High-speed CLD-based TCP assembly offload

1. A method offloading data intensive tasks from a processor, comprising:

receiving at a configurable logic device (CLD) a network packet, the CLD being coupled to a memory having a packet table including a plurality packet data structures for packet assembly;

parsing the network packet to determine that the packet is a TCP segment;

searching the packet table to locate an associated partially assembled packet data structure;

when an associated partially assembled packet data structure is found, inserting the network packet into the associated partially assembled packet data structure;

when an associated partially assembled packet data structure is not found, adding the network packet to an unused packet data structure within the packet table;

determining whether the partially assembled packet data structure contains every segment produced from an original TCP packet;

when every segment is contained in the data structure:

assembling a fully assembled TCP packet from the data in the partially assembled packet data structure; and

transmitting the fully assembled TCP packet to a processor in the same computer system as the CLD; and

when every segment is not contained in the data structure:

determining whether a threshold number of the plurality of packet data structures are in use; and

when a threshold number are in use:

identifying the oldest partially assembled packet data structure;

transmitting without assembling each TCP segment packet contained in the oldest partially assembled packet data structure to the processor; and

clearing the oldest partially assembled packet data structure for subsequent reuse.

2. The method of claim 1 , further comprising, at a regular interval:

inspecting each of the partially assembled packet data structures;

determining that a particular partially assembled packet data structure is older than a predetermined threshold;

transmitting without assembling each TCP segment packet contained in the particular partially assembled packet data structure to the processor; and

clearing the particular partially assembled packet data structure for subsequent reuse.

3. The method of claim 1 , wherein the memory is not shared with the processor.

4. The method of claim 1 , wherein a total number of packet data structures within the memory is controlled by the processor.

5. The method of claim 1 , wherein the threshold number is one less than a total number of packet data structures included within the memory.

6. The method of claim 1 , further comprising applying a bypass filter to the packet to determine whether or not to proceed to the searching step.

7. A tangible, non-transitory computer-readable media comprising a configuration file that when loaded by a configurable logic device (CLD) configures the CLD to:

receive at a CLD a network packet, the CLD being coupled to a memory having a packet table including a plurality packet data structures for packet assembly;

parse the network packet to determine that the packet is a TCP segment;

search the packet table to locate an associated partially assembled packet data structure;

when an associated partially assembled packet data structure is found, insert the network packet into the associated partially assembled packet data structure;

when an associated partially assembled packet data structure is not found, add the network packet to an unused packet data structure within the packet table;

determine whether the partially assembled packet data structure contains every segment produced from an original TCP packet;

when every segment is contained in the data structure:

assemble a fully assembled TCP packet from the data in the partially assembled packet data structure; and

transmit the fully assembled TCP packet to a processor in the same computer system as the CLD; and

when every segment is not contained in the data structure:

determine whether a threshold number of the plurality of packet data structures are in use; and

when a threshold number are in use:

identify the oldest partially assembled packet data structure;

transmit without assembling each TCP segment packet contained in the oldest partially assembled packet data structure to the processor; and

clear the oldest partially assembled packet data structure for subsequent reuse.

8. The media of claim 7 , wherein the configuration file further configures the CLD to, at a regular interval:

inspect each of the partially assembled packet data structures;

determine that a particular partially assembled packet data structure is older than a predetermined threshold;

transmit without assembling each TCP segment packet contained in the particular partially assembled packet data structure to the processor; and

clear the particular partially assembled packet data structure for subsequent reuse.

9. The method of claim 7 , wherein the memory is not shared with the processor.

10. The method of claim 7 , wherein a total number of packet data structures within the memory is controlled by the processor.

11. The method of claim 7 , wherein the threshold number is one less than a total number of packet data structures included within the memory.

12. The method of claim 7 , wherein the configuration file further configures the CLD to apply a bypass filter to the packet to determine whether or not to search the packet table.

13. A computing system, comprising:

a configurable logic device (CLD) configured to:

receive at a CLD a network packet, the CLD being coupled to a memory having a packet table including a plurality packet data structures for packet assembly;

parse the network packet to determine that the packet is a TCP segment;

search the packet table to locate an associated partially assembled packet data structure;

when an associated partially assembled packet data structure is found, insert the network packet into the associated partially assembled packet data structure;

when an associated partially assembled packet data structure is not found, add the network packet to an unused packet data structure within the packet table;

determine whether the partially assembled packet data structure contains every segment produced from an original TCP packet;

when every segment is contained in the data structure:

assemble a fully assembled TCP packet from the data in the partially assembled packet data structure; and

transmit the fully assembled TCP packet to a processor in the same computer system as the CLD; and

when every segment is not contained in the data structure:

determine whether a threshold number of the plurality of packet data structures are in use; and

when a threshold number are in use:

identify the oldest partially assembled packet data structure;

transmit without assembling each TCP segment packet contained in the oldest partially assembled packet data structure to the processor; and

clear the oldest partially assembled packet data structure for subsequent reuse.

14. The computing system of claim 13 , wherein the CLD is further configured to, at a regular interval:

inspect each of the partially assembled packet data structures;

determine that a particular partially assembled packet data structure is older than a predetermined threshold;

transmit without assembling each TCP segment packet contained in the particular partially assembled packet data structure to the processor; and

clear the particular partially assembled packet data structure for subsequent reuse.

15. The method of claim 13 , wherein the memory is not shared with the processor.

16. The method of claim 13 , wherein a total number of packet data structures within the memory is controlled by the processor.

17. The method of claim 13 , wherein the threshold number is one less than a total number of packet data structures included within the memory.

18. The method of claim 13 , wherein the CLD is further configured to apply a bypass filter to the packet to determine whether or not to search the packet table.