Re: [Xen-devel] [PATCH V3 2/5] Config.mk: update OVMF changeset

[Fabio Fantoni <fabio.fantoni@xxxxxxx>]

Il 30/10/2013 12:50, Wei Liu ha scritto:
> On Wed, Oct 30, 2013 at 12:33:50PM +0100, Fabio Fantoni wrote:
> [...]
> > > > For this patches serie:
> > > > Tested-by: Fabio Fantoni <fabio.fantoni@xxxxxxx>
> > > >
> > > > About ovmf:
> > > > I don't have uefi domUs ready, so I did a fast test with windows 7
> > > > pro 64 bit dvd and starts, spice it works.
> > > > The problem found for now is very bad general performance and at the
> > > > end of windows installer loading the domU reboots.
> > > Oh Windows... :-P
> > I know that quality, efficiency, stability and security do not exist
> > in the Microsoft vocabulary...
> > Unfortunately almost all users for desktop/workstation use windows
> > (for now) and even at the time I still use it for work due to some
> > programs that only run on windows (and often are made very badly) :(
> Don't get me wrong. I'm just saying debugging Windows with no source
> code available is beyond my ability. :-)
>
> > > It would be useful to have at least some hypervisor / QEMU logs.
> > I found this error on xl dmesg:
> > > (XEN) mm.c:620:d0 Could not get page ref for pfn ffffffffffffffff
> > I see a recently posted patch about pfn:
> > http://lists.xen.org/archives/html/xen-devel/2013-10/msg02028.html
> > Can this serie solve the problem and I have to try it or has nothing
> > to do with this?
> I don't think that patch is useful to fix this at first glance.
>
> AFAICT this is a limitation of Xen's physmap implementation.  I'm still
> investigating a proper fix for this -- unfortunately that might
> involving modifying Xen's most complicated part so don't hold your
> breath.
>
> > And this is the error that cause reboot of domU:
> > > (XEN) hvm.c:1274:d4 Triple fault on VCPU0 - invoking HVM shutdown
> > > action 1.
> Windows, right? Could you try Linux?
>
> > I not found any other error or relevant information on logs but I
> > attach them if you want see.
> >
> > I also tried with Ubuntu 12.04.3, same pfn error on xl dmesg, same
> > performance problem (some minutes only for arrive to grub-efi of
> > dvd) and after other 10-15 minutes without complete the boot I did a
> > destroy.
> > Also for that logs on attachments.
> Is QMEU busy-looping?
I probably not understand exactly what you mean, the domU seems continue 
to boot apparently but in very slow motion.
Qemu is always active and with low load (on average 5% cpu).

> > Despite windows is bad, it seems that problems persist even in the
> > test linux except the shutdown with vcpu issue.
> >
> > The dom0 used have the installation of this:
> > http://lists.xen.org/archives/html/xen-devel/2013-10/msg01111.html
> > Except the git branch used (here the hvm-improve.t6) and the qemu:
> > QEMU_UPSTREAM_URL ?=
> > git://xenbits.xen.org/staging/qemu-upstream-unstable.git
> > QEMU_UPSTREAM_REVISION ?= master
> Oh please don't use staging QEMU. That introduces a new variable which
> makes it very hard to identify the genuine problem.
>
> Wei.
Is based on latest qemu stable (1.6.1).
I tested 1.6 fora long time without see regression and seems more good 
than 1.3 on something.
Are there specific reasons to use an older version instead?
If there were it would not be a problem for me to recompile/re-test with 
a qemu previous.
> > If you need more informations and/or tests tell me and I'll do/post them.
> >
> > About legacy bios support that I mentioned days ago I found this
> > document about:
> > http://code.coreboot.org/p/seabios/source/tree/master/README.CSM
> > Seem that need ovmf change (and probably also on hvmloader) that are
> > not able to do.
> >
> > Thank for any reply and sorry for my bad english.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel