[FreeNX-kNX] Weird crash of free nx session when browsing cups through the web remotely

[Jeffrey J. Kosowsky]

This is a weird one and has me stumped.

I am using freenx with printer sharing to log into my Fedora8 server
running freenx 0.7.1
The client runs WinXP with NoMachines NX Client for Windows 3.0.0-78.

When I use smbport=139 to enable printer sharing, I am able to print
fine from the command line (using lpr -h locahost:11001 -P
<myprinter>) and also am able to use the cups browser at
http://localhost:11001 to manage the printer and print test pages from
within the nx client (i.e. running the web browser from localhost =
Fedora nx server)

However, when I manually set up a parallel ssh tunnel from my Windoze
machine using -L 11001:localhost:11001 to browse the cups printer
interface from my Windows client, everything works fine until I try to
print a test page. The test page actually prints fine, but shortly
after the test page goes to the printer, the nx client crashes. The
session is still there but the client freezes.

Interestingly, if using my patches (see earlier messages), I set
smbport=cifs (445) and then set up a parallel ssh tunnel to manage the
cups printing (using -R 13001:<windozeclient>:139) along with another
tunnel (-L 11001:localhost:11001) to forward the cups web interface, I
am able to print test pages just fine from the cups web browser on my
Windows machine.

So, anybody have any ideas why the following occurs:
- Printing test pages from the cups web interface on localhost
  independent of whether tunnel is the standard nx/frenx smbfs tunnel
  (<DISPLAY+3000> -> 139) or my own manual tunnel (13001 -> 139).

- Printing test pages from cups web interface browsed remotely on my Windows
  machine (by forwarding 11001 -> 11001) works when printing over my
  own manual tunnel (13001 -> 139) but crashes the nx client when using the
  default smbfs tunnel (<DISPLAY+3000> -> 139).

In fact, I can narrow down the difference further but setting up two
artificial test cases with a common setup of forwarding -R
13001:<mymachine>:139 -L 11001:localhost:11001 and in both cases doing
the printing over port 13001 and the browsing remotely over 11001.
The only difference in the entire setup is then:
	- Case A: smbport=139  -- crashes after sending test page to
	  	   	  printer
	- Case B: smbport=445 -- no crash after sending test page to
	  	   	  printer 

I don't have any clue why I would be seeing such conflicting
behavior. In particular, why does printing a test page over ssh
forwarding of port 11001 kill the nx client session when smbport=139
but not when smbport=445?

I'm stumped...