Sean, can you give us a bit more detail on removing a thread?  I don't 
think I'm real clear on who is being protected from what by the 
quiesce() calls, and just how drivers should "handle their own 
quiescing".  I'm testing code that looks a lot like qnet's 
qdpp_handle_thread() loop, except with your DVF_QUIESCESELF flag.  I can 
/usually /unload the module without any apparent problems, but 
occasionally unloading it results in a SEGV in qnet's 
en_iopkt_tx_pkt().  Causing another module toSEGV suggests that I should 
be quiescing everybody but myself during some critical section, but so 
far the details are eluding me.

Murf