanks for catching this. Andrew> One thing I did wonder about while looking at this fix is whether it Andrew> would be possible to combine at least parts of syms_from_objfile_1 Andrew> with the core of reread_symbols. I did have a go at doing this but Andrew> gave up in the end due to the subtle differences between the two. Andrew> Still, I think that with some more effort this might be possible, and Andrew> this could be a nice clean up. A long time ago, Jan had a patch along these lines. I believe what his did was just throw away the logic in reread_symbols in favor of simply creating a new objfile. I wonder if it's too late to do this now, since objfiles are exposed to Python. Anyway, IMO, if there are subtle differences, they are probably bugs of some sort; and unifying these code paths seems like clearly the right thing to do. Andrew> + /* In syms_from_objfile_1 after calling objfile_set_sym_fns we Andrew> + handle the possibility that objfile->sf might be NULL, which Andrew> + can happen for some obscure objfile formats. We've never Andrew> + handled the NULL case here before, but */ This looks like it got cut off. Andrew> + /* Setup the section offsets structure for this objfile. We use Andrew> + zero section address information here, though it's not clear Andrew> + this will always be correct. If the user originally loaded Andrew> + this objfile with non-zero address information then we're Andrew> + going to loose that here. */ s/loose/lose/ I don't know what else would make sense in this case. Warn the user? The patch seems otherwise reasonable to me. Tom