We are current running QNX 6.4.1 with the ns83815 drivers

Any suggestions on how to attack this problem would be greatly appreciated.

We have multiple ways of causing the TCPIP subsystem to fail, these are but 2 examples:

1) If we directly connect a laptop to the Ethernet card on our pc104 stack (with the appropriate crossover) and 
establish a connection to our data acquisition application and then put the laptop to "sleep" mode, we seem to 
completely lose **ALL**  TCP/IP functionality on the QNX box -- which will only be restored with a complete reboot.  

2) connect QNX to operate with a wireless router on the lan side -- log a laptop into the router and then setup an 
application to talk with QNX (via our DAS app, telnet, ftp, etc) Move the laptop to a fringe location and wait for the 
wireless session to break.  Once the wireless session breaks with QNX, the TCPIP subsystem then random breaks 
approximently 20% of the time in the QNX box -- again requiring a complete reboot.  

Is there any way to detect if the TCP drivers have failed and perhaps reboot these alone?  We are completely stumped by 
this.

Thanks

James Kelley