On 10/03, Mina Almasry wrote:

On Mon, Sep 30, 2024 at 10:18 AM Stanislav Fomichev [off-list ref] wrote:

quoted

Use dualstack socket to support both v4 and v6. v4-mapped-v6 address
can be used to do v4.

The network on test machines I can run ncdevmem on are ipv4 only, and
ipv6 support is not possible for my test setup at all. Probably a mega
noob question, but are these changes going to regress such networks?
Or does the dualstack support handle ipv4 networks seamlessly?

If such regression is there, maybe we can add a -6 flag that enables
ipv6 support similar to how other binaries like nc do it?

As long as your kernel is compiled with IPv6 (which all kernel in the
past 10+ years do) and you don't toggle net.ipv6.bindv6only sysctl to 1
(which defaults to 0 and afaik, none of the distros change it to 1),
AF_INET6 should properly support both v4 and v6 even if you don't have any
v6 environment setup.