Re: 2.6.12-mm1

[Andy Whitcroft]

Greg KH wrote:
> On Mon, Jun 20, 2005 at 02:14:51PM +0100, Andy Whitcroft wrote:

>>Having trouble getting 2.6.12-mm1 to compile on my x86 test
>>boxes other than a basic PC.  I suspect this patch is to 'blame'.
>>
>>>+gregkh-pci-pci-assign-unassigned-resources.patch
>>
>>We seem to need to include setup-bus.o for most x86 architectures
>>regardless of HOTPLUG.  Not sure if this is the right fix, but it
>>seems to work on the systems I have tested.

> Sounds like a NUMA issue, right?  If you don't have HOTPLUG enabled, X86
> should not need setup_bus.  Care to find the real problem here?

Ok.  I've spent some time looking at this and I think my fix is correct
if we assume that the intent of the change in the patch below is correct:

	gregkh-pci-pci-assign-unassigned-resources.patch

This patch adds a call to pci_assign_unassigned_resources() to
pcibios_init().  pcibios_init() is called unconditionally as a
subsys_initcall() from arch/i386/pci/common.c which is an unconditional
link for i386.

@@ -165,6 +165,7 @@
        if ((pci_probe & PCI_BIOS_SORT) && !(pci_probe & PCI_NO_SORT))
                pcibios_sort();
 #endif
+       pci_assign_unassigned_resources();
        return 0;
 }

I am not a PCI guru so I can't comment on whether this call is
reasonable, but if it is then we require setup-bus.o for all i386 platforms.

I will note that in reading the patch the commentry at the top lists
three individual changes which I think I can identify in the patch
itself, I don't feel that the change above falls under any of them?

Ivan, can you shead any light on whether the hunk of your patch above is
one of the three fixes, whether its a fourth fix, and indeed whether its
needed?

-apw

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/