| 摘要 |
A technique for managing bidirectional data flow between a first data bus, such as in a local area network, having a fixed data flow rate once access has been obtained, and a memory or storage device connected to a second data bus having variable latency. Inbound data from the first data bus are temporarily stored in a receive buffer memory and later forwarded to the storage device. Outbound data bursts are retrieved from the same storage device, temporarily stored in a transmit buffer memory, and later transmitted onto the first data bus. The invention controls the steps of forwarding inbound data to the storage device and retrieving outbound data from the storage device, in such a manner as to make efficient use of the second data bus and to minimize the possibility of overflowing the receive buffer memory during a receive operation, or underflowing the transmit buffer memory during a transmit operation. When transmitting is in progress, priority is given to supplying outbound data to the transmit buffer memory, with any inbound data being processed only if the transmit buffer memory is sufficiently full. When receiving is in progress, priority is given to emptying inbound data from the receive buffer memory, with any outbound data being processed only if the receive buffer memory is sufficiently empty. When neither transmitting nor receiving is in progress, priority is giving to emptying the receive buffer memory, then filling the transmit buffer memory.
|
