The changes look fine to me. I can sponsor this for you John.

-Chris.

On 06/20/2013 11:51 AM, Alan Bateman wrote:
On 19/06/2013 17:05, John Zavgren wrote:
Greetings:

Please consider the following change to the NIO DatagramChannel code
that conditionally sets the IP_MULTICAST_ALL socket option to "off"
when the Linux host is running a modern (greater than 2.6.30) kernel.
This option is used to prevent "promiscuous" receptions.

http://cr.openjdk.java.net/~jzavgren/8014377/webrev.03/
<http://cr.openjdk.java.net/%7Ejzavgren/8014377/webrev.03/>

As background, this is a follow-up to the thread on this list a few
weeks ago about setting IP_MULTICAST_ALL so that applications bound to
the same port but joining different multicast groups don't see
interference. Following that discussion then I created a prototype [1]
to check it out and verify that it doesn't have any side effects with
DatagramChannel. John has taken the baton to complete this and get it
working with java.net.MulticastSocket too. As per the original thread,
there doesn't appear to be a solution for IPv6 multicasting at this time.

Anyway, I'm fine with the fix. I think the only difference in John's
webrev is that the comment in Net.c has been moved/changed. I think I
prefer the original but it's not a big deal.

-Alan.

[1] http://cr.openjdk.java.net/~alanb/8014377/webrev/

Reply via email to