[kernel-xen] Regression in kernel-xen-3.11.0-2

[Gordan Bobic]

On 09/11/2013 01:15 AM, Steven Haigh wrote:
> On 11/09/2013 7:48 AM, Gordan Bobic wrote:
>> On 09/07/2013 08:46 AM, Steven Haigh wrote:
>>> On 06/09/13 10:10, Steven Haigh wrote:
>>>> On 05/09/13 21:22, Gordan Bobic wrote:
>>>>> On Thu, 05 Sep 2013 21:14:49 +1000, Steven Haigh <netwiz at crc.id.au>
>>>>> wrote:
>>>>>> On 05/09/13 20:54, Gordan Bobic wrote:
>>>>>>> On Thu, 05 Sep 2013 15:37:31 +1000, Steven Haigh <netwiz at crc.id.au>
>>>>>>> wrote:
>>>>>>>> Hi all,
>>>>>>>>
>>>>>>>> I've built kernel 3.11.0 and placed it in the testing repo. Major
>>>>>>>> changes include optimisations on both the network stack and disk
>>>>>>>> subsystem.
>>
>> It looks like the pciehp driver is back in the kernel rather than built
>> as a module. Is this deliberate or is it a regression?
>
> This is interesting. It seems that a 'make oldconfig' for every new
> kernel I build defaults this to built in vs module. I'm not sure if
> there is reason behind this, or this is just default behaviour.

I have never trusted oldconfig to not break everything. The most sane 
approach I have found that works is to just copy the .config to the 
build directory, do make menuconfig, save, exit, and them rebuild.

> Sorry to be a pain, can you refresh my memory as to why we put this as a
> module? I didn't find anything in a quick scan through past threads in
> this list.

Check the list archive for 27th of April 2013. It's a workaround for a 
hardware/BIOS bug. Also see:
https://bugzilla.redhat.com/show_bug.cgi?id=529153
https://bugzilla.redhat.com/show_bug.cgi?id=908023

Gordan