On Thu, Dec 22, 2005 at 11:45:06PM -0000, Ian Pratt wrote:
> > becomes unstable. I am now testing the same HD with kernel
> > 2.6.12-1.1454_FC4.i686 in another machine (AMD Duron(tm)
> Processor),
> > and following the tip of the friend I incapacitated
> "acpi=off". Until the moment all good is going...
>
> yes there have been feedbacks on the xen user list about
> acpi=off, noapic and 'swiotlb=force swiotlb=8m' kernel
> arguments fixing some bugs raised in recent Xen-3.0 versions.
> It's not fully cooked :-)
swiotlb=force/8m should *not* be required with xen 3.0.0, only with
pre-release versions of the -unstable tree (such as the RH FC4 RPMs).
The vast majority of machines seem to work fine with acpi/apic enabled
-- I've only seen a handful of people say they've needed to turn them
off, and several of these have found that the problem goes away if they
updated their BIOS. I would like to get to the bottom of these reports,
though.
Although Xen doesn't run on all the huge range of PC hardware that Linux
does, it does seem to run fine on a very large subset, in particular the
vast majority of modern server class hardware.
It certainly do run fine on my laptop :-)
The "fully cooked" applied to what we just pushed on fedora rawhide
channel, this is the merge branch and with the Fedora set of config
options also using rawhide kernel. I replied in this context (and didn't
see the Cc' to xen-user which had been added), I didn't want to offence !
Daniel
--
Daniel Veillard | Red Hat
http://redhat.com/
veillard(a)redhat.com | libxml GNOME XML XSLT toolkit
http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine
http://rpmfind.net/