Date: Fri, 20 May 2011 14:01:45 +0200 From: Manuel Bouyer <bouyer@antioche.eu.org> To: Hugo Silva <hugo@barafranca.com> Cc: freebsd-xen@freebsd.org, port-xen@NetBSD.org Subject: Re: FreeBSD HVM+PV boot hangs with phy: run_interrupt_driven_hooks: still waiting after 60 seconds for xenbusb_nop_confighook_cb Message-ID: <20110520120145.GA1015@asim.lip6.fr> In-Reply-To: <20110520113518.GA9031@asim.lip6.fr> References: <4DB7616F.9060109@barafranca.com> <20110520113518.GA9031@asim.lip6.fr>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, May 20, 2011 at 01:35:18PM +0200, Manuel Bouyer wrote:
> >
> > If I create a FreeBSD VM and use file:, XENHVM will boot to completion.
> >
> > If I vnconfig that same file, then it will hang. On lvm storage it will
> > also hang.
> >
> > So it seems to be related to using phy:.
>
> It could be because with file:, the backend driver uses a vnd device attached
> to the file while qemu-dm opens the file directly. In the phy: case, both
> qemu-dm and the backend driver wants to open the device, and this may be
> denied by the kernel.
OK, I understand the problem now:
the block device backend driver wants a block device (e.g. /dev/vnd0d).
qemu-dm wants the raw device (e.g. /dev/rvnd0d). We should pach either
qemu-dm to add the 'r' to the path, or xend to do so.
My feeling it that's it's easier to do in qemu-dm at this time (we can
do a stat(2) first to see if the given path is a block device, and hack the
path if it is).
--
Manuel Bouyer <bouyer@antioche.eu.org>
NetBSD: 26 ans d'experience feront toujours la difference
--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20110520120145.GA1015>