Re: Xen2 removal: now or later

[Christoph Egger <Christoph_Egger%gmx.de@localhost>]

Stephen Borrill wrote:
> On Fri, 31 Oct 2008, Christoph Egger wrote:
>>>> Xen 2 support was only kept in for one reason:
>>>> NetBSD supports PCI passthrough with Xen 2 but not with Xen 3.
>>>>
>>>> The question is: when is the right time to throw away Xen 2 ?
>>>>
>>>> Should it be done now or should we wait for
>>>> a) PCI passthrough with Xen 3 is there
>>>> b) or netbsd-5 got some testing just in case to make pullups
>>>>    for Xen 2 easier ?
> 
> I think given that "NetBSD supports PCI passthrough with Xen 2 but not
> with Xen 3", then option a) must be taken.

Why do you think, -current can't live a short time period w/o any PCI
passthrough support? This fact just can give someone a motivation
to do the work sooner than later :)

>>> It would be good if HEAD and netbsd-5 didn't diverge too quickly,
>>> though.
>>
>> So you're argumenting for wait a little - point b) from above, right?
> 
> I don't see why it's a choice between a) _OR_ b) as long as PCI
> passthrough is pulled up to netbsd-5

This pullup can happen anytime to netbsd-5 independent if Xen 2 support
has been removed in -current or not.

Further, I do NOT have a Xen 2 test machine. The stuff I'm working on is
Xen 3 only. I can and I do only compile testing for Xen 2. For patches
which may have broken Xen 2 *before* branching netbsd-5, I asked bouyer
for the ok to commit since he can actually runtest on Xen 2.

If I break Xen 2 now, my fix for Xen 2 is to remove it. I have a
ready-to-commit patch to do it now:
http://www.netbsd.org/~cegger/rm_xen2.diff


Christoph