MULTICAST ADDRESS LEARNING IN AN INPUT/OUTPUT ADAPTER OF A NETWORK PROCESSOR

• US 20120236857A1
• Filed: 05/25/2012
• Published: 09/20/2012
• Est. Priority Date: 05/18/2010
• Status: Active Grant

• Current Assignee: Intel Corporation
• Original Assignee: LSI Corporation (Broadcom, Inc.)
• Inventors: Manzella, Joseph A., Vora, Nilesh S., Peachey, Ritchie J.
• Granted Patent: US 8,705,531 B2
• Time in Patent Office: Days
• US Class Current: 370/390
• CPC Class Codes: H04L 49/109 Integrated on microchip, e.g. switch-on-chip
• CPC Class Codes: H04L 49/201 Multicast operation; Broadcast operation

Abstract

An apparatus for supporting multicast address learning in a network processor includes a task parameter decoder receiving the packet and determining parameters of the packet, a plurality of unlearned address counters recording a number of the packets to be multicast, a correlator determining destination addresses of the packet, a multicast replicator replicating the packet multiple times, and a task generator generating a generated packet with a bitmap and sending out the generated packet to an I/O adaptor over a task ring interface of the network processor. If the packet is unlearned, the unlearned address counter is then incremented and the unlearned packet is sent back to a special queue in the I/O adaptor over the task ring interface with an index to the bitmap for replicating by the multicast replicator, after replicating, the unlearned address counter is decremented. Methods for replicating unlearned/learned multicast packets within a network processor are included.

Claims

1. An apparatus for supporting multicast address learning for a packet to be sent to multiple destinations coupled to a network processor, the apparatus comprising:

a task parameter decoder receiving the packet from a task ring interface of the network processor through a task start interface and determining at least one parameter of the packet including determining whether the packet corresponds to a multicast packet or a unicast packet;

a plurality of unlearned address counters recording a number of the packets that is in a flow to be multicast and determining whether the packet is being sent out;

a correlator determining destination addresses of the packet based on an output of a parsing and bridging logic for each packet to be generated and transmitted by the network processor;

a multicast replicator receiving the destination addresses determined by the correlator and replicating the packet multiple times based on a number of the destination addresses corresponding to the packet; and

a task generator generating a generated packet with a bitmap and sending out the generated packet to the I/O adaptor over the task ring interface,

wherein the bitmap is analyzed in the I/O adaptor and multiple packets are created for transmitting to at least one processing module of the network processor for further processing based on the destination addresses determined by the correlator,

wherein, if the task parameter decoder determines the packet is an unlearned packet, the unlearned address counter is incremented and the unlearned packet is sent back to a dedicated queue in the task start interface of the I/O adaptor over the task ring interface with an index to the bitmap for replicating by the multicast replicator, thereby reducing bandwidth for the replicating by replicating only header data of each multicast packet.

2. The apparatus of claim 1, wherein the unlearned packet is one of an unlearned multicast packet and an unlearned unicast packet.

3. The apparatus of claim 1, wherein, if the task parameter decoder determines that the packet is a learned multicast packet, the learned multicast packet is sent back to the dedicated queue in the task start interface of the I/O adaptor over the task ring interface with information on the bitmap, such that the I/O adaptor replicates the packet using the multicast replicator, which results in a great reduction in bandwidth since the multicast replication uses the pointers for the majority of data and only replicates the header data of the packet.

4. The apparatus of claim 1, wherein, if the task parameter decoder determines that the packet is a learned unicast packet, then the learned unicast packet is sent to the destination of the packet.

5. The apparatus of claim 1, further comprising at least one virtual flow (Vflow) corresponding to a processing path for the packet between the at least one processing module of the network processor.

6. The apparatus of claim 5, wherein the at least one Vflow is determined by the multicast replicator corresponding to the packet from a Vflow table based on Vflow identifications (IDs) associated with the packet, and the multicast replicator provides the Vflow IDs to the correlator.

7. The apparatus of claim 1, wherein the task parameter decoder provides the parameters of the packet to the multicast replicator, the parameters include a Vflow ID corresponding to the packet, a source address of the packet, at least one destination address of the packet, whether each task corresponds to the unicast packet or the multicast packet, and a size of the packet.

8. The apparatus of claim 7, wherein the task parameter decoder performs a hash operation on one or both of the source address of the packet and the destination address of the packet.

9. The apparatus of claim 8, wherein the task parameter decoder determines whether the packet corresponds to a multicast or unicast packet through checking an indicator of a bridging decision based on the destination addresses of the packet.

10. The apparatus of claim 1, further comprising a packet learning module tracking a number of the packets being processed by the network processor.

11. The apparatus of claim 10, wherein the packet learning module maintains at least one packet destination address corresponding to the packet.

12. The apparatus of claim 11, wherein the at least one packet destination address about the packet is a media access control (MAC) address, and the packet learning module tracks the MAC address of the packet received by the network processor.

13. The apparatus of claim 1, wherein the unlearned address counter calculates a hash of a source/destination address.

14. The apparatus of claim 13, wherein the unlearned address counter is indexed by an address hash and ensures ordering.

15. The apparatus of claim 14, wherein the unlearned address counter is selected by a combination of a physical input port and the address hash of the source/destination address.

16. The apparatus of claim 15, wherein the unlearned address counter is compared to thresholds to keep a multicast bandwidth to a minimum, if the unlearned address counter exceeds the threshold, the packet is dropped before sending to the task generator.

17. The apparatus of claim 1, wherein the output of the parsing and bridge logic sent to the correlator includes a 24 bit bitmap containing the destinations of the packet, a 8 bit hash of a source/destination address, and a 1 bit flag indicating a destination address is learned or unlearned on the destination of the packet, each bit of the 24 bit bitmap responds to one destination, if multiple bits are set, the packet is multicast, if one bit is set, the packet is unicast.

18. A multicast replication logic for supporting multicast address learning for a packet to be sent to multiple destinations coupled to a network processor, the multicast replication logic comprising:

a task parameter decoder receiving the packet from a task ring interface of the network processor through a task start interface and determining at least one parameter of the packet including determining whether the packet corresponds to a multicast packet or a unicast packet;

a plurality of unlearned address counters recording a number of the packets that is in a flow to be multicast and determining whether the packet is being sent out;

a correlator determining destination addresses of the packet based on an output of a parsing and bridging logic for each packet to be generated and transmitted by the network processor;

a multicast replicator receiving the destination addresses determined by the correlator and replicating the packet multiple times based on a number of the destination addresses corresponding to the packet; and

a task generator generating a generated packet with a bitmap and sending out the generated packet to the I/O adaptor over the task ring interface,

wherein the bitmap is analyzed in the I/O adaptor and multiple packets are created for transmitting to at least one processing module of the network processor for further processing based on the destination addresses determined by the correlator,

wherein, if the task parameter decoder determines the packet is an unlearned packet, the unlearned address counter is incremented and the unlearned packet is sent back to a dedicated queue in the task start interface of the I/O adaptor over the task ring interface with an index to the bitmap for replicating by the multicast replicator, thereby reducing bandwidth for the replicating by replicating only header data of each multicast packet.

19. The apparatus of claim 18, wherein the unlearned packet is one of an unlearned multicast packet and an unlearned unicast packet.

20. The apparatus of claim 18, wherein, if the task parameter decoder determines that the packet is a learned multicast packet, the learned multicast packet is sent back to the dedicated queue in the task start interface of the I/O adaptor over the task ring interface with information on the bitmap, such that the I/O adaptor replicates the packet using the multicast replicator, which results in a great reduction in bandwidth since the multicast replication uses the pointers for the majority of data and only replicates the header data of the packet.

21. The apparatus of claim 18, wherein, if the task parameter decoder determines that the packet is a learned unicast packet, then the learned unicast packet is sent to the destination of the packet.

22. A multicast network processor system comprising:

a task ring interface including at least one input/output (I/O) interface, at least one microprocessor and at least one hardware accelerator; and

a switch coupled to the task ring interface and a shared memory,

wherein the task ring interface is coupled through the switch to the shared memory,

wherein the at least one I/O adaptor includes a multicast replication logic generating at least one multicast launch packet within the multicast network processor system,

the multicast replication logic comprising;

a task parameter decoder receiving the packet from a task ring interface of the network processor through a task start interface and determining at least one parameter of the packet including determining whether the packet corresponds to a multicast packet or a unicast packet;

a plurality of unlearned address counters recording a number of the packets that is in a flow to be multicast and determining whether the packet is being sent out;

a correlator determining destination addresses of the packet based on an output of a parsing and bridging logic for each packet to be generated and transmitted by the network processor;

a multicast replicator receiving the destination addresses determined by the correlator and replicating the packet multiple times based on a number of the destination addresses corresponding to the packet; and

a task generator generating a generated packet with a bitmap and sending out the generated packet to the I/O adaptor over the task ring interface,

wherein the bitmap is analyzed in the I/O adaptor and multiple packets are created for transmitting to at least one processing module of the network processor for further processing based on the destination addresses determined by the correlator,

wherein, if the task parameter decoder determines the packet is an unlearned packet, the unlearned address counter is incremented and the unlearned packet is sent back to a dedicated queue in the task start interface of the I/O adaptor over the task ring interface with an index to the bitmap for replicating by the multicast replicator, thereby reducing bandwidth for the replicating by replicating only header data of each multicast packet.

23. The apparatus of claim 22, wherein the unlearned packet is one of an unlearned multicast packet and an unlearned unicast packet.

24. The apparatus of claim 22, wherein, if the task parameter decoder determines that the packet is a learned multicast packet, the learned multicast packet is sent back to the dedicated queue in the task start interface of the I/O adaptor over the task ring interface with information on the bitmap, such that the I/O adaptor replicates the packet using the multicast replicator, which results in a great reduction in bandwidth since the multicast replication uses the pointers for the majority of data and only replicates the header data of the packet.

25. The apparatus of claim 22, wherein, if the task parameter decoder determines that the packet is a learned unicast packet, then the learned unicast packet is sent to the destination of the packet.

26. A method for replicating unlearned multicast packets within a network processor comprising:

receiving packet information on packet destinations from an I/O adaptor of a task ring interface and determining destination addresses by a correlator and sending the destination addresses to a multicast replicator;

incrementing an unlearned address counter;

sending the unlearned multicast packet back to a dedicated queue in the I/O adaptor over the task ring interface with a bitmap;

analyzing the bitmap in the I/O adaptor;

replicating a header of the unlearned multicast packet by the multicast replicator;

decrementing the unlearned address counter after replicating;

generating a generated packet through a task generator; and

sending the generated packet back to the task ring interface with an index to the bitmap and creating a multicast launch packet thereon to at least one processing module of the network processor for further processing.

27. The method of claim 26, further comprising:

determining parameters of the packet including a virtual flow (Vflow) identification (ID) corresponding to the packet, a source address of the packet, at least one destination address of the packet.

28. The method of claim 27, further comprising:

performing a hash operation on the source address and the destination addresses by a task parameter decoder.

29. The method of claim 28, further comprising:

comparing the unlearned address counter to thresholds to keep a multicast bandwidth to a minimum, if the unlearned address counter exceeds the threshold, the packet is dropped before sending to the task generator.

30. The method of claim 26, wherein the receiving packet information on packet destinations includes:

receiving the packet information on the packet destinations by a task start interface;

extracting the bitmap, an address hash and unlearned packet flag in the dedicated queue of the task start interface;

sending the packet out to each location in the bitmap; and

determining an unlearned address counter associated with the packet.

31. A method for replicating learned multicast packets within a network processor comprising:

receiving packet information on packet destinations from an I/O adaptor of a task ring interface and determining destination addresses by a correlator and sending the destination addresses to a multicast replicator;

sending the learned multicast packet back to a special queue in the I/O adaptor over the task ring interface with a bitmap;

analyzing the bitmap in the I/O adaptor;

replicating a header the learned multicast packet by the multicast replicator;

generating a generated packet through a task generator; and

sending the generated packet back to the task ring interface with an index to the bitmap and creating a multicast launch packet thereon to at least one processing module of the network processor for further processing.

32. The method of claim 31, wherein the receiving packet information on packet destinations includes:

receiving the packet information on the packet destinations by a task start interface;

extracting the bitmap, an address hash and unlearned packet flag in the special queue of the task start interface;

sending the packet out to each location in the bitmap; and

determining an unlearned address counter associated with the packet.