> Based on idea contributed to iputils.
> https://github.com/iputils/iputils/commit/8ed7ffc999e2a541e06ee48faf26a323dfe487c2
Comparing some linux ping implementations that used glibc
getaddrinfo() which already converts IPv4-mapped address string to
IPv4 struct for requested AF_INET family
In case of nothing prevents to do this conversion:
1) Inetutils ping:
- explicit ipv4 ping:
% ping -c1 ::ffff:127.0.0.1
PING ::ffff:127.0.0.1 (127.0.0.1): 56 data bytes
64 bytes from 127.0.0.1: icmp_seq=0 ttl=64 time=0,100 ms
- explicit ipv6 ping:
% ping6 -c1 ::ffff:127.0.0.1
PING ::ffff:127.0.0.1 (::ffff:127.0.0.1): 56 data bytes
./ping/ping6: sending packet: Network is unreachable
2) Iputils ping with reverted two workarounds (one related to
rewriting requested ipv4 family and the last ones with target string
manipulations):
- explicitly requested ipv4 family:
$ ping -c1 -4 ::ffff:127.0.0.1
PING ::ffff:127.0.0.1 (127.0.0.1) 56(84) bytes of data.
64 bytes from localhost.localdomain (127.0.0.1): icmp_seq=1 ttl=64 time=0.065 ms
- explicitly requested ipv6 family:
$ ping -c1 -6 ::ffff:127.0.0.1
PING ::ffff:127.0.0.1 (::ffff:127.0.0.1) 56 data bytes
--- ::ffff:127.0.0.1 ping statistics ---
1 packets transmitted, 0 received, 100% packet loss, time 0ms
So, IPv4-mapped addresses are pinged in the same way if nothing
prevents getaddrinfo() to unmap them for requested AF_INET family
> + hostname = strrchr(hostname, ':') + 1;
>
If do that for AF_UNSPEC, suppose it'd be better to use data already
converted by getaddrinfo() without any string handling and additional
conversion calls, something like
```
struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)ai->ai_addr;
int rc = IN6_IS_ADDR_V4MAPPED(&sa6->sin6_addr);
if (rc) {
struct sockaddr_in sa4 = { .sin_family = AF_INET, .sin_addr.s_addr
= ((uint32_t*)&sa6->sin6_addr)[3] };
memcpy(ai->ai_addr, &sa4, sizeof(sa4));
ai->ai_addrlen = sizeof(sa4);
ai->ai_family = AF_INET;
}
```
Implicit or auto cases (when IP family is not requested explicitly
like in case of iputils `ping ::ffff:127.0.0.1`)
are under question in this case
Kind Regards
_______________________________________________
busybox mailing list
[email protected]
https://lists.busybox.net/mailman/listinfo/busybox
