From owner-freebsd-alpha@FreeBSD.ORG Fri Mar 12 12:41:49 2004 Return-Path: Delivered-To: freebsd-alpha@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 234AD16A4CE for ; Fri, 12 Mar 2004 12:41:49 -0800 (PST) Received: from mail3.speakeasy.net (mail3.speakeasy.net [216.254.0.203]) by mx1.FreeBSD.org (Postfix) with ESMTP id F0E5343D39 for ; Fri, 12 Mar 2004 12:41:48 -0800 (PST) (envelope-from jhb@FreeBSD.org) Received: (qmail 10402 invoked from network); 12 Mar 2004 20:41:48 -0000 Received: from dsl027-160-063.atl1.dsl.speakeasy.net (HELO server.baldwin.cx) ([216.27.160.63]) (envelope-sender ) encrypted SMTP for ; 12 Mar 2004 20:41:48 -0000 Received: from 10.50.40.205 (gw1.twc.weather.com [216.133.140.1]) by server.baldwin.cx (8.12.10/8.12.10) with ESMTP id i2CKfe2A043821 for ; Fri, 12 Mar 2004 15:41:45 -0500 (EST) (envelope-from jhb@FreeBSD.org) From: John Baldwin To: alpha@FreeBSD.org Date: Fri, 12 Mar 2004 15:43:03 -0500 User-Agent: KMail/1.6 MIME-Version: 1.0 Content-Disposition: inline Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Message-Id: <200403121543.03123.jhb@FreeBSD.org> X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on server.baldwin.cx Subject: Testers Needed!! X-BeenThere: freebsd-alpha@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the Alpha List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 12 Mar 2004 20:41:49 -0000 Ok, two patches that I need someone to test. First off, I have some fixes to allow ptrace() to be marked MP safe again w/o generating a lock order reversal. This patch is at http://www.FreeBSD.org/~jhb/patches/alpha.ptrace.patch Note that you will have to do a 'make init_sysent.c' in sys/kern after patching to get the updated ptrace flag before building a new kernel. Second patch is to enable ithread preemption on Alpha which should reduce interrupt latency and provide a noticeable performance improvement (it did on i386 when we turned it on). The last time I was playing with having this on, I would get lockups under load on my SMP test system (a DS20). I think the reason it locked up was due to migration resulting in the thread that was interrupted by an interrupt returning back to PAL on a different CPU and that this violated some assumption made by PAL. To work around this, I use the new sched_pin() API to pin the interrupted thread around ithread preemption in this patch. So, this patch needs to be tested on both UP and SMP under load (-j 256 world build loops, etc.). I'm also curious if it does help performance at all, or at least the interactive "feel". Patch is at http://www.FreeBSD.org/~jhb/patches/alpha_preempt.patch Thanks. -- John Baldwin <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.org