Re: RFC: Making QEMU honour 'readonly' flag for disks

["Daniel P. Berrange" <berrange@redhat.com>]

On Wed, Jul 30, 2008 at 10:16:08AM +0100, Daniel P. Berrange wrote:
> On Wed, Jul 30, 2008 at 10:01:29AM +0100, Ian Jackson wrote:
> > Daniel P. Berrange writes ("[Xen-devel] RFC: Making QEMU honour 'readonly' flag for disks"):
> > > The Xen disk configuration syntax allows a block device to be marked as
> > > readonly, exclusive writable or shared writeable. The xen hotplug scripts
> > > will clash for clashing configs between domains, but  it is upto the
> > > backend drivers to actually enforce the readonly flag on I/O operations.
> > > The paravirt backend disk driver does this fine, but QEMU's emulated
> > > backend driver does not.
> > 
> > I still think this is a fix we should have but your patch is very
> > intrusive.  Is there some reason why you didn't just invent
> > BDRV_O_RDONLY_NO__ACTUALLY__READONLY ?  A new parameter to bdrv_new
> > seems quite wrong.
> > 
> > (It's a shame that the existing BDRV_O_RDONLY does something strange
> > and probably wrong, but we probably don't want to fix that in our
> > branch.)
> 
> Yeah, I'm actually attempting to fix that craziness, which is why
> I've not posted an update yet. I think it may actually be easier
> to fix that it appears....

Take a look at my patch on qemu-devel subject of

  "PATCH: Control over drive open modes for backing file"

If that gets into QEMU, the xen specific bit will merely involve changing
the xenstore.c file to pass correct flags to bdrv_open() instead of using
the default '0'.

Regards,
Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|