On Tue, Feb 15, 2011 at 10:28:08PM +0200, Gleb Natapov wrote:
> On Tue, Feb 15, 2011 at 02:21:41PM -0600, Anthony Liguori wrote:
> > On 02/15/2011 02:07 PM, Bruce Rogers wrote:
> > >Hi,
> > >
> > >We just noticed an issue flagged by a libvirt based test. This same
> > >command line didn't used to fail, and I wanted to be sure that this is
> > >behaving as intended.
> > >
> > >When the following command line is used on the current qemu version:
> > >
> > >x86_64-softmmu/qemu-system-x86_64 -kernel /boot/vmlinuz -drive
> > >file=~/disk0.raw,if=none,id=foo,boot=on -device virtio-blk-pci,drive=foo
> > >
> > >We get the following error reported:
> > >Two devices with same boot index 0
> > >
> > >Previous versions of qemu did not flag this as an error condition.
> >
> > Upstream QEMU does not have a boolean boot flag although I guess we
> > ignore it in -drive which sucks :-/
> >
> > In upstream QEMU, the BIOS can boot just fine from a virtio device.
> > What you're seeing is that we've apparently overloaded the boot flag
> > in upstream qemu to mean boot index.
> >
> > Gleb, what's the right invocation here?
> >
> Just drop boot=on. Qemu-kvm registers extboot and some other bootrom
> (which one?) with the same boot index. This should be fixed, but
Ah the second one is linux loader.
> dropping boot=on is the right solution in any case. Actually I want to
> remove extboot from qemu-kvm at all. It will not make it upstream
> anyway.
>
>
> > Regards,
> >
> > ANthony Liguori
> >
> > >I can see that we are indicating two different boot sources here, so I
> > >would guess the command line is invalid, but wanted to be sure.
> > >
> > >Bruce
> > >
> > >
>
> --
> Gleb.
--
Gleb.
