SAMBA digest 2491

[David Lee]

Re:

> Date:   Mon, 17 Apr 2000 14:44:57 +0200 (CEST)
> From:   Niels Kokholm <kokholm at math.ku.dk>
> To:     samba at samba.org
> Subject: Problem: Samba 2.0.7pre4 --with-utmp on OSF1
> 
> I tried configuring Samba 2.0.7pre4 on a OSF1 machine (actually a DEC
> Alpha running Digital Unix 4.0d) with --with-utmp. 
> 
> The autoconf system sets the variable HAVE_UTMPX_H
> 
> The linking of smbd aborts with the message 
> ----------------------------------------------------------
> ld:
> Unresolved:
> getutmpx
> utmpxname
> updwtmpx
> *** Exit 1
> ----------------------------------------------------------
> 
> If I insert into smbd/connection.c
> ----------------------------------------------------------
> #if defined (OSF1)
> #undef HAVE_UTMPX_H
> #endif
> ----------------------------------------------------------
> right after 
> ----------------------------------------------------------
> /*
>  * Apparently AIX has utmpx.h but doesn't implement it.
>  * The test for this ought to be (a) more automatic (b) elsewhere.
>  */
> #if defined (AIX)
> #undef HAVE_UTMPX_H
> #endif
> ----------------------------------------------------------
> The linking of smbd works ok and logging to the utmp file seems to
> work. There is no logging to the wtmp file, but that would be obvious from
> the comments in smbd/connection.c for this case:
> ----------------------------------------------------------
>      /* *** Hmmm.  Appending wtmp (as distinct from overwriting utmp) has
>         me baffled.  How is it to be done? *** */
> ----------------------------------------------------------

Sorry for the delay in replying: I was on holiday.

(I write as the person who has been coordinating the utmp development.)

Thanks for the report.  The "utmp" functionality is brand new to
"official" Samba at 2.0.7 .  It has turned out that each operating system
has its own little quirks in this area. 

Even during the course of 2.0.7pre1 through to pre3, the utmp code in
Samba was evolving rapidly to improve portability.  And, of course, this
tends to run counter to a principal aim of such a phase, which is to
stabilise the existing code.  So it was deliberately frozen (agreed by
Jeremy and me) at "pre3". 

Meanwhile (and while not on holiday!), I have continued to develop a
version of the utmp code with the prime aim of increasing portability to
different operating systems.

Now that 2.0.7 is released, I hope rapidly to retrofit my most recent
changes to that, and make them available.  I am reasonably confident that
these changes should address the issues you raise.  Indeed, I hope you
won't mind if, in a few days, I ask you to check them on your OS.

-- 

:  David Lee                                I.T. Service          :
:  Systems Programmer                       Computer Centre       :
:                                           University of Durham  :
:  http://www.dur.ac.uk/~dcl0tdl            South Road            :
:                                           Durham                :
:  Phone: +44 191 374 2882                  U.K.                  :