Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757924AbZFQJOP (ORCPT ); Wed, 17 Jun 2009 05:14:15 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1764988AbZFQJNu (ORCPT ); Wed, 17 Jun 2009 05:13:50 -0400 Received: from fgwmail5.fujitsu.co.jp ([192.51.44.35]:48436 "EHLO fgwmail5.fujitsu.co.jp" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757924AbZFQJNr (ORCPT ); Wed, 17 Jun 2009 05:13:47 -0400 Message-ID: <4A38B3B4.1020304@jp.fujitsu.com> Date: Wed, 17 Jun 2009 18:13:24 +0900 From: Kenji Kaneshige User-Agent: Thunderbird 2.0.0.21 (Windows/20090302) MIME-Version: 1.0 To: Andrew Patterson CC: linux-pci@vger.kernel.org, torvalds@linux-foundation.org, linux-kernel@vger.kernel.org, jbarnes@virtuousgeek.org Subject: Re: [PATCH 0/1] Recurse when searching for empty slots in resources trees References: <20090616220419.14021.84524.stgit@bob.kio> In-Reply-To: <20090616220419.14021.84524.stgit@bob.kio> Content-Type: text/plain; charset=ISO-2022-JP Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4948 Lines: 147 Andrew Patterson wrote: > I recently ran into a resource collision problem where PCI hot-plug > operations are failing for certain PCI topologies. One case > illustrating the problem is using a QLogic PCIe HBA in a slot with a > PCIe root port as its parent bus. Here is an abbreviated lspci output > for this topology: > > -+-[0000:c2]---00.0-[0000:c3-fb]--+-00.0 QLogic Corp. 8Gb Fibre Channel HBA > | \-00.1 QLogic Corp. 8Gb Fibre Channel HBA > > > > c2:00.0 PCI bridge: PCIe Root Port (prog-if 00 [Normal decode]) > Bus: primary=c2, secondary=c3, subordinate=fb, sec-latency=0 > I/O behind bridge: 00001000-0000ffff > Memory behind bridge: f0000000-fdffffff > Prefetchable memory behind bridge: 0000080780000000-00000807ffffffff > > c3:00.0 Fibre Channel: QLogic Corp. 8Gb Fibre Channel HBA > Region 0: I/O ports at 8001100 [size=256] > Region 1: Memory at f0284000 (64-bit, non-prefetchable) [size=16K] > Region 3: Memory at f0100000 (64-bit, non-prefetchable) [size=1M] > Expansion ROM at f0240000 [disabled] [size=256K] > > c3:00.1 Fibre Channel: QLogic Corp. 8Gb Fibre Channel HBA > Region 0: I/O ports at 8001000 [size=256] > Region 1: Memory at f0280000 (64-bit, non-prefetchable) [size=16K] > Region 3: Memory at f0000000 (64-bit, non-prefetchable) [size=1M] > Expansion ROM at f0200000 [disabled] [size=256K] > > After boot, the resource tree looks like: > > f0000000-fdffffff : PCI Bus 0000:c3 > f0000000-fdffffff : PCI Bus 0000:c2 > f0000000-f00fffff : 0000:c3:00.1 > f0000000-f00fffff : qla2xxx > f0100000-f01fffff : 0000:c3:00.0 > f0100000-f01fffff : qla2xxx > f0200000-f023ffff : 0000:c3:00.1 > f0240000-f027ffff : 0000:c3:00.0 > f0280000-f0283fff : 0000:c3:00.1 > f0280000-f0283fff : qla2xxx > f0284000-f0287fff : 0000:c3:00.0 > f0284000-f0287fff : qla2xxx > > Note that PCI Bus 0000:c2 is a child of PCI Bus 0000:c3 and has an > identical address range. According to the lspci output, PCI Bus 0000:c2 is a parent of PCI Bus 0000:c3. But they are upside down in resource tree. I guess this is the root cause of the problem. I think the reason why it happen is ia64 (I believe you are using ia64 environment) uses pci_claim_resource(), which calls insert_resource(), to insert resources. In my understanding, if two resources having an identical address range are inserted using insert_resource(), the latter one becomes parent of the first one. I made a sample patch, though I don't know if it fixes the problem. I hope this would help you. But please note that it is NOT well tested. Thanks, Kenji Kaneshige arch/ia64/pci/pci.c | 32 +++++++++++++++++++++++--------- 1 file changed, 23 insertions(+), 9 deletions(-) Index: 20090612/arch/ia64/pci/pci.c =================================================================== --- 20090612.orig/arch/ia64/pci/pci.c +++ 20090612/arch/ia64/pci/pci.c @@ -301,9 +301,10 @@ static __devinit acpi_status add_window( } static void __devinit -pcibios_setup_root_windows(struct pci_bus *bus, struct pci_controller *ctrl) +pcibios_setup_root_windows(struct pci_bus *bus) { int i, j; + struct pci_controller *ctrl = bus->sysdata; j = 0; for (i = 0; i < ctrl->windows; i++) { @@ -371,9 +372,6 @@ pci_acpi_scan_root(struct acpi_device *d * such quirk. So we just ignore the case now. */ pbus = pci_scan_bus_parented(NULL, bus, &pci_root_ops, controller); - if (pbus) - pcibios_setup_root_windows(pbus, controller); - return pbus; out3: @@ -456,15 +454,29 @@ pcibios_fixup_resources(struct pci_dev * { struct pci_bus_region region; int i; + struct resource *devr, *busr; for (i = start; i < limit; i++) { - if (!dev->resource[i].flags) + char *dtype = i < PCI_BRIDGE_RESOURCES ? "device" : "bridge"; + + devr = &dev->resource[i]; + if (!devr->flags) continue; + region.start = dev->resource[i].start; region.end = dev->resource[i].end; - pcibios_bus_to_resource(dev, &dev->resource[i], ®ion); - if ((is_valid_resource(dev, i))) - pci_claim_resource(dev, i); + pcibios_bus_to_resource(dev, devr, ®ion); + if (!is_valid_resource(dev, i)) + continue; + + busr = pci_find_parent_resource(dev, devr); + if (!busr || request_resource(busr, devr)) { + dev_err(&dev->dev, "BAR %d: %s of %s %pR\n", i, + busr ? "address space collision on" : + "no parent found for", + dtype, devr); + devr->flags = 0; + } } } @@ -487,7 +499,9 @@ pcibios_fixup_bus (struct pci_bus *b) { struct pci_dev *dev; - if (b->self) { + if (pci_is_root_bus(b)) + pcibios_setup_root_windows(b); + else { pci_read_bridge_bases(b); pcibios_fixup_bridge_resources(b->self); } -- 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/