Ticket History #569: PATCH: AIX/DUNIX phantom wtmp entries

From schwim@whatmore.Stanford.EDU Thu Mar 19 08:36:30 1998
Received: from MIT.EDU (SOUTH-STATION-ANNEX.MIT.EDU [18.72.1.2]) by rt-11.MIT.EDU (8.7.5/8.7.3) with SMTP id IAA28114 for <bugs@RT-11.MIT.EDU>; Thu, 19 Mar 1998 08:36:29 -0500
Received: from whatmore.Stanford.EDU by MIT.EDU with SMTP
id AA01651; Thu, 19 Mar 98 08:36:23 EST
Received: (from schwim@localhost)
by whatmore.Stanford.EDU (8.8.8/8.8.8) id FAA06160;
Thu, 19 Mar 1998 05:36:16 -0800 (PST)
Message-Id: <199803191336.FAA06160@whatmore.Stanford.EDU>
Date: Thu, 19 Mar 1998 05:36:16 -0800 (PST)
From: Larry Schwimmer <schwim@whatmore.Stanford.EDU>
To: krb5-bugs@MIT.EDU
Subject: PATCH: AIX/DUNIX phantom wtmp entries


Responsible-Changed-From-To: gnats-admin->tlyu
Responsible-Changed-By: tlyu
Responsible-Changed-When: Sun Apr 5 21:21:33 1998
Responsible-Changed-Why:

Refiled

State-Changed-From-To: open-analyzed
State-Changed-By: tlyu
State-Changed-When: Sun Apr 5 21:29:11 1998
State-Changed-Why:

Weird.


From: Tom Yu <tlyu@MIT.EDU>
To: schwim@whatmore.Stanford.EDU
Cc: krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Sun, 5 Apr 1998 21:33:52 -0400

That's really weird. I presume that NO_UT_PID is not getting defined
on either system?

---Tom

From: Larry Schwimmer <schwim@whatmore.Stanford.EDU>
To: tlyu@MIT.EDU (Tom Yu)
Cc: krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Sun, 5 Apr 1998 21:52:57 -0700 (PDT)

You (Tom Yu) write:

Correct. They have ut_pid in struct utmp. I don't think
their last deals with it properly.

yours,
Larry

State-Changed-From-To: analyzed-feedback
State-Changed-By: tlyu
State-Changed-When: Mon Apr 6 20:30:41 1998
State-Changed-Why:

Should be fixed.

src/util/pty/update_utmp.c 1.17


From: Tom Yu <tlyu@MIT.EDU>
To: schwim@whatmore.Stanford.EDU
Cc: krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Mon, 6 Apr 1998 20:30:35 -0400

I've committed a patch to our sources that should fix this problem by
simply not writing out a LOGIN_PROCESS type entry to wtmp ever.
Please see if this works for you. I will probably include it in the
next patch release.

---Tom

Index: update_utmp.c
===================================================================
RCS file: /cvs/krbdev/krb5/src/util/pty/update_utmp.c,v
retrieving revision 1.16
retrieving revision 1.17
diff -u -r1.16 -r1.17
--- update_utmp.c 1998/03/12 23:12:52 1.16
+++ update_utmp.c 1998/04/06 23:38:11 1.17
@@ -194,5 +194,9 @@

#endif /* HAVE_SETUTENT */

- return ptyint_update_wtmp(&ent, host, userbuf);
+ /* Don't record LOGIN_PROCESS entries. */
+ if (process_type == PTY_LOGIN_PROCESS)
+ return 0;
+ else
+ return ptyint_update_wtmp(&ent, host, userbuf);
}

From: Larry Schwimmer <schwim@whatmore.Stanford.EDU>
To: tlyu@MIT.EDU (Tom Yu)
Cc: krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Mon, 6 Apr 1998 17:51:58 -0700 (PDT)

You (Tom Yu) write:

I added it into our source tree; I'll test it out. If this
will be added into the default tree, will the LOGIN_PROCESS
pty_update_utmp calls in appl/bsd/krlogind.c and
appl/telnet/telnetd/sys_term.c also be removed (since they're
not doing anything at that point)?

yours,
Larry

From: Tom Yu <tlyu@MIT.EDU>
To: Larry Schwimmer <schwim@whatmore.Stanford.EDU>
Cc: tlyu@MIT.EDU (Tom Yu), krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Mon, 6 Apr 1998 20:53:29 -0400

Not certain whether the calls to update_utmp that use a type of
LOGIN_PROCESS will be removed, since some systems do want that to
appear in utmp.

---Tom

From: Larry Schwimmer <schwim@whatmore.Stanford.EDU>
To: Cc: krb5-bugs@MIT.EDU
Subject: Re: pty/569: PATCH: AIX/DUNIX phantom wtmp entries
Date: Wed, 15 Apr 1998 17:47:28 -0700 (PDT)

You (Tom Yu) write:

Seems to work fine. Thanks.

yours,
Larry Schwimmer
schwim@leland.stanford.edu
Leland Systems Group

State-Changed-From-To: feedback-closed
State-Changed-By: tlyu
State-Changed-When: Thu Apr 16 22:32:53 1998
State-Changed-Why:

Thanks; closing this out now.

Submitter-Id: net
Originator: Larry Schwimmer
Confidential: no
Synopsis: AIX and DUNIX last do not understand psuedo entries
Severity: non-critical
Priority: low
Category: krb5-appl
Class: sw-bug
Release: 1.0.5
Environment: AIX, Digital UNIX
Description:

krlogind and telnetd call pty_update_utmp with "rlogin" and
"LOGIN" entries, respectively. This call generates a wtmp entry for
these psuedousers. last on AIX and Digital UNIX reports these
entries (as a login with no logout).

I partially understand why the call is made, but I'm not
sure it's really needed. tcpd and syslog provide logging information
about connections, so the wtmp entry seems unnecessary. For most
systems, it is not a problem, but for systems whose last does not
discard them, it is confusing and can be misleading.

Fix:

Either remove the code completely or ifdef it out for
the systems whose last does not understand it.

--- appl/bsd/krlogind.c.orig Wed Mar 18 17:42:26 1998
+++ appl/bsd/krlogind.c Wed Mar 18 17:43:12 1998
@@ -702,9 +702,10 @@

#ifndef NO_UT_PID
{
-
+#if !defined(_AIX) && !defined(__osf__)
pty_update_utmp(PTY_LOGIN_PROCESS, getpid(), "rlogin", line,
""/*host*/, PTY_TTYSLOT_USABLE);
+#endif
}
#endif

--- appl/telnet/telnetd/sys_term.c.orig Fri Feb 6 19:41:47 1998
+++ appl/telnet/telnetd/sys_term.c Wed Mar 18 00:59:24 1998
@@ -1093,9 +1093,11 @@
close(syncpipe[0]);

} else {
-
+
+#if !defined(_AIX) && !defined(__osf__)
pty_update_utmp (PTY_LOGIN_PROCESS, getpid(), "LOGIN", line,
host, PTY_TTYSLOT_USABLE);
+#endif
getptyslave(autologin);

/* Notify our parent we're ready to continue.*/