From owner-freebsd-current@FreeBSD.ORG Tue Mar 1 17:21:58 2011 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 2E9EA106566B; Tue, 1 Mar 2011 17:21:58 +0000 (UTC) (envelope-from dim@FreeBSD.org) Received: from tensor.andric.com (cl-327.ede-01.nl.sixxs.net [IPv6:2001:7b8:2ff:146::2]) by mx1.freebsd.org (Postfix) with ESMTP id E3D7D8FC08; Tue, 1 Mar 2011 17:21:57 +0000 (UTC) Received: from [IPv6:2001:7b8:3a7:0:348f:5488:ad29:7fa] (unknown [IPv6:2001:7b8:3a7:0:348f:5488:ad29:7fa]) (using TLSv1 with cipher DHE-RSA-CAMELLIA256-SHA (256/256 bits)) (No client certificate requested) by tensor.andric.com (Postfix) with ESMTPSA id 0304B5C59; Tue, 1 Mar 2011 18:21:56 +0100 (CET) Message-ID: <4D6D2B3A.5050907@FreeBSD.org> Date: Tue, 01 Mar 2011 18:22:02 +0100 From: Dimitry Andric Organization: The FreeBSD Project User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.2; en-US; rv:1.9.2.15pre) Gecko/20110227 Lanikai/3.1.9pre MIME-Version: 1.0 To: Doug Barton References: <11C58791-9103-4C0E-8B35-EE46A4A6D3D4@kientzle.com> <4D6B54F6.2000804@FreeBSD.org> In-Reply-To: <4D6B54F6.2000804@FreeBSD.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: freebsd-current@freebsd.org Subject: Re: Can't update CLang-based system X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 01 Mar 2011 17:21:58 -0000 On 2011-02-28 08:55, Doug Barton wrote: > I have a different problem on r219092. Everything builds find, but > "linking kernel.debug" hangs forever. It can't even be killed with ^C. > My existing system is r218985M, which was built with clang. This is my > first time trying to build a system with clang ON a system that was > itself built with clang (if that makes sense). In r219139, I've checked in a temporary workaround for this problem. Long story short, there's a bug in clang's integrated assembler, which sometimes causes malloc initialization to hang indefinitely in threaded statically linked executables (such as the bootstrap copy of ctfmerge). You might need to rebuild and install libc first, to get a good copy of ctfmerge, though. This is because the bootstrap copy of ctfmerge is linked with /usr/lib/libc.a, which might contain a bad version of malloc.o.