Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262439AbTE0B5K (ORCPT ); Mon, 26 May 2003 21:57:10 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262445AbTE0B5K (ORCPT ); Mon, 26 May 2003 21:57:10 -0400 Received: from holomorphy.com ([66.224.33.161]:61895 "EHLO holomorphy") by vger.kernel.org with ESMTP id S262439AbTE0B5I (ORCPT ); Mon, 26 May 2003 21:57:08 -0400 Date: Mon, 26 May 2003 19:10:02 -0700 From: William Lee Irwin III To: Andrew Morton Cc: davem@redhat.com, andrea@suse.de, davidsen@tmr.com, haveblue@us.ibm.com, habanero@us.ibm.com, mbligh@aracnet.com, linux-kernel@vger.kernel.org Subject: Re: userspace irq balancer Message-ID: <20030527021002.GD8978@holomorphy.com> Mail-Followup-To: William Lee Irwin III , Andrew Morton , davem@redhat.com, andrea@suse.de, davidsen@tmr.com, haveblue@us.ibm.com, habanero@us.ibm.com, mbligh@aracnet.com, linux-kernel@vger.kernel.org References: <20030527000639.GA3767@dualathlon.random> <20030526.171527.35691510.davem@redhat.com> <20030527004115.GD3767@dualathlon.random> <20030526.174841.116378513.davem@redhat.com> <20030527015307.GC8978@holomorphy.com> <20030526185920.64e9751f.akpm@digeo.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20030526185920.64e9751f.akpm@digeo.com> Organization: The Domain of Holomorphy User-Agent: Mutt/1.5.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1166 Lines: 27 William Lee Irwin III wrote: >> In >> the userspace implementation the reprogramming is done infrequently >> enough to make even significant cost negligible; in-kernel the cost >> is entirely uncontrolled and the rate of reprogramming unlimited. On Mon, May 26, 2003 at 06:59:20PM -0700, Andrew Morton wrote: > eh? > #define MAX_BALANCED_IRQ_INTERVAL (5*HZ) > #define MIN_BALANCED_IRQ_INTERVAL (HZ/2) The number of interrupt sources on a system ends up scaling this up to numerous IO-APIC RTE reprograms and ioapic_lock acquisitions per-second (granted, with a 5s timeout between reprogramming storms) where it competes against IO-APIC interrupt acknowledgements. Making the lock per- IO-APIC would at least put a bound on the number of competitors mutually interfering with each other, but a tighter bound on the amount of work than NR_IRQS would be more useful than that. -- wli - 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/