Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1765800AbXEVIMU (ORCPT ); Tue, 22 May 2007 04:12:20 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1759018AbXEVIGg (ORCPT ); Tue, 22 May 2007 04:06:36 -0400 Received: from mtagate1.de.ibm.com ([195.212.29.150]:63389 "EHLO mtagate1.de.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756823AbXEVIGR (ORCPT ); Tue, 22 May 2007 04:06:17 -0400 From: muli@il.ibm.com To: ak@suse.de Cc: linux-kernel@vger.kernel.org, discuss@x86-64.org, jdmason@kudzu.us, michaelv@il.ibm.com, guillaume.thouvenin@ext.bull.net, muli@il.ibm.com Subject: [PATCH 11/15] x86-64: Calgary - reserve TCEs with the same address as MEM regions Date: Tue, 22 May 2007 04:06:00 -0400 Message-Id: <11798211673144-git-send-email-muli@il.ibm.com> X-Mailer: git-send-email 1.4.4 In-Reply-To: <1179821167443-git-send-email-muli@il.ibm.com> References: <11798211643534-git-send-email-muli@il.ibm.com> <11798211643659-git-send-email-muli@il.ibm.com> <11798211643144-git-send-email-muli@il.ibm.com> <11798211663773-git-send-email-muli@il.ibm.com> <1179821166685-git-send-email-muli@il.ibm.com> <11798211661041-git-send-email-muli@il.ibm.com> <11798211672108-git-send-email-muli@il.ibm.com> <1179821167481-git-send-email-muli@il.ibm.com> <11798211673003-git-send-email-muli@il.ibm.com> <11798211672822-git-send-email-muli@il.ibm.com> <1179821167443-git-send-email-muli@il.ibm.com> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 2472 Lines: 94 From: Muli Ben-Yehuda This works around a bug where DMAs that have the same addresses as some MEM regions do not go through. Not clear yet if this is due to a mis-configuration or something deeper. Signed-off-by: Muli Ben-Yehuda --- arch/x86_64/kernel/pci-calgary.c | 67 ++++++++++++++++++++++++++++++++++++++ 1 files changed, 67 insertions(+), 0 deletions(-) diff --git a/arch/x86_64/kernel/pci-calgary.c b/arch/x86_64/kernel/pci-calgary.c index a49b19d..9efecf3 100644 --- a/arch/x86_64/kernel/pci-calgary.c +++ b/arch/x86_64/kernel/pci-calgary.c @@ -1538,3 +1538,70 @@ static int __init calgary_parse_options( return 1; } __setup("calgary=", calgary_parse_options); + +static void __init calgary_fixup_one_tce_space(struct pci_dev *dev) +{ + struct iommu_table *tbl; + unsigned int npages; + int i; + + tbl = dev->sysdata; + + for (i = 0; i < 4; i++) { + struct resource *r = &dev->resource[PCI_BRIDGE_RESOURCES + i]; + + /* Don't give out TCEs that map MEM resources */ + if (!(r->flags & IORESOURCE_MEM)) + continue; + + /* 0-based? we reserve the whole 1st MB anyway */ + if (!r->start) + continue; + + /* cover the whole region */ + npages = (r->end - r->start) >> PAGE_SHIFT; + npages++; + + printk(KERN_DEBUG "Calg: dev %p [%x] tbl %p reserving " + "0x%Lx-0x%Lx [0x%x pages]\n", dev, dev->bus->number, + tbl, r->start, r->end, npages); + + iommu_range_reserve(tbl, r->start, npages); + } +} + +static int __init calgary_fixup_tce_spaces(void) +{ + struct pci_dev *dev = NULL; + void* tce_space; + + if (no_iommu || swiotlb || !calgary_detected) + return -ENODEV; + + printk(KERN_DEBUG "Calgary: fixing tce spaces\n"); + + do { + dev = pci_get_device(PCI_VENDOR_ID_IBM, PCI_ANY_ID, dev); + if (!dev) + break; + if (!is_cal_pci_dev(dev->device)) + continue; + if (!translate_phb(dev)) + continue; + + tce_space = bus_info[dev->bus->number].tce_space; + if (!tce_space) + continue; + + calgary_fixup_one_tce_space(dev); + + } while (1); + + return 0; +} + +/* + * We need to be call after pcibios_assign_resources (fs_initcall level) + * and before device_initcall. + */ +rootfs_initcall(calgary_fixup_tce_spaces); -- 1.4.4 - 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/