On Mar 11, 2014, at 11:40 AM, Gisle Vanem <gva...@yahoo.no> wrote:
> Applying this patch fixes it though:
Checked into the trunk and pushed.
> Time for an INET6 aware util.c print-function for these cases?
ip6addr_string() is a macro wrapper for getname6().
getname6() uses both gethostbyaddr() and inet_ntop() with AF_INET6, so it
currently requires those two functions to exist and requires AF_INET6 to be
defined and supported by those functions.
If a platform has inet_ntop() and AF_INET6, and inet_ntop() supports AF_INET6,
but it doesn't have gethostbyaddr() or it has one that doesn't support
AF_INET6, we could have getname6() just use that and skip gethostbyaddr(); I
don't know if there are any such platforms, however.
Otherwise, we'd have to duplicate inet_ntop() and use that; fortunately, we
*have* duplicated it, in missing/inet_ntop.c. Unfortunately, it only supports
AF_INET6 if INET6 is defined. What we could do, I guess, is:
if AF_INET6 isn't defined for the platform, define it ourselves, with
some appropriately-chosen value;
always build getname6() and always define ip6addr_string();
have a #define to indicate whether gethostbyaddr() supports IPv6, and
use that to control whether to use it in getname6();
and then get rid of all the "#ifdef INET6" that only check whether we have
ip6addr_string() and getname6().
That might not work as we want on systems that have inet_ntop() and either
don't define AF_INET6 or don't support it in inet_ntop(), however. I don't
know whether there are any such platforms.
_______________________________________________
tcpdump-workers mailing list
tcpdump-workers@lists.tcpdump.org
https://lists.sandelman.ca/mailman/listinfo/tcpdump-workers
