Re: [Qemu-devel] [PATCH v3 09/11] authz: add QAuthZListFile object type for a file access control list

["Dr. David Alan Gilbert" <dgilbert@redhat.com>]

* Daniel P. Berrangé (berrange@redhat.com) wrote:
> On Wed, Jun 20, 2018 at 11:58:01AM +0100, Dr. David Alan Gilbert wrote:
> > * Daniel P. Berrangé (berrange@redhat.com) wrote:
> > > Add a QAuthZListFile object type that implements the QAuthZ interface. This
> > > built-in implementation is a proxy around the QAtuhZList object type,
> > > initializing it from an external file, and optionally, automatically
> > > reloading it whenever it changes.
> > > 
> > > To create an instance of this object via the QMP monitor, the syntax
> > > used would be:
> > > 
> > >       {
> > >         "execute": "object-add",
> > >         "arguments": {
> > >           "qom-type": "authz-list-file",
> > >           "id": "authz0",
> > >           "parameters": {
> > >             "filename": "/etc/qemu/vnc.acl",
> > > 	    "refresh": "yes"
> > >           }
> > >         }
> > >       }
> > > 
> > > If "refresh" is "yes", inotify is used to monitor the file,
> > > automatically reloading changes. If an error occurs during reloading,
> > > all authorizations will fail until the file is next successfully
> > > loaded.
> > 
> > I'm curious about the 'refresh' stuff:
> >   a) If refresh=no is there a way to explicitly ask for a refresh
> >      when some tool knows it's finished with fiddling with the file.
> 
> If refresh=no, then you can still use  object_del + object_add to
> recreate the ACL object which will cause new content to be picked
> up.

But if I have a VNC/NBD/etc listening, won't it be bound to the old
object, so I can't delete the old object?

> >   b) am I right in thinking you're actually monitoring the directory
> >      for changes and retriggering on directory changes?  So it'll
> >      trigger on the normal trick of renaming the modified file over the
> >      original?
> 
> I originally made the mistake of trying to monitor the file
> and discovered the awfulness of inotify :-(
> 
> So, yeah, now we monitor the directory so we can trigger
> reloads even when the file is editted by doing a tmpfile+rename
> dance.
> 
> The inotify support in the early patch in this series is intelligent
> so that if multiple objects all monitor the same directory we only
> use a single inotify watch.
> 
> >   c) I'm a little concerned about making it automatic; if all the qemus
> >      on a  system all wake up and reparse the file at the same time.
> 
> I don't think its a big deal in this case because the files in
> question are very small. Once one QEMU reads it, the kernel will
> have content in I/O cache, and the CPU overhead of parsing the JSON
> will be lost in the noise given likely file sizes IMHO.

OK.

Dave

> Regards,
> Daniel
> -- 
> |: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
> |: https://libvirt.org         -o-            https://fstop138.berrange.com :|
> |: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
--
Dr. David Alan Gilbert / dgilbert@redhat.com / Manchester, UK