From owner-freebsd-current@FreeBSD.ORG Wed Oct 27 17:52:39 2004 Return-Path: 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 10E1A16A4CF for ; Wed, 27 Oct 2004 17:52:39 +0000 (GMT) Received: from mail4.speakeasy.net (mail4.speakeasy.net [216.254.0.204]) by mx1.FreeBSD.org (Postfix) with ESMTP id ABB0643D5C for ; Wed, 27 Oct 2004 17:52:38 +0000 (GMT) (envelope-from jhb@FreeBSD.org) Received: (qmail 24208 invoked from network); 27 Oct 2004 17:52:38 -0000 Received: from dsl027-160-063.atl1.dsl.speakeasy.net (HELO server.baldwin.cx) ([216.27.160.63]) (envelope-sender ) encrypted SMTP for ; 27 Oct 2004 17:52:37 -0000 Received: from [10.50.40.221] (gw1.twc.weather.com [216.133.140.1]) (authenticated bits=0) by server.baldwin.cx (8.12.11/8.12.11) with ESMTP id i9RHqMSX021506; Wed, 27 Oct 2004 13:52:33 -0400 (EDT) (envelope-from jhb@FreeBSD.org) From: John Baldwin To: freebsd-current@FreeBSD.org Date: Wed, 27 Oct 2004 13:52:04 -0400 User-Agent: KMail/1.6.2 References: <20041022171134.73B1.SHINO@fornext.org> <20041022171924.73B3.SHINO@fornext.org> In-Reply-To: <20041022171924.73B3.SHINO@fornext.org> MIME-Version: 1.0 Content-Disposition: inline Message-Id: <200410271352.04069.jhb@FreeBSD.org> Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on server.baldwin.cx cc: Shunsuke SHINOMIYA cc: obrien@FreeBSD.org Subject: Re: disabling interrupt storm protection X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 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: Wed, 27 Oct 2004 17:52:39 -0000 On Friday 22 October 2004 04:26 am, Shunsuke SHINOMIYA wrote: > > Hi, > > has there been any ways to disable interrupt storm protection from > > sys/kern/kern_intr.c:1.109? > > Don't try sysctl hw.intr_storm_threshold=0. It mask all interrupt > sources. Bah, that's supposed to work but I guess bde's enhancements broke it. Can you try this patch: --- //depot/vendor/freebsd/src/sys/kern/kern_intr.c 2004/09/05 02:10:52 +++ //depot/projects/smpng/sys/kern/kern_intr.c 2004/10/27 17:41:06 @@ -485,14 +485,22 @@ struct intrhand *ih; /* and our interrupt handler chain */ struct thread *td; struct proc *p; +#ifdef HACK int count, warming, warned; +#else + int count, warned; +#endif td = curthread; p = td->td_proc; ithd = (struct ithd *)arg; /* point to myself */ KASSERT(ithd->it_td == td && td->td_ithd == ithd, ("%s: ithread and proc linkage out of sync", __func__)); +#ifdef HACK warming = 10 * intr_storm_threshold; +#else + count = 0; +#endif warned = 0; /* @@ -514,7 +522,9 @@ CTR4(KTR_INTR, "%s: pid %d: (%s) need=%d", __func__, p->p_pid, p->p_comm, ithd->it_need); +#ifdef HACK count = 0; +#endif while (ithd->it_need) { /* * Service interrupts. If another interrupt @@ -548,9 +558,29 @@ if ((ih->ih_flags & IH_MPSAFE) == 0) mtx_unlock(&Giant); } + +#ifndef HACK + /* + * If we detect an interrupt storm, pause with the + * source masked until the next hardclock tick. + */ + if (intr_storm_threshold != 0 && + count >= intr_storm_threshold) { + if (!warned) { + printf( + "Interrupt storm detected on \"%s\"; throttling interrupt source\n", + p->p_comm); + warned = 1; + } + tsleep(&count, td->td_priority, "istorm", 1); + count = 0; + } else + count++; +#endif + if (ithd->it_enable != NULL) { ithd->it_enable(ithd->it_vector); - +#ifdef HACK /* * Storm detection needs a delay here * to see slightly delayed interrupts @@ -565,8 +595,10 @@ DELAY(1); --warming; } +#endif } +#ifdef HACK /* * If we detect an interrupt storm, sleep until * the next hardclock tick. We sleep at the @@ -574,7 +606,8 @@ * to ensure that we see slightly delayed * interrupts. */ - if (count >= intr_storm_threshold) { + if (intr_storm_threshold != 0 && + count >= intr_storm_threshold) { if (!warned) { printf( "Interrupt storm detected on \"%s\"; throttling interrupt source\n", @@ -598,6 +631,7 @@ count = INT_MAX - 1; } count++; +#endif } WITNESS_WARN(WARN_PANIC, NULL, "suspending ithread"); mtx_assert(&Giant, MA_NOTOWNED); @@ -610,6 +644,9 @@ mtx_lock_spin(&sched_lock); if (!ithd->it_need) { TD_SET_IWAIT(td); +#ifndef HACK + count = 0; +#endif CTR2(KTR_INTR, "%s: pid %d: done", __func__, p->p_pid); mi_switch(SW_VOL, NULL); CTR2(KTR_INTR, "%s: pid %d: resumed", __func__, p->p_pid); First, can you test that your system works with the default threshold, and second, can you test that setting the threshold to zero disables the storm detection? -- John Baldwin <>< http://www.FreeBSD.org/~jhb/ "Power Users Use the Power to Serve" = http://www.FreeBSD.org