IPv6

50 views
Skip to first unread message

Felix Morgner

unread,
May 1, 2014, 9:18:20 AM5/1/14
to umurmur...@googlegroups.com
Hi

I've implemented a basic version of IPv6 support for umurmur. I would be glad if you took a look at the code in my fork and
would provide comments. The implementation currently works as follows:

- Two new commandline parameters were added (-A, -B) which work analogously to the -a, -b flags but for IPv6
- Two new config parameters were added (bindaddr6, bindport6)
- IPv4 and IPv6 connections can be configured separately.
- If no addresses are provided both IPv4 and IPv6 bind to the "wildcard address"
- Currently the only way to "disable" IPv6 is to bind to localhost (::1)
- Address base banning is not implemented yet

Again, please provide feedback and file bugs in my fork (if thats possible).

Felix

signature.asc

Felix Morgner

unread,
May 4, 2014, 4:43:51 PM5/4/14
to umurmur...@googlegroups.com
Hi

Just wanted to let you know that I pushed a fix for the banning issue, plus some code cleanup to the branch.

Felix

Felix Morgner

unread,
May 22, 2014, 12:33:14 PM5/22/14
to umurmur...@googlegroups.com
Hi

I just finished the patch that hopefully should handle systems without IPv6 support. Sadly I dont't have access to a non v6 system right now and thus could not fully test the patch yet.

> On a related note, should ipv6 support be a compile-time option, like a '--disable-ipv6' switch to the configure script? I'm thinking of scenarios where the target system lacks support and will give compile and/or link errors. These systems should be scarce nowadays I suppose and I think this will mess up the code quite a bit...

I think we shouldn't do something about this right now, as IPv6 support lost its experimental marking in Linux with version 2.6.12 (in 2005 if i remember correctly) and on other systems the situation is similar. I suppose we should deal with it, when the issue arises. Being able to avoid it all together would make it a lot easier to maintain the code, as handling this issue would hugely increase code size ( we would need to handle the address types separately for these platforms).

Felix
signature.asc

Martin Johansson

unread,
Jun 17, 2014, 2:48:01 PM6/17/14
to Felix Morgner, umurmur-general
Hi!

Sorry for my late response (again). I wanted to review the code before merging your pull request and since it is quite a lot of changes it takes time. This far I think it looks good and I have tested it successfully on OpenWRT (Linux uClibc). I found a couple of things but intend to merge anyway and fix these after. The only things I noted was:

o missing util.c in src/Makefile.am
o I'm suspicious about the ipv6 address writing/reading to the banfile in ban.c

/Martin

Reply all
Reply to author
Forward
0 new messages