This is a list of bugs for my NetBSD source collection for Linux.

Please email me, if you can give more detailed bug reports. Don't just
post to one of the newsgroups and think, I will see it...

If you feel like not working enough, you could just pick up one of the
mentioned bugs and find a bug-fix for it. Many bug reports are very
detailed.

I included the mail address of the people who reported the bugs, if it
seems necessary to me.

(* I have fixed "lprm". Please test, if it really works!!)

(* "telnet" always stripps of the highest bit. It should be 8-bit...
  Just try "telnet localhost". This bug is fixed as of version 0.03.)

* After a while of running the machine, the rpc.portmapper begins to fork
  constantly. This only seems to happen when I have the rpc.ugidd also
  running.
  I have run the portmapper in debug mode and when it does go wild I get
  a pile of debug saying that it is about to enter the main switch in the
  portmapper source code.
  The problem does not happen on the two machines that I have at home networked.
  This only happens at work on a network with DEC 5000/240/260 and PC's etc.

  Joe Ceklosky <jceklosk@grctechs.grci.com> 

  (Could someone please mail me, where I can find the newest version of
  rpc.ugidd? I want to include it into my net source collection on funet.fi)

(* This bug should be fixed with the newest alpha kernel. The problem might
  have been, that the broadcast flag wasn't set for the loopback device.

  "rusers" segmentation faults, if invoked without any arguments (host names).
  (broadcast to all hosts on the local network)
  rusers with a list of hostnames works fine.)

(* See above, this bug should also be fixed with the newest alpha kernel.
  rpc.rusersd and rpc.rwalld cannot be run from inetd, but must be started
  from the /etc/rc* scripts as constantly running daemons.
  Inetd will start up dozen of rpc.rusersd processes and eventually terminates
  them, saying "rusersd/rpc/udp server failing (looping), service terminated".)

* rpc.rwalld prints two messages instead of one. (fixed??)

* ftp to a slow site and do "ls -ltr" a lot of times. ftp will get stuck and
  there will be a "CLOSE" connection, that will not vanish for days.
  (I am not sure, if this is still true for newer kernels.)

* Do "telnet localhost chargen" and then hit Ctrl-C. The process will hang.
  You will have to go to another console and kill it with a HANGUP signal.
  There have been some Linux specific changes to signal handling, maybe
  we should convert back?

* "rsh" is known to have a problem: A stopped rsh command will block the
  port, so that other rsh commands will hang. Alan Cox has promised to fix
  the kernel for this. This should be gone with a very new kernel. (??)

* I have changed "timed" to compile for Linux, but I cannot test it.
  Could someone tell me, if it is working?
  It is not working, but someone is working on it...

* Can someone make a real good sample /etc/printcap ????
  I need more examples, like printing via network.

* There is an option "-I" for rpcgen to make programs usable for "inetd".
  Then we maybe get a running "rusersd" ??  NO!!!
  The problem is maybe: All programs that use rpc broadcasts cannot be used
  together with the inetd.   (???)

* Note, this problem seems to be with a GNU version of 'last'. I have never
  seen that version and is probably not a problem with rlogind/telnetd.

  Subject: Problem with rlogind/telnetd and wtmp entries
  I have noticed the following problem with the wtmp entries written by
  the rlogind/telnetd provided in the net package.  An excerpt from the
  output of last will serve better than an attempted explanation of the
  problem, so here goes.  Notice how the logout time on ttyp1 is the
  same as the login time for the next session on that tty, and that
  quinlan has been logged out since sometime yesterday but that last
  doesn't seem to know this yet.
  [~]$ last
  quinlan  ttyp1        ftp.yggdrasil.co Wed Jun 22 16:28   still logged in
  imurdock ttyS1                         Wed Jun 22 12:27 - 12:34  (00:07)
  imurdock ttyS1                         Wed Jun 22 12:23 - 12:24  (00:01)
  imurdock ttyS1                         Wed Jun 22 12:22 - 12:22  (00:00)
  imurdock ttyS1                         Wed Jun 22 12:20 - 12:21  (00:01)
  imurdock ttyS1                         Wed Jun 22 12:17 - 12:18  (00:01)
  quinlan  ttyp1        ftp.yggdrasil.co Wed Jun 22 00:58 - 16:28  (15:29)
  [...]
  Using the `--complain' option of GNU last, the following information
  is provided about the wtmp entries:
  [~]$ last --complain
  quinlan  ttyp1        ftp.yggdrasil.co Wed Jun 22 16:28   still logged in
  [...]
  problem: logout time for next entry assumed to be last login time on tty
  quinlan  ttyp1        ftp.yggdrasil.co Wed Jun 22 00:58 - 16:28  (15:29)
  [...]
  I find it hard to believe that rlogind/telnetd is not writing a logout
  entry to wtmp at all.  In any case, somehow it doesn't like the way
  they do it.  Any ideas?

* I use rpcgen from rpc-0.9. I should upgrade to the newer available version.
  But I have heard rumors, that they don't work ok for all source code.
  If newer versions work ok, I should delete source code for rpcgen and
  rpcinfo from my source collection.
  I couldn't find newer versions with archie. Where are they?

* Vesa Ruokonen has suggested to use ncurses instead of bsd curses for talk.
  I will do that after the next version of ncurses is released.

Florian  La Roche   flla@stud.uni-sb.de
