From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 5885 invoked by alias); 21 Dec 2005 10:30:23 -0000 Received: (qmail 5867 invoked by uid 22791); 21 Dec 2005 10:30:22 -0000 X-Spam-Check-By: sourceware.org Received: from md.dent.med.uni-muenchen.de (HELO md.dent.med.uni-muenchen.de) (138.245.179.2) by sourceware.org (qpsmtpd/0.31) with SMTP; Wed, 21 Dec 2005 10:30:22 +0000 Received: (qmail 3224 invoked by uid 211); 21 Dec 2005 10:30:17 -0000 Date: Wed, 21 Dec 2005 10:30:00 -0000 Message-ID: <20051221103017.3223.qmail@md.dent.med.uni-muenchen.de> From: Wolfram Gloger To: libc-hacker@sources.redhat.com In-reply-to: <20051219222554.GL4625@sunsite.mff.cuni.cz> (message from Jakub Jelinek on Mon, 19 Dec 2005 23:25:54 +0100) Subject: Re: [PATCH] Fix mtrace vs. mudflap coexistence References: <20051219222554.GL4625@sunsite.mff.cuni.cz> Mailing-List: contact libc-hacker-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: libc-hacker-owner@sourceware.org X-SW-Source: 2005-12/txt/msg00037.txt.bz2 Hi, > When an application linked with alternate malloc implementation that > internally calls the glibc (next) malloc (such as when built with > mudflap) is mtraced and the alternate malloc doesn't allow reentrancy, > the application hangs. An alternative viewpoint would be that the alternate malloc implementation which _knows_ it is calling the glibc malloc (and which _knows_ that itself is not reentrant-safe...) needs to reset the glibc malloc hooks to zero before calling dlsym (RTLD_NEXT, "malloc") and friends. Of course this would lose the ability to mtrace that particular alternate malloc implementation, however with your patch we effectively lose the ability to mtrace _any_ alternate malloc implementation. I'm unsure which is better. Regards, Wolfram.