IPv6: #259 #253#276
Conversation
|
@AkihiroSuda can you enable ipv6 for the Docker containers? |
| int is_udp; | ||
| struct in_addr host_addr; | ||
| int is_ipv4; | ||
| int is_ipv6; |
There was a problem hiding this comment.
We shouldn’t need is_ipv4.
Checking !is_ipv6 should be enough.
There was a problem hiding this comment.
We can pass tcp6/4 as proto to explicitly just enable either.
If we pass tcp as proto, ipv6 is enabled and the host address
is either 0.0.0.0 or ::0, both are enabled.
| (namespace)$ json='{"execute": "list_hostfwd"}' | ||
| (namespace)$ echo -n $json | nc -U /tmp/slirp4netns.sock | ||
| {"return": {"entries": [{"id": 42, "proto": "tcp", "host_addr": "0.0.0.0", "host_port": 8080, "guest_addr": "10.0.2.100", "guest_port": 80}]}} | ||
| {"return": {"entries": [{"id": 42, "proto": "tcp", "host_addr": "0.0.0.0", "host_addr6": "::", "host_port": 8080, "guest_addr": "10.0.2.100", "guest_addr6": "fd00::100", "guest_port": 80}]}} |
There was a problem hiding this comment.
I’m not sure we should mix up v4 and v6 in a single entry.
There was a problem hiding this comment.
What should we return in api_handle_req_add_hostfwd when we add both tcp4 and tcp6? An array of IDs?
There was a problem hiding this comment.
Avoid adding tcp4 and tcp6 in a single request/response.
Mixing up them in a single request/response makes it hard to implement RootlessKit integration.
https://github.com/rootless-containers/rootlesskit/blob/c5481c1cfd6f6c42c8ec1868fbc543f2b0a916a5/pkg/api/openapi.yaml#L70-L90
There was a problem hiding this comment.
I tried to implement it like Go's net listen, which does listen on both 4 and 6 if you set tcp with global ip 0.0.0.0 or :: if I understood it correctly.
If we don't need that I can change it like that?
- host_addr: 0.0.0.0, proto: tcp -> v4
- host_addr: 0.0.0.0, proto: tcp6 -> error
- host_addr: ::, proto: tcp -> v6
- host_addr: ::, proto: tcp4 -> error
|
Thanks for working on this. Could you use real name for signing?
Do you mean Docker containers on the CI? I don’t think we have IPv6 there. |
- implement forwarding - implement random CIDR Signed-off-by: Jasmin Fazlic <superfassl@gmail.com>
Yes, ok too bad, I guess we can then test the port forwarding in the vagrant box. |
giuseppe
left a comment
giuseppe
left a comment
There was a problem hiding this comment.
left some comments.
Could you please describe what you are doing in the commit message?
| goto finish; | ||
| } | ||
| #if SLIRP_CONFIG_VERSION_MAX >= 3 | ||
| int flags = (fwd->is_udp ? SLIRP_HOSTFWD_UDP : 0); |
There was a problem hiding this comment.
could we also check that SLIRP_HOSTFWD_UDP is defined?
| free(fwd); | ||
| goto finish; | ||
| } | ||
| flags |= SLIRP_HOSTFWD_V6ONLY; |
There was a problem hiding this comment.
SLIRP_HOSTFWD_V6ONLY might not be defined
| goto finish; | ||
| } | ||
| flags |= SLIRP_HOSTFWD_V6ONLY; | ||
| if (slirp_add_hostxfwd(slirp, (const struct sockaddr *)&fwd->host6, |
There was a problem hiding this comment.
we need to check for slirp_add_hostxfwd at configure time
| wrc = write(fd, api_ok, strlen(api_ok)); | ||
| if (fwd->is_ipv4) { | ||
| #if SLIRP_CONFIG_VERSION_MAX >= 3 | ||
| if (slirp_remove_hostxfwd(slirp, |
There was a problem hiding this comment.
same for slirp_remove_hostxfwd
There was a problem hiding this comment.
Edit: nevermind
Checking for SLIRP_HOSTFWD_V6ONLY at the moment would imply SLIRP_HOSTFWD_UDP and the xfwd functions. Can we just check for that to enable the IPv6 code path?
There was a problem hiding this comment.
sure, that should work as well
|
What's current status? |
Sorry, I'm out. |
|
I've been running this branch built with libslirp from master for well over a month now and I'm yet to experience any issues. |
|
I'm interested in working on this. Can I hijack this PR and @pfandl's work ? |
Yes, please |
Do you still plan this? |
5 participants
Signed-off-by: fassl superfassl@gmail.com