MVME162 RTEMS exit(0) actions/control

[Ken Fitch]

I've successfully (I think) built RTEMS 4.5.0-beta3 for a Solaris 
2.5.1 host and a MVME162 target (after some messing around with 
workarounds for problems in the build procedures).  

For historical reasons, the MVME162 board is configured
to use serial port 2 for the MBUG interface.  If I build the 
hello_world_c program, edit the S8 S-record to avoid immediate
execution, download the resulting test.exe using MBUG (162-Bug)'s "LO" 
command, and then run the program with a "G 100000", it does write out the
expected messages to serial port 1, but then continues to spew hex data
until ultimately the processor stops (apparently because the code gets
overwritten).

What is the expected behavior when the exit(0) call is made?  I expected
(probably without any basis) that the appropriate trap call to get me
back to the 162-Bug> prompt would happen.  I didn't expect to see data
coming out the serial port.  

I scanned the mailing list archives on the old OAR site, but didn't
find anything I thought was relevant.  

Is this indicative of a problem in the installation/build or with the
MVME162 board configuration, or with my lack of experience/understanding
with/or RTEMS?  If it's the latter, what should I be reading to explain
it all to me?

Info about the target system: 

162-Bug>cnfg
Board (PWA) Serial Number = "2039983     "
Board Identifier          = "PWB-MVME162-020A"
Artwork (PWA) Identifier  = "01-W3884B65B    "
MPU Clock Speed           = "2500"
...

Ken Fitch