Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1760866AbZCTAcl (ORCPT ); Thu, 19 Mar 2009 20:32:41 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755799AbZCTAcb (ORCPT ); Thu, 19 Mar 2009 20:32:31 -0400 Received: from www.tglx.de ([62.245.132.106]:44411 "EHLO www.tglx.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754137AbZCTAca (ORCPT ); Thu, 19 Mar 2009 20:32:30 -0400 Date: Fri, 20 Mar 2009 01:31:41 +0100 (CET) From: Thomas Gleixner To: LKML cc: rt-users , Ingo Molnar , Steven Rostedt , Peter Zijlstra , Carsten Emde , Clark Williams , Frank Rowand Subject: [Announce] 2.6.29-rc78rt1 In-Reply-To: Message-ID: References: User-Agent: Alpine 2.00 (LFD 1167 2008-08-23) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2391 Lines: 73 We are pleased to announce the next update to our new preempt-rt series. - port forward to 2.6.29-rc8 - disable -rt conflicting config options - hotplug cpu fixes (peterz) - slab/pagealloc lock breaks (peterz / tglx) - sigqueue caching for -rt tasks - posixtimer thread avoid useless wakeups - various build fixes (mingo, frank ....) - lots of tracer updates from -tip (check the tip git logs) The outstanding improvement is the slab/pagealloc change which breaks and splits locking and brought down worst case latencies in problematic use cases from >500us to <100us. As a side note: There seems to be a wide spread underestimation of the problem spots exposed by preempt-rt. The usual shrug off answer is: "I don't care about -rt. Come back if you can expose the same problem in the mainline kernel." This is a fundamentally wrong answer. preempt-rt mostly exposes existing latency spots and magnifies them Reducing latencies in -rt by a factor 5 will be not that prominent in a non-rt setup, but the problematic code area will still produce measureable latency problems. I'm well aware of the tradeoff between determinitic behaviour and throughput, but problematic spots (e.g. lock contentions) hurt both. So can we please put down the stupid "I don't care about -rt" attitudes and accept that we have to think about the mutual benefits of deterministic and throughput aspects without hurting each other ? Download locations: http://rt.et.redhat.com/download/ http://www.kernel.org/pub/linux/kernel/projects/rt/ Information on the RT patch can be found at: http://rt.wiki.kernel.org/index.php/Main_Page to build the 2.6.29-rc8-rt1 tree, the following patches should be applied: http://www.kernel.org/pub/linux/kernel/v2.6/testing/linux-2.6.29-rc8.tar.bz2 http://www.kernel.org/pub/linux/kernel/projects/rt/patch-2.6.29-rc8-rt1.bz2 The broken out patches are also available at the same download locations. Enjoy ! tglx P.S.: ARM/PowerPC support is in the pipeline and will be available with -rt2 (hopefully :) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/