From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S261868AbTJGHuo (ORCPT ); Tue, 7 Oct 2003 03:50:44 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S261875AbTJGHuo (ORCPT ); Tue, 7 Oct 2003 03:50:44 -0400 Received: from smtpout.mac.com ([17.250.248.86]:9429 "EHLO smtpout.mac.com") by vger.kernel.org with ESMTP id S261868AbTJGHuj (ORCPT ); Tue, 7 Oct 2003 03:50:39 -0400 Message-ID: <1448383.1065513016965.JavaMail.pwaechtler@mac.com> Date: Tue, 07 Oct 2003 09:50:16 +0200 From: Peter Waechtler To: Jakub Jelinek Subject: Re: POSIX message queues Cc: Jamie Lokier , Ulrich Drepper , Krzysztof Benedyczak , linux-kernel@vger.kernel.org, Manfred Spraul , Michal Wronski Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sunday, October 05, 2003, at 08:32PM, Jakub Jelinek wrote: >> Speaking of librt - I should not have to link in pthreads and the >> run-time overhead associated with it (locking stdio etc.) just so I >> can use shm_open(). Any chance of fixing this? > >That overhead is mostly gone in current glibcs (when using NPTL): >a) e.g. locking is done unconditionally even when libpthread is not present > (it is just lock cmpxchgl, inlined) a "lock cmpxchg" is > 100 cycles (according to a recent Linux Journal article from Paul McKenney: 107ns on 700MHz PentiumIII) But I think you will have benchmarked the alternatives? BTW, what are they? you suggested naming the syscall number symbols NR_mq_open instead of NR_sys_mq_open. In the stub I want to overload some syscalls (e.g. mq_open) but others not (e.g. mq_timedsend). How to deal with that?