On Thu, Dec 19, 2013 at 11:32:32AM -0700, Eric Blake wrote:
On 12/19/2013 08:23 AM, Daniel P. Berrange wrote:
> On Thu, Dec 19, 2013 at 08:15:09AM -0700, Eric Blake wrote:
>> diff --git a/src/libvirt-qemu.c b/src/libvirt-qemu.c
>> index db52c65..849932d 100644
>> + if (dom &&
>> + !(VIR_IS_CONNECTED_DOMAIN(dom) && dom->conn == conn)) {
>> + virLibConnError(VIR_ERR_INVALID_CONN, __FUNCTION__);
>> + virDispatchError(conn);
>> + return -1;
>> + }
>> + virCheckNonNullArgGoto(cb, error)
>
>
> I have a gut feeling that we should restrict use of this API to
> authenticated users only. So add a check for a read-only connection
> here
Hmm. It would match what we have for qemu-monitor-command; on the other
hand, it differs from what we have for normal events (libvirt events are
fine on a read-only connection). I guess an argument in favor of
requiring write privileges is that qemu may expose details in its events
that are better left internal to libvirt (that is, libvirt has a chance
to scrub details before handing information to read-only guests, but
with raw event handling, there is no scrubbing). And it's always easier
to relax things later than it is to add read-write restrictions later
and break existing callers that had grown used to read-only.
Yes, my concern is that there could one day be a QEMU event that includes
some sensitive information in its parameters. Normal users should never
need to use this API anyway, so restricting its access shouldn't really
hurt people.
Daniel
--
|:
http://berrange.com -o-
http://www.flickr.com/photos/dberrange/ :|
|:
http://libvirt.org -o-
http://virt-manager.org :|
|:
http://autobuild.org -o-
http://search.cpan.org/~danberr/ :|
|:
http://entangle-photo.org -o-
http://live.gnome.org/gtk-vnc :|