Re: Small mrouted improvements

Claudio Jeker <[email protected]> Thu, 16 Jul 2026 08:54:57 +0200
Newsgroups gmane.os.openbsd.tech
Message-ID <[email protected]>
On Wed, Jul 15, 2026 at 04:39:11PM +0200, Alexander Bluhm wrote:
> On Sun, Jun 28, 2026 at 12:48:34PM +0100, Russell Howe wrote:
> > I saw a couple of opportunities to improve mrouted's output.
> > 
> > First, if SIOCGETVIFCNT fails, it's useful to know the error code and
> > which vif it failed for:
> > 
> > diff --git usr.sbin/mrouted/vif.c usr.sbin/mrouted/vif.c
> > index 82c2f1b481b..b25f33150d4 100644
> > --- usr.sbin/mrouted/vif.c
> > +++ usr.sbin/mrouted/vif.c
> > @@ -1333,8 +1333,7 @@ dump_vifs(FILE *fp)
> >         }
> >         v_req.vifi = vifi;
> >         if (ioctl(udp_socket, SIOCGETVIFCNT, (char *)&v_req) == -1) {
> > -           logit(LOG_WARNING, 0,
> > -               "SIOCGETVIFCNT fails");
> > +           logit(LOG_WARNING, errno, "SIOCGETVIFCNT fails for vif %d", vifi);
> >         }
> >         else {
> >             fprintf(fp, "                         pkts in : %ld\n",
> > 
> 
> I have commited this part.
> 
> > Second, when logging the ratelimit for a multicast peer, include the
> > units.
> > 
> > diff --git usr.sbin/mrouted/config.c usr.sbin/mrouted/config.c
> > index 9ba913f577f..a3394f2660c 100644
> > --- usr.sbin/mrouted/config.c
> > +++ usr.sbin/mrouted/config.c
> > @@ -101,7 +101,7 @@ config_vifs_from_kernel(void)
> >         v->uv_acl         = NULL;
> >         v->uv_addrs       = NULL;
> > 
> > -       logit(LOG_INFO,0,"installing %s (%s on subnet %s) as vif #%u - rate=%d",
> > +       logit(LOG_INFO,0,"installing %s (%s on subnet %s) as vif #%u - rate=%dbit/s",
> >             v->uv_name, inet_fmt(addr, s1), inet_fmts(subnet, mask, s2),
> >             numvifs, v->uv_rate_limit);
> > 
> 
> Man page says unit is kilo bit per second.
> Kernel feature was removed here:
> https://github.com/openbsd/src/commit/9500e0064b973fbd2fac185909ed03e90413f2e2
> 
> Would it be better to remove the remains of rate limit in the daemon?

Agreed. We will most probably not bring this feature back.

-- 
:wq Claudio