Method and apparatus for a configurable packet routing, buffering and scheduling scheme to optimize throughput with deadlock prevention in SRIO-to-PCIe bridges
View Patent ↗A method and apparatus for a configurable packet routing, buffering and scheduling scheme to optimize throughput with deadlock prevention in SRIO-to-PCIe Bridges have been described. A single level enqueue method with dynamic buffering and dequeuing based on packet re-ordering is disclosed. Single level packet routing and scheduling to meet SRIO and PCIe rules to enqueue packets based on FType/TType is disclosed. Backpressure based on ingress watermarks for different packet types is disclosed. Use of a circular-reorder queue (CRQ) for both ingress and egress allows packet reordering and packet passing.
1. A method for bridging between a Serial RapidIO® (SRIO) port and a PCI Express® (PCIe) port the method comprising:
checking for a backpressure for packet types denoted Messaging Response (MR), Maintenance Write Response (MWR), Doorbell Maintenance (DB), Messaging (M), Posted (P), Completion (CPL), Non-Posted (NP), and RFDID where RFDID is: R: Reserved Ftype 0 , 1 , 3 , 4 , 12 , 14 ; F: Flow control (Ftype 7 ); D: Data Streaming Ftype 9 ; ID: Implementation Defined (Ftype 15 ) on said SRIO port;
when there is said backpressure for packet types then passing packet types to said PCIe port based on predefined rules during dequeue operations in said SRIO port,
wherein said dequeue operations uses a dequeue block directly coupled to a circular-reorder queue (CRQ), said CRQ implemented in hardware and having a last-free-location pointer to an open location in said CRQ, a next-to-send pointer in said CRQ, a first-free-location pointer in said CRQ, and a last-packet-enqueued pointer in said CRQ, and wherein said CRQ has an acknowledgement identification for each of said received packet types; and
when there is no said backpressure for packet types then dequeuing packets strictly in a time ordered sequence in said SRIO port wherein said dequeue packets uses a dequeue block directly coupled to a circular-reorder queue (CRQ), said CRQ implemented in hardware and having a last-free-location pointer to an open location in said CRQ, a next-to-send pointer in said CRQ, a first-free-location pointer in said CRQ, and a last-packet-enqueued pointer in said CRQ.
2. An apparatus for bridging between a Serial RapidIO® (SRIO) port and a PCIExpress® (PCIe) port comprising:
means for checking for a backpressure for packet types denoted Messaging Response (MR), Maintenance Write Response (MWR), Doorbell Maintenance (DB), Messaging (M), Posted (P), Completion (CPL), Non-Posted (NP), and RFDID where RFDID is: R: Reserved Ftype 0 , 1 , 3 , 4 , 12 , 14 ; F: Flow control (Ftype 7 ); D: Data Streaming Ftype 9 ; ID: Implementation Defined (Ftype 15 ) on said SRIO port;
means for, when there is said backpressure for packet types, passing packet types to said PCIe port based on predefined rules during dequeue operations in said SRIO port, wherein said dequeue operations uses a dequeue block directly coupled to a circular-reorder queue (CRQ), said CRQ implemented in hardware and having a last-free-location pointer to an open location in said CRQ, a next-to-send pointer in said CRQ, a first-free-location pointer in said CRQ, and a last-packet-enqueued pointer in said CRQ, and wherein said CRQ has an acknowledgement identification for each of said received packet types, and
means for, when there is no said backpressure for packet types, dequeuing packets strictly in a time ordered sequence in said SRIO port wherein said dequeue packets uses a dequeue block directly coupled to a circular-reorder queue (CRQ), said CRQ implemented in hardware and having a last-free-location pointer to an open location in said CRQ, a next-to-send pointer in said CRQ, a first-free-location pointer in said CRQ, and a last-packet-enqueued pointer in said CRQ.
3. The apparatus of claim 2 further comprising:
means for buffering dynamically said SRIO Ftype packets; and
means for buffering dynamically said SRIO Ttype packets.
4. The apparatus of claim 3 wherein said means for buffering dynamically comprises means for dynamic buffer allocation using a virtual output queue (VOQ) structure.
5. The apparatus of claim 4 further comprising:
means for generating a backpressure towards a link-partner based on a plurality of programmable ingress watermarks, wherein said plurality of programmable ingress watermarks are programmable for different packet types.