From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtp.polymtl.ca (smtp.polymtl.ca [132.207.4.11]) by sourceware.org (Postfix) with ESMTPS id 3F04D3857C56 for ; Tue, 6 Apr 2021 01:36:31 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.3.2 sourceware.org 3F04D3857C56 Received: from simark.ca (simark.ca [158.69.221.121]) (authenticated bits=0) by smtp.polymtl.ca (8.14.7/8.14.7) with ESMTP id 1361aPvY001732 (version=TLSv1/SSLv3 cipher=ECDHE-RSA-AES256-GCM-SHA384 bits=256 verify=NOT) for ; Mon, 5 Apr 2021 21:36:30 -0400 DKIM-Filter: OpenDKIM Filter v2.11.0 smtp.polymtl.ca 1361aPvY001732 Received: from [10.0.0.11] (192-222-157-6.qc.cable.ebox.net [192.222.157.6]) (using TLSv1.3 with cipher TLS_AES_128_GCM_SHA256 (128/128 bits) key-exchange X25519 server-signature RSA-PSS (2048 bits) server-digest SHA256) (No client certificate requested) by simark.ca (Postfix) with ESMTPSA id C3CEA1E590 for ; Mon, 5 Apr 2021 21:36:25 -0400 (EDT) Subject: Re: [PATCH] sim/m32c: fix memory leaks in opc2c To: gdb-patches@sourceware.org References: <20210405145856.3925296-1-simon.marchi@polymtl.ca> From: Simon Marchi Message-ID: Date: Mon, 5 Apr 2021 21:36:25 -0400 User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:78.0) Gecko/20100101 Thunderbird/78.7.1 MIME-Version: 1.0 In-Reply-To: Content-Type: text/plain; charset=utf-8 Content-Language: en-US Content-Transfer-Encoding: 7bit X-Poly-FromMTA: (simark.ca [158.69.221.121]) at Tue, 6 Apr 2021 01:36:25 +0000 X-Spam-Status: No, score=-4.2 required=5.0 tests=BAYES_00, DKIM_SIGNED, DKIM_VALID, DKIM_VALID_AU, DKIM_VALID_EF, NICE_REPLY_A, RCVD_IN_MSPIKE_H3, RCVD_IN_MSPIKE_WL, SPF_HELO_PASS, SPF_PASS, TXREP autolearn=ham autolearn_force=no version=3.4.2 X-Spam-Checker-Version: SpamAssassin 3.4.2 (2018-09-13) on server2.sourceware.org X-BeenThere: gdb-patches@sourceware.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: Gdb-patches mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 06 Apr 2021 01:36:34 -0000 On 2021-04-05 5:51 p.m., Mike Frysinger wrote: > On 05 Apr 2021 14:46, Simon Marchi via Gdb-patches wrote: >> On 2021-04-05 12:23 p.m., Mike Frysinger wrote:> On 05 Apr 2021 10:58, Simon Marchi via Gdb-patches wrote: >>>> Fix the leak in dump_lines by free-ing the elements of varnames. >>> >>> i dislike stuffing a bunch of free's at the end of a program's lifetime just >>> to satisfy a tool that is not normally used. which isn't to say LSAN isn't >>> useful, just that i think we should do better. >> >> Why though? Because it adds execution time where not necessary? > > when the process exits, the kernel releases all the memory at once. there's > no point to calling free() on all your allocated buffers before exiting. it > only wastes time with the C library heap accounting & syscalls. > > i get that we're talking about opc2c here which is used only twice to generate > two other files, so in the larger scheme of things, it's barely noise. i'm > trying to define standard patterns we can apply in general for "the next one". I understand. I'm not that worried about that kind of performance hit though. >>> in other codebases, i've done things like: >>> #ifdef __SANITIZE_ADDRESS__ >>> # define ENABLE_CLEANUP_ONEXIT 1 >>> #else >>> # define ENABLE_CLEANUP_ONEXIT 0 >>> #endif >>> >>> then this could be written: >>> >>> if (ENABLE_CLEANUP_ONEXIT) { >>> for (i = 0; i < vn; i++) >>> free (varnames[i]); >>> } >>> >>> wdyt ? feel free to bikeshed the name. not sure if there's prior art in >>> the tree currently. >> >> I find this complexity a bit unnecessary, versus just free-ing the >> stuff. But I don't really mind, we can do as you like, I just want to >> my build to be fixed! >> >> Note that the igen tool doesn't free anything, so it's next on the list >> of things that break the -fsanizite=address build. I started to update >> it to free things, it's a bit tedious but it should be do-able. >> >> Another option would be to change the Makefile to call igen with the >> ASAN_OPTIONS=detect_leaks=0 environment variable. > > ah yes, this is exactly what i mean wrt "the tip of the iceberg" and it being > "a slippery slope" ;). first it's the small build tools, then the larger > build tools, then the programs we actually install, ... > > maybe an alternative would be to convert these to C++. then it would handle > stack/heap resources for us. If you convert to C++ and the memory is managed automatically, isn't it exactly the same (runtime-wise) as free-ing everything by hand in C? Although I'm always in favor of using C++ for just not having to think about free-ing stuff. Simon