| To: | "Michael Richardson" <mcr@xxxxxxxxxxxxxxxxxxxxxx> |
|---|---|
| Subject: | Re: [uml-devel] multicast on loopback: mcast networking at the cottage |
| From: | stian@xxxxxxxx |
| Date: | Mon, 25 Jul 2005 19:20:57 +0200 (CEST) |
| Cc: | "user-mode-linux-devel" <user-mode-linux-devel@xxxxxxxxxxxxxxxxxxxxx>, netdev@xxxxxxxxxxx |
| Importance: | Normal |
| In-reply-to: | <6048.1122248129@marajade.sandelman.ottawa.on.ca> |
| References: | <6048.1122248129@marajade.sandelman.ottawa.on.ca> |
| Sender: | netdev-bounce@xxxxxxxxxxx |
| User-agent: | SquirrelMail/1.4.3a-0.f1.1 |
> I suspect that for people who want to do multicast between hosts on an > actual wire won't want to have just a route to lo, they may want a route > for 224.0.0.0/4 on all of their interfaces. Maybe that should be > automatic? > > I think that applications that are multicast aware are supposed to > figure out which interfaces that they want to bind to anyway, so they > should not be confused by the multiple routes. If I remember things right (I might be wrong here): ipv6 does this automatically, so should ipv4 have done too if multicast is supported on a device. Multicast aware programs should always bind to a device, or broadcast on all devices for discovery. This is not uml-specific, but vanilla kernel. Some distroes add multicast-routing for ipv4 during init scripts. Stian |
| <Prev in Thread] | Current Thread | [Next in Thread> |
|---|---|---|
| ||
| Previous by Date: | multicast on loopback: mcast networking at the cottage, Michael Richardson |
|---|---|
| Next by Date: | PPP dropping packets?, Antti Salmela |
| Previous by Thread: | multicast on loopback: mcast networking at the cottage, Michael Richardson |
| Next by Thread: | PPP dropping packets?, Antti Salmela |
| Indexes: | [Date] [Thread] [Top] [All Lists] |