Lets try this push again.. 2.5.1p2 bugs left.

[Gert Doering]

Hi,

On Sat, Feb 24, 2001 at 09:49:28PM +0100, Gert Doering wrote:
> > Hmm, It's working here with both openssh client and ssh 1.2.27
> > But I'm not compiling with skey support.
> 
> Tried w/o skey, but no difference (besides skey not working anymore):
> 
> debug: Doing challenge reponse authentication.
> debug: No challenge.
> debug: Doing password authentication.
> gert at greenie.muc.de's password:
> debug: Requesting pty.
> debug: Requesting shell.
> debug: Entering interactive session.
> Received disconnect from 193.149.48.161: Command terminated on signal 11.
> debug: Calling cleanup 0x9dbc(0x0)
> debug: Calling cleanup 0x19464(0x0)
> debug: Calling cleanup 0x1e898(0x0)
> 
> sending a single command ("ssh host id -l") works and gives the expected
> result.

Update:

 - running "sshd -d" from gdb works - no signal 11
 - running "sshd" from the command line (luid is set already) works
 - running "sshd -D" from the command line works
 - running "sshd -D" from /etc/inittab does *not* work (signal 11).

This makes debugging somewhat "interesting".

In syslog, I found the following message:

Feb 24 21:57:06 greenie sshd.x[11686]: Disconnecting: Command terminated on signal 11.

What does this mean?  "child shell", or "sshd" terminated?


Doing further testing:

 - "ssh host command" does NOT have the problem
 - "ssh host ksh" (my login shell) does NOT have the problem
 - "ssh -t host ksh" (shell with pty) does NOT have the problem
 - "ssh -T host" (login, no pty) does NOT have the problem
 - "ssh host" -> something crashes, reproducible


Ok, next:

First I had the following inittab entry, reliably crashing:

dm3:23:respawn:/u/softadm/openssh_cvs/sshd.x -p 24 -D -f /u/softadm/openssh_cvs/sshd_config 

For debugging, I changed this to

dm3:23:respawn:/u/softadm/openssh_cvs/sshd.x -p 24 -D -f /u/softadm/openssh_cvs/sshd_config 2>>/tmp/sshd.err >>/tmp/sshd.out

(with various numbers of "-d" arguments).  And, voila, *no more crashes*.

If I remove the ">>/tmp/sshd.out" part, the crash is back.

So what I think this boils down to is some hard-to-track error that occurs
only if *stdout* is not connected to anything upon sshd startup.

... and it doesn't even crash sshd but the child process?!


Now I have to give up and ask for some hints where to look in the
code...?

gert


-- 
USENET is *not* the non-clickable part of WWW!
                                                           //www.muc.de/~gert/
Gert Doering - Munich, Germany                             gert at greenie.muc.de
fax: +49-89-35655025                        gert.doering at physik.tu-muenchen.de