OK. I've finally been able to see this bug in action. It turns out to be 
a Comanche bug. Apparently, there are times when Comanche does not know 
the remote address. This causes it to send nil, rather than a byte array 
to NetNameResolver. This causes a lock on the semaphore. WEIRD!

Peace and Luck!

Je77

On Mon, Aug 09, 2004 at 01:41:24PM -0400, Jochen F. Rick wrote:
> I think I found a bug in Squeak's networking and I was wondering if 
> somebody could help me out. It may be causing the Swikis to have problems.
> 
> The bug appears to be in NetNameResolver. NetNameResolver uses a 
> semaphore to keep it from resolving more than one address at a time. This 
> seems OK, except that sometimes that semaphore gets locked down. My guess 
> is that there was an error (i.e. primitive failed, etc.) and that caused 
> the semaphore to remain locked. Unfortunately, I can't replicate this 
> happening as it happens so rarely. So, I'm unable to track it down.
> 
> Could someone who knows the code see if there is a way to avoid this from 
> happening. For instance, could we put an ifError: thing in there?
> 
> BTW, so far, I've only observed this bug happening on solaris and 
> linux. Is there a chance of a VM bug?
> 
> Peace and Luck!
> 
> Je77
> 
> -- 
> Jochen "Je77" Rick                               jochen.rick at cc.gatech.edu
> PhD Student                                           http://www.je77.com/
> Georgia Tech College of Computing                       work: 404-385-1105

-- 
Jochen "Je77" Rick                               jochen.rick at cc.gatech.edu
PhD Student                                           http://www.je77.com/
Georgia Tech College of Computing                       work: 404-385-1105