mmm@cup.portal.com (Mark Robert Thorson) (07/18/90)
In reading the Tech Ref for models 50 and 60, I notice the following warning on page 2-22: "WARNING: If both an adapter and a portion of the system board are selected for setup at the same time, bus contention will result; no useful programming can take place and damage to the hardware can occur." This refers to the fact that the I/O space addresses used by the POS (such as the adapter ID) are shared by all system board and expansion bus peripherals. If you enable more than one device (using bits in ports 94h and 96h), and if you then try to read one of the shared I/O addresses, you can get a buffer fight. Does anyone know how real the possibility of burning out a buffer is? I imagine that the probability would be low, because the fight would only last for one bus cycle. Would even a tight loop be able to exercise the problem to the point of causing a failure? If so, this would seem like an incredible blunder on IBM's part. Some kid could write a virus which would cause actual hardware damage. Do the models 70 and 80 have the same problem?