Fwd: possible bug in httptunnel (htc)

[Philip Yarra]

OK, here http://www.nocrew.org/software/httptunnel.html lists 3.0.5 as stable, 
gnu.org http://directory.fsf.org/httptunnel.html lists 3.3 as stable as of 
2001, so I grabbed 3.3 and tried that... problem solved, it seems.

It might be worth updating the httptunnel page to mark 3.3 as stable?

Regards, Philip.

----------  Forwarded Message  ----------

Subject: possible bug in httptunnel (htc)
Date: Tue, 16 Aug 2005 01:06 pm
From: Philip Yarra <philip@utiba.com>
To: bug-httptunnel@gnu.org

Hi all, I have downloaded and had a play with httptunnel-3.0.5. I think I
 have hit a bug in the htc tool. My steps are:

1) start hts on remote machine
2) start htc on local machine, pointing to proxy server
3) start ssh application to use the htc tunnel
4) successful ssh login to remote machine
5) logout of remote machine
6) check that htc is still running (it is)
7) start ssh application to re-use the htc tunnel. It's here that things
 start going screwy: my local machine suddnly has 100% CPU utilisation, 100%
 RAM usage (1 GB physical RAM), and swap starts to get chewed up. I know it's
 rather unscientific, but I've attached a screengrab of the resource use -
 it's rather graphic. You can clearly see the sudden spikes in RAM and swap
 usage.

I'm having trouble proving it is htc causing this behaviour, since at this
point the machine is pretty well dead in the water, and my running top
process gets edged off the CPU. I am fairly sure it is htc, as if I manually
kill and re-start htc between steps 6 and 7, this problem does not occur.

When time allows I'll try to search for the cause. My suspicion is that
there's some runaway loop, possibly that's malloc'ing something (having seen
this sort of issue in my own code before). Does this sound possible?

Frustratingly, when I compile with debug enabled, this problem does not
 occur, so I'm having a tough time locating the problem. It doesn't seem to
 matter if I run with debugging on, so long as it was compiled with debug...
 I'm going to have a snoop around the #ifdef DEBUG_MODE blocks for possible
 issues.

In case you're interested, the remote machine is Mandrake 9.2 (kernel 2.4.22,
gcc version 3.3.1) and the local machine is RedHat WS 3 (kernel
2.4.21-32.ELsmp, gcc version 3.2.3).

Please let me know if I can assist with solving this issue. If I find it
 first I'll submit a patch.

Regards, Philip.

-------------------------------------------------------

-----------------
Utiba Pty Ltd 
This message has been scanned for viruses and
dangerous content by Utiba mail server and is 
believed to be clean.

ram_eater.png
Description: PNG image