Abstract: A method and apparatus for arbitrating between command streams. The method unblocks high priority commands which are blocked and then selects any remaining high priority commands. Normal priority commands are selected after the high priority commands. A memory controller described includes a command queue block having a plurality of command queues, each being coupled to receive a different type of command. The memory controller also includes arbitration logic which, among other things, selects high priority read commands before high priority write commands. Memory interface logic generates memory accesses performing commands selected by the arbitration logic.

Abstract: A method and apparatus for ordering memory access commands. A command ordering circuit which is described includes a plurality of command slots which receive memory access commands. A page register stores a value indicating a last page accessed by a prior memory access command. Comparators compare the value in the page register to values stored in the command slots, and an arbiter receives outputs from the comparators and selects a command from one of the slots. According to the method described, memory accesses are reordered depending on the portion of memory accessed. A first memory access command requesting access to a first portion of memory is issued. Additional memory access commands also referencing the first portion of memory are issued until a count is reached. After the count is reached, a second memory access command which references a second portion of memory is issued.

Abstract: A bridge circuit adapted to be associated with a PCI and a secondary bus circuits which bridge circuit includes circuitry for storing an indication that a particular PCI bus master has attempted an access of the secondary bus and has been forced to retry that access, circuitry for masking any retry until the bus is again available, and circuitry for providing an interval during which a retrying PCI bus master is guaranteed access to the secondary bus in favor of a bus master on the secondary bus after the bus is relinquished so that a sequence of retry operations causing a loss of bandwidth on the PCI bus is not generated.

Abstract: Error correction circuitry attempts to detect and correct on the fly erroneous words within random access memory (RAM) within a computer system. RAM errors are scrubbed or corrected back in the memory without delaying the memory access cycle. Rather, the address of the section or row of RAM that contains the correctable error is latched for later used by an interrupt-driven firmware memory-error scrub routine. This routine reads and rewrites each word within the indicated memory section--the erroneous word is read, corrected on-the-fly as it is read, and then rewritten back into memory correctly. If the size of the memory section exceeds a predetermined threshold, then the process of reading and re-writing that section is divided into smaller sub-processes that are distributed in time using a delayed interrupt mechanism. Duration of each memory scrubbing subprocess is kept short enough that the response time of the computer system is not impaired with the housekeeping task of scrubbing RAM memory errors.

Abstract: A method for converting signals from one arbitration and management protocol to another. The conversion is performed by at least three state machines. The conversion circuit converts a set of signals on the first bus to a bus request signal on the second bus. The conversion circuit also converts a signal from the second bus and a set of signals on the first bus to bus grant and memory acknowledge signals on the first bus.

Abstract: A bridge coupling a primary bus to two secondary buses. The bridge contains three interfaces, one for the primary bus and the other two for the two secondary buses. Control circuitry is included within the bridge to support the execution of a transaction initiated by a bus master upstream of the bridge to a target downstream of the bridge. The bridge also supports the execution of a transaction initiated by a bus master coupled to either one of the secondary buses to a target upstream of the bridge.

Abstract: A method and apparatus for regulating the deferral of a transaction issued on a bus by a processor in a computer system is disclosed. A bus transaction recorder coupled to the bus processes encoded signals from the transaction issued on the bus. A line coupled to the bus sends an indication signal when a pending transaction request is issued on the bus. A CPU latency timer times the current transaction on the bus when a new pending transaction is waiting on the bus. The CPU latency timer outputs an expiration signal when the transaction takes more than a predetermined amount of time to complete. A transaction processor unit is coupled to the bus transaction recorder, the line, and the CPU latency timer.

Abstract: A computer system including amongst its components a host bus coupled to a processor, an intermediate (PCI) bus, an expansion (ISA or EISA) bus, a host bridge coupled between the host and intermediate busses, and an expansion bridge coupled between the intermediate and expansion busses, is disclosed. The host bridge incorporates data buffer management circuitry which examines a write request presented to the host bridge to determine whether the write request is to a device not coupled to the expansion bus. If the write request is to a device not coupled to the expansion bridge, the buffer management allows the write buffer to accept write data associated within the write request. If not, the buffer management circuitry prevents the write buffer from accepting the write data associated with the write request. The data buffer management circuitry may be configured to determine specifically whether the write request is to a graphics frame buffer.

Abstract: A register address space is defined with a capacity large enough to accommodate substantial growth in the number of required registers. Unused register locations are reserved for future use. Access requests directed to reserved addresses are redirected to a physical register containing the same stored value that would be returned if a physical register were associated with the reserved address to which the access was originally directed. The physical register is separate from any central processing unit.

Abstract: A bridge circuit providing for efficient data transfer between a first bus and a second bus in a computer system. The bridge circuit receives an address indicating a memory location storing a data segment requested to be transferred from the first bus to the second bus. Fetch circuitry fetches the requested data from the first bus and prefetches one or more additional data segments stored in memory locations sequentially following the memory location storing the requested data. The prefetched data segments are stored in a buffer for immediate access by subsequent data transfer requests. Supply circuitry transfers each data segment from the buffer to the second bus in response to receiving an address corresponding to the particular data segment on the address input circuitry.

Abstract: A personal computer (PC) card insertion method and apparatus uses a subset of connector ground terminals and pins, located at either end of the connector, for detecting the onset of a card insertion. The host PC card slot connector has pull-up resistors for keeping the subset of ground terminals at a high logic level (V.sub.CC). Also, the subset of pins are made longer than the signal pins so that when an insertion of a PC card begins, the grounding of one or more of the subset of pins indicates that a PC card insertion has begun, allowing the host system to take the necessary precautions to ensure an orderly acceptance of the card without any undesirable system affects that might otherwise result. Also, a logic network for using the subset of connector terminals as additional grounding connections is provided upon completion of the insertion.

Abstract: A latching mechanism captures an address transmitted on a multiplexed address/data bus and preserves it for the full bus cycle. A transparent latch with a multiplexed feedback path allows the address to be quickly captured and decoded. An additional multiplexer and latch cooperate with the first mentioned latch to keep the address stable for a sufficient time to allow latching by slower memory elements. Additional elements are provided to automatically increment the address for multiple data burst operation.