x86/amd/numa: Fix northbridge quirk to assign correct NUMA node
authorDaniel J Blueman <daniel@numascale.com>
Thu, 13 Mar 2014 11:43:01 +0000 (19:43 +0800)
committerIngo Molnar <mingo@kernel.org>
Fri, 14 Mar 2014 10:05:36 +0000 (11:05 +0100)
For systems with multiple servers and routed fabric, all
northbridges get assigned to the first server. Fix this by also
using the node reported from the PCI bus. For single-fabric
systems, the northbriges are on PCI bus 0 by definition, which
are on NUMA node 0 by definition, so this is invarient on most
systems.

Tested on fam10h and fam15h single and multi-fabric systems and
candidate for stable.

Signed-off-by: Daniel J Blueman <daniel@numascale.com>
Acked-by: Steffen Persvold <sp@numascale.com>
Acked-by: Borislav Petkov <bp@suse.de>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/1394710981-3596-1-git-send-email-daniel@numascale.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
arch/x86/kernel/quirks.c

index 7c6acd4..ff898bb 100644 (file)
@@ -529,7 +529,7 @@ static void quirk_amd_nb_node(struct pci_dev *dev)
                return;
 
        pci_read_config_dword(nb_ht, 0x60, &val);
-       node = val & 7;
+       node = pcibus_to_node(dev->bus) | (val & 7);
        /*
         * Some hardware may return an invalid node ID,
         * so check it first: