From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 11021 invoked by alias); 24 Feb 2006 22:41:21 -0000 Received: (qmail 10973 invoked by uid 22791); 24 Feb 2006 22:41:20 -0000 X-Spam-Check-By: sourceware.org Received: from nevyn.them.org (HELO nevyn.them.org) (66.93.172.17) by sourceware.org (qpsmtpd/0.31.1) with ESMTP; Fri, 24 Feb 2006 22:41:19 +0000 Received: from drow by nevyn.them.org with local (Exim 4.54) id 1FClch-0001mu-OR; Fri, 24 Feb 2006 17:41:15 -0500 Date: Fri, 24 Feb 2006 23:10:00 -0000 From: Daniel Jacobowitz To: Michael Snyder Cc: gdb-patches@sourceware.org Subject: Re: Some more multi-fork fixes Message-ID: <20060224224115.GA6792@nevyn.them.org> Mail-Followup-To: Michael Snyder , gdb-patches@sourceware.org References: <20060115190757.GA10454@nevyn.them.org> <43DA7BFE.2070100@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <43DA7BFE.2070100@redhat.com> User-Agent: Mutt/1.5.8i X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2006-02/txt/msg00466.txt.bz2 On Fri, Jan 27, 2006 at 12:01:02PM -0800, Michael Snyder wrote: > >Some other problems I noticed, but did not fix: > > > >The "follow child, print pids" test is victim of a race condition; > >sometimes > >the "$pid done" printout from the inferior will be interleaved with the GDB > >output, causing expect to mark the test as a failure. > > > >delete-fork really ought to call waitpid after deleting the fork; otherwise > >the process lives on as a zombie. > > That's fine with me too. Well, I'd been sort of hoping that you (as the developer who contributed this huge pile of new code) would take care of its reported bugs. I'm up against a bit of a brick wall at the moment. I've got patches in my source tree to unify the single-threaded and multi-threaded code in the Linux native layer into a single combined target. But since this is an inherently single-threaded operation the multi-threaded code totally falls down - switching forks doesn't preserve things like the LWP list, leading to messy timeouts. I'm going to find some minimally painful way to make the two of them talk to each other. -- Daniel Jacobowitz CodeSourcery