Re: [Patch] Janitorial cleanup of GET_INDEX macro in arch/i386/pci/fixup.c

[Jeff Garzik <jgarzik@pobox.com>]

Greg KH wrote:
> On Mon, Jun 27, 2005 at 03:19:11PM -0700, Linus Torvalds wrote:
> 
>>
>>On Mon, 27 Jun 2005, Neil Horman wrote:
>>
>>>Patch to clean up the implementation of the GET_INDEX macro in the i386 pci
>>>fixup code so that it uses the PCI_DEVFN macro, rather than re-implements it.
>>
>>This looks wrong:
>>
>>
>>>-#define GET_INDEX(a, b) ((((a) - PCI_DEVICE_ID_INTEL_MCH_PA) << 3) + ((b) & 7))
>>>+#define GET_INDEX(a, b) PCI_DEVFN((a - PCI_DEVICE_ID_INTEL_MCH_PA),b)
>>
>>that first argument looks like it has parentheses at the wrong place, it 
>>should be
>>
>>	(a) - PCI_DEVICE_ID_INTEL_MCH_PA
>>
>>rather than
>>
>>	(a - PCI_DEVICE_ID_INTEL_MCH_PA)
>>
>>methinks.
>>
>>Other than that... Greg?
> 
> 
> I'd like to say yes, but I'll get an ack by the pci express people from
> Intel first (PCI_DEVFN masks off bits that might be needed here, don't
> really know...)  Also, this is only used for an array index, not a
> pci devfn memory access (look at how it is used in the code...)
> 
> I'll put it in my tree for now, and let it get testing, I would not
> recommend it for yours just yet.

Please let me know, as I suggested this patch to Neil.

It sure seems like the code wants a real PCI devfn, even though it is 
obviously doing a table index.

Comments?

	Jeff