Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!uunet!kaukau.comp.vuw.ac.nz!mark
From: mark@comp.vuw.ac.nz (Mark Davies)
Newsgroups: comp.sys.hp
Subject: dump/rdump under 7.0
Message-ID: <1990Apr09.222130.10246@kaukau.comp.vuw.ac.nz>
Date: 9 Apr 90 22:21:30 GMT
Sender: news@kaukau.comp.vuw.ac.nz (News Admin)
Organization: Dept. of Comp. Sci., Victoria Uni. of Wellington, New Zealand.
Lines: 36

I have tracked down my problem with dump under hpux 7.0.  If the group
operator has any members there is a possibility that dump will get stuck
broadcasting messages to those members.

The probable cause is someone forgot about the difference in structure in
/etc/utmp when porting dump from BSD (I should have known -- see next
posting).

Working from the bsd source the following patch to the broadcast routine
fixes the problem. (Temporary fix: don't list any members in group operator).

*** dumpoptr.c~	Mon May 22 11:40:18 1989
--- dumpoptr.c	Mon Apr  9 15:10:15 1990
***************
*** 181,187 ****
--- 181,191 ----
  	while (!feof(f_utmp)){
  		if (fread(&utmp, sizeof (struct utmp), 1, f_utmp) != 1)
  			break;
+ #ifndef __hpux
  		if (utmp.ut_name[0] == 0)
+ #else
+ 		if (utmp.ut_type != USER_PROCESS || utmp.ut_name[0] == '\0')
+ #endif
  			continue;
  		nusers++;
  		for (np = gp->gr_mem; *np; np++){


Thanks to Michael J. Corrigan <mcorrigan@UCSD.EDU> for trying out some
tests on his system and to Curt Vandetta <curt@sylvestr.oce.orst.edu> for
sending me his port of BSD dump which was the starting point of my own port
which exposed the original problem with HP's.

cheers
mark