From owner-freebsd-xen@FreeBSD.ORG Fri May 15 14:14:58 2015 Return-Path: Delivered-To: freebsd-xen@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:1900:2254:206a::19:1]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by hub.freebsd.org (Postfix) with ESMTPS id 5BA6FD82 for ; Fri, 15 May 2015 14:14:58 +0000 (UTC) Received: from SMTP02.CITRIX.COM (smtp02.citrix.com [66.165.176.63]) (using TLSv1 with cipher RC4-SHA (128/128 bits)) (Client CN "mail.citrix.com", Issuer "Cybertrust Public SureServer SV CA" (verified OK)) by mx1.freebsd.org (Postfix) with ESMTPS id E3F181182 for ; Fri, 15 May 2015 14:14:57 +0000 (UTC) X-IronPort-AV: E=Sophos;i="5.13,434,1427760000"; d="scan'208";a="265501264" Message-ID: <5555FF4A.8010702@citrix.com> Date: Fri, 15 May 2015 16:14:34 +0200 From: =?UTF-8?B?Um9nZXIgUGF1IE1vbm7DqQ==?= User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.9; rv:31.0) Gecko/20100101 Thunderbird/31.6.0 MIME-Version: 1.0 To: "Eggert, Lars" CC: "freebsd-xen@freebsd.org" Subject: Re: Xen dom0 "interrupt storm detected on "irq16:"; throttling interrupt source" References: <55531131.7040404@citrix.com> <65D7289A-9261-4F02-88E5-B2D137C268C1@netapp.com> <5553651D.3000400@citrix.com> <5555DFBC.9030306@citrix.com> <683335F9-E04E-4F8C-B1EF-0047E63E00AD@netapp.com> In-Reply-To: <683335F9-E04E-4F8C-B1EF-0047E63E00AD@netapp.com> Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: 8bit X-DLP: MIA2 X-BeenThere: freebsd-xen@freebsd.org X-Mailman-Version: 2.1.20 Precedence: list List-Id: Discussion of the freebsd port to xen - implementation and usage List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 15 May 2015 14:14:58 -0000 Hello, El 15/05/15 a les 14.22, Eggert, Lars ha escrit: > On 2015-5-15, at 13:59, Roger Pau Monné wrote: >> Can you provide the bootlog with the patch applied? > > Below. That's weird, AFAICT the patch I've send you should force isci to use the legacy interrupt, but it seems it still tries to use MSI-X. I've attached another patch that will force isci to print some messages, can you try that? It should be applied on a clean tree, since it also contains the change to xen_msi.c I've send you before, and of course you should not set hw.isci.force_legacy_interrupts when testing it. >> Also, can you try if there's any difference when setting: >> hw.isci.force_legacy_interrupts=1 >> on /boot/loader.conf? > > That seems to elimintate the storm. The isci line is now simply: > > isci0: port 0x6000-0x60ff mem 0xde07c000-0xde07ffff,0xddc00000-0xddffffff irq 16 at device 0.0 on pci10 Can you confirm the device works properly with hw.isci.force_legacy_interrupts=1? Thanks, Roger. --- diff --git a/sys/dev/isci/isci_interrupt.c b/sys/dev/isci/isci_interrupt.c index 52c64f7..f82b9de 100644 --- a/sys/dev/isci/isci_interrupt.c +++ b/sys/dev/isci/isci_interrupt.c @@ -137,6 +137,8 @@ isci_interrupt_setup(struct isci_softc *isci) isci->num_interrupts = max_msix_messages; pci_alloc_msix(isci->device, &isci->num_interrupts); + printf("ISCI: isci->num_interrupts: %u max_msix_messages: %u\n", + isci->num_interrupts, max_msix_messages); if (isci->num_interrupts == max_msix_messages) use_msix = TRUE; } diff --git a/sys/x86/xen/xen_msi.c b/sys/x86/xen/xen_msi.c index 0f678b1..21b542a 100644 --- a/sys/x86/xen/xen_msi.c +++ b/sys/x86/xen/xen_msi.c @@ -114,6 +114,7 @@ int xen_msix_alloc(device_t dev, int *irq) { + *irq = 0; return (ENXIO); }