Re: [libvirt] [PATCH v2 5/5] network: check accept_ra before enabling ipv6 forwarding
On Thu, 2017-03-23 at 12:09 -0400, John Ferlan wrote:
On 03/23/2017 04:44 AM, Cedric Bosdonnat wrote:
> On Wed, 2017-03-22 at 07:16 -0400, John Ferlan wrote:
> > [...]
> >
> > > +
> > > +static int
> > > +virNetDevIPCheckIPv6ForwardingCallback(const struct nlmsghdr *resp,
> > > + void *opaque)
> > > +{
> > > + struct rtmsg *rtmsg = NLMSG_DATA(resp);
> > > + int accept_ra = -1;
> > > + struct rtattr *rta;
> > > + char *ifname = NULL;
> > > + struct virNetDevIPCheckIPv6ForwardingData *data = opaque;
> > > + int ret = 0;
> > > + int len = RTM_PAYLOAD(resp);
> > > + int oif = -1;
> > > +
> > > + /* Ignore messages other than route ones */
> > > + if (resp->nlmsg_type != RTM_NEWROUTE)
> > > + return ret;
> > > +
> > > + /* Extract a few attributes */
> > > + for (rta = RTM_RTA(rtmsg); RTA_OK(rta, len); rta = RTA_NEXT(rta,
len)) {
> > > + switch (rta->rta_type) {
> > > + case RTA_OIF:
> > > + oif = *(int *)RTA_DATA(rta);
> > > +
> > > + if (!(ifname = virNetDevGetName(oif)))
> > > + goto error;
> > > + break;
> >
> > Did you really mean to break from the for loop if ifname is set? This
> > breaks only from the switch/case. Of course Coverity doesn't know much
> > more than you'd be going back to the top of the for loop and could
> > overwrite ifname again. It proclaims a resource leak...
>
> In my dev version I was also getting the RTA_DST attribute to print some
> debugging message that I removed in the submitted version. The break was
> here between the switch cases.
>
> In the current case I don't really care if we break out of the loop or not.
> As there aren't that many attributes in an rtnetlink message to loop over,
> breaking wouldn't gain a lot of cycles.
>
> What I don't get though is what this break is actually doing? isn't it
> for the switch case even though there is no other case after it?
>
> --
> Cedric
So should this be changed to:
if (rta->rta_type == RTA_OIF) {
oif = *(int *)RTA_DATA(rta);
if (!(ifname = virNetDevGetName(oif)))
goto error;
break;
}
leaving two questions in my mind
1. Can there be more than one RTA_OIF
2. If we don't finish the loop (e.g. we break out), then does one of
the subsequent checks fail?
IMHO getting more than one RTA_OIF per message would be a bug from the
kernel: a route is associated to one device. Think of these messages like
the lines printed by 'ip r'
Is it more of a problem that we find *two* RTA_OIF's and thus
should add a:
if (ifname) {
VIR_DEBUG("some sort of message");
goto error;
}
prior to the virNetDevGetName call and then remove the break;?
I can remove the break without that for sure.
John (who doesn't know the answer!)
BTW: The issue from patch4 is resolved by my 22 patch bomb from yesterday
Yep, seen that one: thanks a lot
--
Cedric
>
> > John
> > > + }
> > > + }
> > > +
> > > + /* No need to do anything else for non RA routes */
> > > + if (rtmsg->rtm_protocol != RTPROT_RA)
> > > + goto cleanup;
> > > +
> > > + data->hasRARoutes = true;
> > > +
> > > + /* Check the accept_ra value for the interface */
> > > + accept_ra = virNetDevIPGetAcceptRA(ifname);
> > > + VIR_DEBUG("Checking route for device %s, accept_ra: %d",
ifname, accept_ra);
> > > +
> > > + if (accept_ra != 2 && VIR_APPEND_ELEMENT(data->devices,
data->ndevices, ifname) < 0)
> > > + goto error;
> > > +
> > > + cleanup:
> > > + VIR_FREE(ifname);
> > > + return ret;
> > > +
> > > + error:
> > > + ret = -1;
> > > + goto cleanup;
> > > +}
> >
> >