Re: The most recent attempt to build 2.6.29-rc8

[Jeremy Fitzhardinge <jeremy@goop.org>]

M A Young wrote:
> On Wed, 25 Mar 2009, M A Young wrote:
>
>> On Wed, 25 Mar 2009, Jeremy Fitzhardinge wrote:
>>
>>> M A Young wrote:
>>>> On Sun, 22 Mar 2009, Boris Derzhavets wrote:
>>>>
>>>>> Yes, i've noticed also that after build completed
>>>>>
>>>>> CONFIG_XEN_XENBUS_FRONTEND
>>>>> gets reset to "m"
>>>>
>>>> I think the attached patch fixes the XEN_XENBUS_FRONTEND 
>>>> configuration dependencies.
>>>
>>> (This patch was full of ^M's)
>>
>> That's strange because it wasn't when I sent it. I wonder if it got 
>> mangled in-transit.
>>
>>> Unfortunately this didn't help me; it still ended up making this 
>>> 'm', which doesn't work.  The simple fix is, I think, to make this a 
>>> "bool", so it gets built in.  Its not really big enough to bother with.
>>
>> I have been using XEN_XENBUS_FRONTEND=m on some of my kernels so it 
>> didn't cause me a problem, though of course if any of the dependent 
>> modules were "y" then that would indeed be incorrect. I did have an 
>> alternate way of doing it that I could try.
>
> I should probably have said that the alternate idea was to remove the 
> selects, then make XEN_XENBUS_FRONTEND default y and depends on 
> XEN_FBDEV_FRONTEND || XEN_NETDEV_FRONTEND || XEN_BLKDEV_FRONTEND || 
> XEN_KBDDEV_FRONTEND

Doesn't scale very well as we add more frontends though.

I don't see a huge benefit in making it separately compilable anyway; 
it's pretty small, and I can't think of much plausible use for a Xen 
kernel without frontends.  (On the other hand, making all of xenbus a 
single module might be useful, so that a non-Xen boot doesn't have to 
carry it.)

    J