From owner-freebsd-current@FreeBSD.ORG Fri Aug 4 20:54:05 2006 Return-Path: X-Original-To: current@FreeBSD.org Delivered-To: freebsd-current@FreeBSD.ORG Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 741A816A4DF; Fri, 4 Aug 2006 20:54:05 +0000 (UTC) (envelope-from grehan@freebsd.org) Received: from dommail.onthenet.com.au (dommail.OntheNet.com.au [203.13.70.57]) by mx1.FreeBSD.org (Postfix) with ESMTP id 1FCA943D45; Fri, 4 Aug 2006 20:54:03 +0000 (GMT) (envelope-from grehan@freebsd.org) Received: from [10.33.24.110] (nat-198-95-226-228.netapp.com [198.95.226.228]) by dommail.onthenet.com.au (MOS 3.5.7-GR) with ESMTP id CAA48629 (AUTH peterg@ptree32.com.au); Sat, 5 Aug 2006 06:54:00 +1000 (EST) Message-ID: <44D3B416.1060508@freebsd.org> Date: Fri, 04 Aug 2006 13:54:46 -0700 From: Peter Grehan User-Agent: Mozilla/5.0 (X11; U; FreeBSD amd64; en-US; rv:1.8b) Gecko/20051014 MIME-Version: 1.0 To: Jason Evans References: <44D12AC0.90009@sippysoft.com> <20060802230803.GA32778@troutmask.apl.washington.edu> <44D13BC7.8010309@FreeBSD.org> <44D168A8.1060201@freebsd.org> <44D173B7.6040302@FreeBSD.org> <44D1757D.4010103@ptree32.com.au> <44D3A642.1000905@FreeBSD.org> In-Reply-To: <44D3A642.1000905@FreeBSD.org> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Cc: Maxim Sobolev , "current@freebsd.org" , Peter Grehan , Steve Kargl , freebsd-ppc@FreeBSD.org Subject: Re: Looks like threading is b0rken on FreeBSD/powerpc 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: Fri, 04 Aug 2006 20:54:05 -0000 >> jemalloc. From the backtrace, t's doing a TLS allocation, where I'm >> certain phkmalloc didn't do anything like that. > > Indeed, jemalloc does add a dependency on TLS for PIC versions, though > it can be configured to run (more slowly) without TLS, as Marcel pointed > out. This appears to be a bug in the TLS implementation, rather than in > jemalloc; the appropriate fix is to make TLS work. Exactly. I was pointing out the difference, sorry if it appeared to be putting blame on jemalloc. later, Peter.