Fix handling of session key for Kerberos5. I don't think this should
authorSam Hartman <hartmans@mit.edu>
Tue, 10 Oct 1995 03:13:48 +0000 (03:13 +0000)
committerSam Hartman <hartmans@mit.edu>
Tue, 10 Oct 1995 03:13:48 +0000 (03:13 +0000)
fix the mutual authentication bug with beta 4, but this should help
forwarding credentials and should also help if someone actually
defines ENCRYPTION.

If user defines ENCRYPTION, telnetd has the nasty behavior of
encrypting one side of the conversation, without telling the user; in
particular, only server output is encrypted.  Fixed by sending a do
encrypt if encryption defined; fix taken from Athena telnetd.

git-svn-id: svn://anonsvn.mit.edu/krb5/trunk@6955 dc483132-0cff-0310-8789-dd5450dbe970

src/appl/telnet/telnetd/ChangeLog
src/appl/telnet/telnetd/authenc.c
src/appl/telnet/telnetd/telnetd.c

index 727cfa1beb43f43f3f90956461568f0caf5c1495..da63e0e126c66fc741cc151750881841757e211b 100644 (file)
@@ -1,3 +1,8 @@
+Mon Oct  9 23:01:36 1995  Sam Hartman  <hartmans@tertius.mit.edu>
+
+       * telnetd.c (getterminaltype): If ENCRYPTION defined, send do
+        encryption, *not* just will encrypt. if you don't do this, then you won't encrypt the input to the server; your password prompt is encrypted, but your password isn't.
+
 Tue Aug  8 11:26:54 1995  Sam Hartman  <hartmans@pao.mit.edu>
 
        * sys_term.c: Fail if the slave side dies.
index 3aec7f9d2e7aefe963098c96dbf2150182a9fce7..5736698941dde3b0f6789d3032a2e499ef194243 100644 (file)
@@ -87,3 +87,7 @@ telnet_gets(prompt, result, length, echo)
        return((char *)0);
 }
 #endif /* defined(AUTHENTICATION) || defined(ENCRYPTION) */
+
+
+
+
index 9fc2bd8a40938e56b93d8965bbe11c8f0fbb3feb..b91343ebf7230dcf092b63e6593cfa9706ffd135 100644 (file)
@@ -650,6 +650,7 @@ getterminaltype(name)
 
 #ifdef ENCRYPTION
     send_will(TELOPT_ENCRYPT, 1);
+    send_do(TELOPT_ENCRYPT, 1);
 #endif /* ENCRYPTION */
     send_do(TELOPT_TTYPE, 1);
     send_do(TELOPT_TSPEED, 1);