Re: [PATCH 0/3, v8] PCI, ACPI: Physical PCI slot objects

[Gary Hade <garyhade@us.ibm.com>]

On Tue, Mar 04, 2008 at 05:42:52PM -0700, Alex Chiang wrote:
> Hi Greg, all,
> 
> This is v8 of my "introduce pci_slot" series, to be considered for
> inclusion in linux-next to get more test exposure and shake out
> the bugs.
> 
> The major change in this revision is removing an earlier patch
> which removed the 'path' sysfs attribute on SGI machines. After
> review from Jesse Barnes and Prarit Bhargava, they kindly
> informed me that 'path' is actually an SGI physical path, and not
> a PCI address. My patch would have definitely broken userspace on
> SGI machines, so I eliminated it, and 'path' remains on SGI boxes.
> 
> This brings the patch series down to a total of 3 patches:
> 
> 	0001-Construct-one-fakephp-slot-per-pci-slot.patch
> 	0002-Introduce-pci_slot.patch
> 	0003-ACPI-PCI-slot-detection-driver.patch
> 
> Thanks for all the code reviews from various people.
> 
> I'm now just looking for some people to beat on this code and let
> me know where I screwed up. :)

Hi Alex,
The x3950 M2 on which the earlier reported Oops (invalid opcode
while in acpi_ev_notify_dispatch) reproduced is unfortunately still
tied up because of a customer critical project.

I have been able to try these new patches on a single node
x3850 [2 hotpluggable PCI-X slots, 4 hotpluggable PCIe slots]
and a 2 node x3950 [12 total (6/node) hotpluggable PCI-X slots]
and found no problems.

Gary

-- 
Gary Hade
System x Enablement
IBM Linux Technology Center
503-578-4503  IBM T/L: 775-4503
garyhade@us.ibm.com
http://www.ibm.com/linux/ltc