following is my thinking, it may be wrong.
want to run dom0 on VTi, we still need to support of paravitualization, madison
is still being used.
have two branch, one VT-i dom0, the other is para dom0.
focus on VT-i dom0 branch.
we still support para dom0, if there are bugs, we need to fix it,
but we will not update xenolinux to new release version, we just try to make all
functions work well.
other side, It's impossible Redhat would like to maintain two modified linux
version, one is for hybrid, the other is for para.
may choose modified linux version for hybrid, it is more easier for
Alex and all,
we, in Bull, are using Xen-ia64 to provide logical
partitioning for Linux physical partitions in our mainframe systems running on
Some of these systems have been
delivered with Madison CPUs which are non-VT.
So I would vote for keeping the support of paravirtualization.
On the technical side, I agree with the
long-term objective, but performance is a major point and I think we should not
discard paravirtualization until we have verified that there is no performance
regression with most of the CPUs installed in Itanium systems.
Knowing that performance will be better with Tukwila is a
valuable information for a long term objective but Tukwila is not what is
So my opinion is that
removing paravirtualization may be a good choice for the long term, but deciding
to do it now is premature.
Envoyé par : xen-ia64-devel-bounces@xxxxxxxxxxxxxxxxxxx
Objet : [Xen-ia64-devel] Time for hybrid
Is it possible? Is it a good idea? What are
some of the issues? We
would lose support for non-VT capable processors
(pre-Montecito), but is
that so bad? Is it a "fast track" to upstream
Linux Xen/ia64 support?
Let me know your thoughts.
HP Open Source & Linux
Xen-ia64-devel mailing list