In article <20051121231824.GA2930@antioche.eu.org>,
	Manuel Bouyer <bouyer@antioche.eu.org> writes:
> On Sun, Nov 20, 2005 at 10:25:39PM +0000, Johnny C. Lam wrote:
>> 
>> I think XEN0 should look more like GENERIC in terms of the the network
>> and disk controller support that is built-in.  XENU makes sense to be
>> stripped down as it can only use the devices that are exported by the
>> privileged domain, but XEN0 needs to provide the network and disks to
>> the guest domains, so it should support as many as it can by default.
>> It was a minor stumbling block for me to install Xen for the first
>> time when I saw that the amr(4) RAID controller in my rather common
>> Dell PowerEdge server wasn't supported by XEN0 and that I would need
>> to compile my own kernel.
>> 
>> I appreciate that NetBSD/xen is still a work-in-progress, but this
>> change should make it easier for interested parties to test NetBSD/xen
>> on their modern i386 hardware without needing to compile their own
>> kernels.  Since Xen needs modern hardware to run anyway, I don't think
>> the difference between a smaller and a larger default kernel are going
>> to be significant details to most Xen users.
> 
> I added a buch of PCI devices to XEN0. It would be nice to have common file 
> with device to be inclued by various kernel configs ... :)

Absolutely. It would be nice, too, to have them factored somewhat, so
a user could comment out, say, all the EISA devices without having to
suffer hand-patching every time a new one is added to GENERIC.


Frederick