ipv6 leaks

[xeverse]

not sure whats the current
status of the ipv6 support dev
but can we at least prevent
ipv6 leaks with the ipv4?

[JsBergbau]

Current status is no IPv6 support, since I don't have IPv6, since I don't see currently any benefit. It makes things only more complicated, since you have to think about IPv4 und IPv6 rules.

To prevent IPv6 leak, force your application to use IPv4. Maybe via your own DNS-Server that only answers A records, but no AAAA records. If that is not possible, don't set an IPv6 address. Then your program can't access via IPv6.

[xeverse]

no6v thats too bad( i used to
disable aaaa on dnscrypt/dnsmasq.
that effectively kills off v6 systemwide.
but thats sumkinda perversion. sux innit.
ive been experimenting with yggdrasil lately
so i've decided to revive my ipv6 proto.

can we just disable/prevent these
ipv6 leaks in the v4 code somehow?

if your ISP denies you IPv6:
Yes, you can still have real IPv6.
Easiest modern method: dual-stack VPS
& WireGuard tunneling your prefix home.
Old school: HE 6in4 brokers.

[witrose1230]

Hi, I'm a technology enthusiast interested in dual-stack networking. I've analyzed the code and noticed that currently the library only handles IPv4 connections (AF_INET). IPv6 connections are not intercepted, which leads to IPv6 leaks as you observed.

I propose to add IPv6 support to bindToInterface.c by:

1. Adding conditional handling for AF_INET6 in the connect() wrapper.
2. Implementing IPv6 source IP binding (BIND_SOURCE_IP for IPv6 addresses).
3. Making DNS override and BIND_EXCLUDE work with IPv6 addresses.
4. Ensuring SO_BINDTODEVICE works for both IPv4 and IPv6 sockets (already works).

This would allow the library to properly bind IPv6 connections to the specified interface, preventing IPv6 leaks and providing full dual-stack support.

I'll start implementing and submit a pull request for review. Let me know if there are any specific requirements or edge cases you'd like me to consider.