summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorPan, Jacob jun <jacob.jun.pan@intel.com>2010-02-25 01:59:55 +0100
committerH. Peter Anvin <hpa@zytor.com>2010-02-25 18:04:19 +0100
commita92d152ef9dd89c578ca2ec7118e9de8fb74a75f (patch)
tree90ee3c6e1b69e1858f38d994092d1e9d2d86d958
parentx86, pci: Add sanity check for PCI fixed bar probing (diff)
downloadlinux-a92d152ef9dd89c578ca2ec7118e9de8fb74a75f.tar.xz
linux-a92d152ef9dd89c578ca2ec7118e9de8fb74a75f.zip
x86, numaq: Make CONFIG_X86_NUMAQ depend on CONFIG_PCI
The NUMAQ initialization sets x86_init.pci.init to pci_numaq_init, which obviously isn't defined if CONFIG_PCI isn't defined. This dependency was implicit in the past, because pci_numaq_init was invoked from arch/x86/pci/legacy.c, which itself was conditioned on CONFIG_PCI. I suspect that no NUMA-Q machines without PCI were ever built, so instead of complicating the code by adding #ifdefs or stub functions, just disable this bit of the configuration space. [ hpa: rewrote the checkin comment ] Signed-off-by: Jacob Pan <jacob.jun.pan@intel.com> LKML-Reference: <43F901BD926A4E43B106BF17856F0755A321EE1F@orsmsx508.amr.corp.intel.com> Signed-off-by: H. Peter Anvin <hpa@zytor.com>
-rw-r--r--arch/x86/Kconfig1
1 files changed, 1 insertions, 0 deletions
diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
index 0ab2dcef7d84..f0322949328e 100644
--- a/arch/x86/Kconfig
+++ b/arch/x86/Kconfig
@@ -425,6 +425,7 @@ config X86_32_NON_STANDARD
config X86_NUMAQ
bool "NUMAQ (IBM/Sequent)"
depends on X86_32_NON_STANDARD
+ depends on PCI
select NUMA
select X86_MPPARSE
---help---