multi-threaded exception handling mechanism

[Luke Dalessandro]

I apologize if this isn't an appropriate place for this question.

I have encountered performance behavior that leads me to believe that a lock 
exists somewhere in the multi-threaded exception mechanism implementation.

Can someone give me a pointer on where to look in the gcc source to confirm 
this suspicion? I'm happy to deal with the issue on my own, but I don't know 
where to start.

Thanks,
Luke