[libvirt] [PATCH 11/23] Fix leak of iterators in virDBusMessageIterEncode

[Daniel P. Berrange]

On Wed, Sep 25, 2013 at 10:45:00AM -0600, Eric Blake wrote:
> On 09/25/2013 08:51 AM, Daniel P. Berrange wrote:
> > From: "Daniel P. Berrange" <berrange at redhat.com>
> > 
> > If virDBusMessageIterEncode hits an OOM condition it often
> > leaks the memory associated with the dbus iterator object
> > 
> > Signed-off-by: Daniel P. Berrange <berrange at redhat.com>
> > ---
> 
> >  
> >  cleanup:
> > +    while (nstack > 0) {
> > +        DBusMessageIter *thisiter = iter;
> > +        VIR_DEBUG("Popping iter=%p", iter);
> > +        if (virDBusTypeStackPop(&stack, &nstack, &iter,
> > +                                &types, &nstruct, &narray) < 0)
> > +            goto cleanup;
> 
> I'm worried that this will infloop.  What guarantee do we have that
> nstack reduces when virDBusTypeStackPop fails?

It can only return -1, if nstack == 0, so I think we're fine there

static int virDBusTypeStackPop(virDBusTypeStack **stack,
                               size_t *nstack,
                               DBusMessageIter **iter,
                               const char **types,
                               size_t *nstruct,
                               size_t *narray)
{
    if (*nstack == 0) {
        virReportError(VIR_ERR_INTERNAL_ERROR, "%s",
                       _("DBus type stack is empty"));
        return -1;
    }

    *iter = (*stack)[(*nstack) - 1].iter;
    *types = (*stack)[(*nstack) - 1].types;
    *nstruct = (*stack)[(*nstack) - 1].nstruct;
    *narray = (*stack)[(*nstack) - 1].narray;
    VIR_DEBUG("Popped '%s'", *types);
    VIR_SHRINK_N(*stack, *nstack, 1);

    return 0;
}

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 :|