Ticket History #1435: cygwin does not have inet_ntop when compiling kdc

From liebman@zod.com Sun Apr 20 19:42:27 2003
Received: from fort-point-station.mit.edu (FORT-POINT-STATION.MIT.EDU [18.7.7.76]) by krbdev.mit.edu (8.9.3) with ESMTP
id TAA18610; Sun, 20 Apr 2003 19:42:27 -0400 (EDT)
From: liebman@zod.com
Received: from zod.com (66-215-26-54.pas-mres.charterpipeline.net [66.215.26.54])
by fort-point-station.mit.edu (8.12.4/8.9.2) with ESMTP id h3KNgQGm002403
for <krb5-bugs@mit.edu>; Sun, 20 Apr 2003 19:42:27 -0400 (EDT)
Received: from liebman by zod.com with local (Exim 4.14)
id HDO1UE-0001HK-9S
for krb5-bugs@mit.edu; Sun, 20 Apr 2003 16:42:14 -0700
To: krb5-bugs@mit.edu
Subject: cygwin patch for keb5-current and 1.3alpha2 patch #2
Reply-To: liebman@zod.com
Cc:
X-send-pr-version: 3.99
Message-Id: <EHDO1UE-0001HK-9S@zod.com>
Date: Sun, 20 Apr 2003 16:42:14 -0700


none

System: CYGWIN_NT-5.0 XYZZY 1.3.22(0.78/3/2) 2003-03-18 09:20 i686 unknown unknown Cygwin


when compiling the kdc it fails with inet_ntop as undefined
configure and type make
Only use inet_ntop if HAVE_INET_NTOP is defined. The following patch
edits the 4 files in the kdc that use inet_ntop.



diff -u -r krb5-current-dist/src/kdc/dispatch.c krb5-current/src/kdc/dispatch.c
--- krb5-current-dist/src/kdc/dispatch.c 2002-09-17 00:00:34.000000000 -0700
+++ krb5-current/src/kdc/dispatch.c 2003-04-20 14:03:26.000000000 -0700
@@ -55,8 +55,20 @@
const char *name = 0;
char buf[46];

+#ifdef HAVE_INET_NTOP
name = inet_ntop (ADDRTYPE2FAMILY (from->address->addrtype),
from->address->contents, buf, sizeof (buf));
+#else
+ if (from->address->addrtype == ADDRTYPE_INET) {
+ struct sockaddr_in *sin
+ = (struct sockaddr_in *)from->address->contents;
+ if (name = inet_ntoa (sin->sin_addr))
+ {
+ strcpy (buf, name);
+ name = buf;
+ }
+ }
+#endif
if (name == 0)
name = "[unknown address type]";
krb5_klog_syslog(LOG_INFO,
diff -u -r krb5-current-dist/src/kdc/do_as_req.c krb5-current/src/kdc/do_as_req.c
--- krb5-current-dist/src/kdc/do_as_req.c 2002-11-04 00:00:44.000000000 -0800
+++ krb5-current/src/kdc/do_as_req.c 2003-04-20 13:57:28.000000000 -0700
@@ -89,9 +89,14 @@
ktypes2str(ktypestr, sizeof(ktypestr),
request->nktypes, request->ktype);

+#ifdef HAVE_INET_NTOP
fromstring = inet_ntop(ADDRTYPE2FAMILY (from->address->addrtype),
from->address->contents,
fromstringbuf, sizeof(fromstringbuf));
+#else
+ if (from->address->addrtype == ADDRTYPE_INET)
+ fromstring = (char *) inet_ntoa(*(struct in_addr *)from->address->contents);
+#endif
if (!fromstring)
fromstring = "<unknown>";

diff -u -r krb5-current-dist/src/kdc/do_tgs_req.c krb5-current/src/kdc/do_tgs_req.c
--- krb5-current-dist/src/kdc/do_tgs_req.c 2003-04-02 00:00:46.000000000 -0800
+++ krb5-current/src/kdc/do_tgs_req.c 2003-04-20 13:57:28.000000000 -0700
@@ -103,9 +103,15 @@
if ((retval = setup_server_realm(request->server)))
return retval;

+#ifdef HAVE_INET_NTOP
fromstring = inet_ntop(ADDRTYPE2FAMILY(from->address->addrtype),
from->address->contents,
fromstringbuf, sizeof(fromstringbuf));
+#else
+ if (from->address->addrtype == ADDRTYPE_INET)
+ fromstring =
+ (char *) inet_ntoa(*(struct in_addr *)from->address->contents);
+#endif
if (!fromstring)
fromstring = "<unknown>";

diff -u -r krb5-current-dist/src/kdc/network.c krb5-current/src/kdc/network.c
--- krb5-current-dist/src/kdc/network.c 2003-01-04 00:00:42.000000000 -0800
+++ krb5-current/src/kdc/network.c 2003-04-20 13:57:28.000000000 -0700
@@ -742,10 +742,24 @@
if (cc == -1) {
char addrbuf[46];
krb5_free_data(kdc_context, response);
+#ifdef HAVE_INET_NTOP
if (inet_ntop(((struct sockaddr *)&saddr)->sa_family,
addr.contents, addrbuf, sizeof(addrbuf)) == 0) {
strcpy(addrbuf, "?");
}
+#else
+ {
+ char* addrstr;
+ if (addrstr = inet_ntoa(((struct sockaddr_in *)&saddr)->sin_addr))
+ {
+ strncpy(addrbuf, addrstr, sizeof(addrbuf));
+ }
+ else
+ {
+ strcpy(addrbuf, "?");
+ }
+ }
+#endif
com_err(prog, errno, "while sending reply to %s/%d",
addrbuf, faddr.port);
return;

If changes are needed in that many places, I'd rather have
port-sockets.h provide an inline version of inet_ntop, to keep the KDC
code simpler, that calls sprintf, since inet_ntoa isn't thread-safe.

Does Cygwin support IPv6 currently?
Do we need to worry about AF_INET6 addresses?

Ken

Ken,
I'll give it a try and let you know!
-- Chris

Ken
It seems to be working. I've installed and authenticated! :-) FYI,
cygwin does not have ipv6 support. Thanks!

-- Chris