Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S932840Ab2K3L1U (ORCPT ); Fri, 30 Nov 2012 06:27:20 -0500 Received: from mail.skyhub.de ([78.46.96.112]:36106 "EHLO mail.skyhub.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750897Ab2K3L1S (ORCPT ); Fri, 30 Nov 2012 06:27:18 -0500 Date: Fri, 30 Nov 2012 12:27:14 +0100 From: Borislav Petkov To: Daniel J Blueman Cc: Ingo Molnar , Thomas Gleixner , H Peter Anvin , Steffen Persvold , x86@kernel.org, linux-kernel@vger.kernel.org Subject: Re: [PATCH 3/4 v8] AMD64 EDAC: Fix PCI function lookup Message-ID: <20121130112714.GC6869@liondog.tnic> Mail-Followup-To: Borislav Petkov , Daniel J Blueman , Ingo Molnar , Thomas Gleixner , H Peter Anvin , Steffen Persvold , x86@kernel.org, linux-kernel@vger.kernel.org References: <1354265060-22956-1-git-send-email-daniel@numascale-asia.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline In-Reply-To: <1354265060-22956-1-git-send-email-daniel@numascale-asia.com> User-Agent: Mutt/1.5.21 (2010-09-15) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 4880 Lines: 147 On Fri, Nov 30, 2012 at 04:44:19PM +0800, Daniel J Blueman wrote: > Fix locating sibling memory controller PCI functions by using the > correct PCI domain and use Northbridge only if found. Tested on > multi-socket server and multi-server, multi-socket NumaConnect setup. > > v7: Refactor patches grouping changes > v8: Restructure searching for PCI function for clarity; use Northbridge > only if found > > Signed-off-by: Daniel J Blueman Ok, looks good. I ended up cleaning it up the code for F15h a bit and fixing it up. Here's the final version: -- commit 75cdc74742bbf368dfc52bb5041274c95d85a68f Author: Daniel J Blueman Date: Fri Nov 30 16:44:19 2012 +0800 amd64_edac: Fix PCI function lookup Fix locating sibling memory controller PCI functions by using the correct PCI domain and use a northbridge descriptor only if found. We need to at least warn if it wasn't found so that it gets fixed and we don't go off with wrong results. Signed-off-by: Daniel J Blueman [Boris: remove wrong comment, sanitize code and warn if NB desc lookup fails] Signed-off-by: Borislav Petkov diff --git a/drivers/edac/amd64_edac.c b/drivers/edac/amd64_edac.c index 60e93faaea18..e22a35224841 100644 --- a/drivers/edac/amd64_edac.c +++ b/drivers/edac/amd64_edac.c @@ -983,10 +983,29 @@ static u64 get_error_address(struct mce *m) return addr; } +static struct pci_dev *pci_get_related_function(unsigned int vendor, + unsigned int device, + struct pci_dev *related) +{ + struct pci_dev *dev = NULL; + + while ((dev = pci_get_device(vendor, device, dev))) { + if (pci_domain_nr(dev->bus) == pci_domain_nr(related->bus) && + (dev->bus->number == related->bus->number) && + (PCI_SLOT(dev->devfn) == PCI_SLOT(related->devfn))) + break; + } + + return dev; +} + static void read_dram_base_limit_regs(struct amd64_pvt *pvt, unsigned range) { + struct amd_northbridge *nb; + struct pci_dev *misc, *f1 = NULL; struct cpuinfo_x86 *c = &boot_cpu_data; int off = range << 3; + u32 llim; amd64_read_pci_cfg(pvt->F1, DRAM_BASE_LO + off, &pvt->ranges[range].base.lo); amd64_read_pci_cfg(pvt->F1, DRAM_LIMIT_LO + off, &pvt->ranges[range].lim.lo); @@ -1000,30 +1019,32 @@ static void read_dram_base_limit_regs(struct amd64_pvt *pvt, unsigned range) amd64_read_pci_cfg(pvt->F1, DRAM_BASE_HI + off, &pvt->ranges[range].base.hi); amd64_read_pci_cfg(pvt->F1, DRAM_LIMIT_HI + off, &pvt->ranges[range].lim.hi); - /* Factor in CC6 save area by reading dst node's limit reg */ - if (c->x86 == 0x15) { - struct pci_dev *f1 = NULL; - u8 nid = dram_dst_node(pvt, range); - u32 llim; + /* F15h: factor in CC6 save area by reading dst node's limit reg */ + if (c->x86 != 0x15) + return; - f1 = pci_get_domain_bus_and_slot(0, 0, PCI_DEVFN(0x18 + nid, 1)); - if (WARN_ON(!f1)) - return; + nb = node_to_amd_nb(dram_dst_node(pvt, range)); + if (WARN_ON(!nb)) + return; - amd64_read_pci_cfg(f1, DRAM_LOCAL_NODE_LIM, &llim); + misc = nb->misc; + f1 = pci_get_related_function(misc->vendor, PCI_DEVICE_ID_AMD_15H_NB_F1, misc); + if (WARN_ON(!f1)) + return; - pvt->ranges[range].lim.lo &= GENMASK(0, 15); + amd64_read_pci_cfg(f1, DRAM_LOCAL_NODE_LIM, &llim); - /* {[39:27],111b} */ - pvt->ranges[range].lim.lo |= ((llim & 0x1fff) << 3 | 0x7) << 16; + pvt->ranges[range].lim.lo &= GENMASK(0, 15); - pvt->ranges[range].lim.hi &= GENMASK(0, 7); + /* {[39:27],111b} */ + pvt->ranges[range].lim.lo |= ((llim & 0x1fff) << 3 | 0x7) << 16; - /* [47:40] */ - pvt->ranges[range].lim.hi |= llim >> 13; + pvt->ranges[range].lim.hi &= GENMASK(0, 7); - pci_dev_put(f1); - } + /* [47:40] */ + pvt->ranges[range].lim.hi |= llim >> 13; + + pci_dev_put(f1); } static void k8_map_sysaddr_to_csrow(struct mem_ctl_info *mci, u64 sys_addr, @@ -1713,23 +1734,6 @@ static struct amd64_family_type amd64_family_types[] = { }, }; -static struct pci_dev *pci_get_related_function(unsigned int vendor, - unsigned int device, - struct pci_dev *related) -{ - struct pci_dev *dev = NULL; - - dev = pci_get_device(vendor, device, dev); - while (dev) { - if ((dev->bus->number == related->bus->number) && - (PCI_SLOT(dev->devfn) == PCI_SLOT(related->devfn))) - break; - dev = pci_get_device(vendor, device, dev); - } - - return dev; -} - /* * These are tables of eigenvectors (one per line) which can be used for the * construction of the syndrome tables. The modified syndrome search algorithm -- Regards/Gruss, Boris. -- 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/