Safety node in interconnect data buses

Abstract

In safety-critical computer systems, fault tolerance is an important design requirement. Data buses for on-chip interconnection in these processor-based systems are exposed to risk arising from faults in the interconnect itself or in any of the connected peripherals. To provide sufficient fault tolerance, a safety node is inserted between an upstream master section and a downstream slave section of an on-chip bus hierarchy or network. The safety node provides a programmable timeout monitor for detecting a timeout condition for a transaction. If timeout has occurred, the safety node transmits a dummy response back to the master, assumes the role of a master, and waits for the slave device to respond. Furthermore, the safety node rejects any subsequent requests by any of the masters on the upstream section by transmitting a dummy response to those subsequent requests, thus enabling these masters to avoid deadlock or stall.