Embedded transport acceleration architecture

• US 7,305,493 B2
• Filed: 11/27/2002
• Issued: 12/04/2007
• Est. Priority Date: 11/27/2002
• Status: Active Grant

• Current Assignee: Tahoe Research Limited (f/k/a Learndale Limited) (Vector Capital Corporation)
• Original Assignee: Intel Corporation
• Inventors: McAlpine, Gary L., Shah, Hemal V., Foong, Annie, Minturn, David B., Regnier, Greg J., Saletore, Vikram A.
• Primary Examiner(s): Pollack; Melvin H.
• Application Number:
  US10/305,738
• Publication Number:
  US 20040103225A1
• Time in Patent Office: 1,833 Days
• Field of Search: 709/200, 709/208, 709/229, 709/232, 709/233, 709/249, 709/250, 719/311, 710/52, 710/105, 710/118, 370/395.52
• US Class Current: 709/250
• CPC Class Codes: G06F 9/543 User-generated data transfer, e.g. clipboards, dynamic data exchange [DDE], object linking and embedding [OLE]
• CPC Class Codes: H04L 69/10 Streamlined, light-weight or high-speed protocols, e.g. express transfer protocol [XTP] or byte stream
• CPC Class Codes: H04L 69/16 Implementation or adaptation of Internet protocol [IP], of transmission control protocol [TCP] or of user datagram protocol [UDP]
• CPC Class Codes: H04L 69/161 Implementation details of TCP/IP or UDP/IP stack architecture; Specification of modified or new header fields
• CPC Class Codes: H04L 69/162 involving adaptations of sockets based mechanisms secure socket layer H04L63/168
• CPC Class Codes: H04L 69/163 In-band adaptation of TCP data exchange; In-band control procedures
• CPC Class Codes: H04L 69/32 Architecture of open systems interconnection [OSI] 7-layer type protocol stacks, e.g. the interfaces between the data link level and the physical level

Abstract

An apparatus and a system may include an adaptation module, a plurality of Direct Transport Interfaces (DTIs), a DTI accelerator, and a Transport Control Protocol/Internet Protocol (TCP/IP) accelerator. The adaptation module may provide a translated sockets call from an application program to one of the DTIs, where an included set of memory structures may couple the translated sockets call to the DTI accelerator, which may in turn couple the set of memory structures to the TCP/IP accelerator. An article may include data causing a machine to perform a method including: receiving an application program sockets call at the adaptation module, deriving a translated sockets call from the application program sockets call, receiving the translated sockets call at a DTI, coupling the translated sockets call to a DTI accelerator using a set of memory structures in the DTI, and coupling the set of memory structures to a TCP/IP accelerator.

Claims

1. An apparatus, comprising:

an adaptation module;

a plurality of direct transport interfaces (DTIs), wherein each one of the plurality of DTIs includes a set of memory structures, and wherein the adaptation module is capable of providing a translated sockets call associated with an application program to a selected one of the plurality of DTIs; and

a DTI accelerator, wherein the set of memory structures included in the selected one of the plurality of DTIs is capable of coupling the translated sockets call to the DTI accelerator, and wherein the DTI accelerator is capable of coupling the set of memory structures included in the selected one of the plurality of DTIs to a Transport Control Protocol/Internet Protocol (TCP/IP) accelerator.

2. The apparatus of claim 1, wherein the application program is capable of communicating directly with the TCP/IP accelerator using the translated sockets call.

3. The apparatus of claim 1, wherein the DTI accelerator and the TCP/IP accelerator are capable of associating a client connection request associated with a client with a Master DTI context, completing a connection sequence with the client, and associating the client with a child DTI.

4. The apparatus of claim 1, wherein the DTI accelerator and the TCP/IP accelerator are capable of associating a doorbell with a DTI and transferring a TCP/IP packet from an application buffer.

5. The apparatus of claim 4, further comprising:

a zero-copy send data path capable of transferring send data directly from the application buffer to the TCP/IP protocol accelerator; and

a one-copy send data path capable of transferring the send data through a DTI send buffer to the TCP/IP protocol accelerator.

6. The apparatus of claim 1, wherein the DTI accelerator and the TCP/IP accelerator are capable of associating a TCP/IP packet with a child DTI and transferring the TCP/IP packet to an application buffer.

7. The apparatus of claim 6, further comprising:

a zero-copy receive data path capable of transferring receive data directly from the TCP/IP protocol accelerator to the application buffer; and

a one-copy receive data path capable of transferring the receive data through a DTI receive buffer to the application buffer.

8. The apparatus of claim 1, further comprising:

a DTI buffer pool managed by the adaptation module, wherein the buffer pool enables dynamic allocation of the set of memory structures included in each one of the plurality of DTIs.

9. The apparatus of claim 1, further comprising:

a kernel agent to initialize the DTI accelerator and the TCP/IP accelerator;

a set of control tables shared by the kernel agent, the DTI accelerator, and the TCP/IP accelerator;

a raw packet driver to provide a default communication path between the DTI accelerator and an operating system; and

an exception agent to process packets not processed by the TCP/IP accelerator or the operating system.

10. The apparatus of claim 9, further comprising:

an interrupt proxy to receive an interrupt associated with the adaptation module from the DTI accelerator, the interrupt proxy to relay the interrupt to the adaptation module to enable processing of the interrupt.

11. A system, comprising:

an apparatus including an adaptation module, a plurality of direct transport interfaces (DTIs), wherein each one of the plurality of DTIs includes a set of memory structures, and wherein the adaptation module is capable of providing a translated sockets call associated with an application program to a selected one of the plurality of DTIs, and a DTI accelerator, wherein the set of memory structures included in the selected one of the plurality of DTIs is capable of coupling the translated sockets call to the DTI accelerator, and wherein the DTI accelerator is capable of coupling the set of memory structures included in the selected one of the plurality of DTIs to a Transport Control Protocol/Internet Protocol (TCP/IP) accelerator; and

a general purpose input/output interconnect interface capable of being communicatively coupled to the apparatus.

12. The system of claim 11, further including:

a DTI buffer pool managed by the adaptation module, wherein the DTI buffer pool enables dynamic creation and destruction of each one of the plurality of DTIs.

13. The system of claim 11, further comprising:

a kernel agent to manage the DTI accelerator and the TCP/IP accelerator.

14. The system of claim 13, wherein the kernel agent includes an address translation table, a context hash table, a context table, a routing table, a set of kernel space queues, a plurality of interrupt event queues, and a kernel buffer pool.

15. A method comprising:

receiving an application sockets call from an application program at an adaptation module;

deriving a translated sockets call from the application sockets call;

receiving the translated sockets call at a direct transport interface (DTI) having a set of memory structures;

coupling the translated sockets call to a DTI accelerator using the set of memory structures; and

coupling the set of memory structures to a transport control protocol/internet protocol (TCP/IP) accelerator.

16. The method of claim 15, further comprising:

associating a send queue and a receive queue with the DTI to enable posting send and receive descriptors, respectively.

17. The method of claim 15, further comprising:

associating a doorbell address with the DTI to enable posting event notices from the adaptation module to the DTI accelerator.

18. The method of claim 15, further comprising:

associating an event queue with the DTI to enable posting event notices from the DTI accelerator to the adaptation module.

19. The method of claim 15, further comprising:

transferring send data directly from an application buffer associated with the application program to the TCP/IP accelerator.

20. The method of claim 15, further comprising:

transferring send data indirectly from an application buffer associated with the application program to the TCP/IP accelerator through a send buffer associated with the DTI.

21. The method of claim 15, further comprising:

transferring receive data directly from the TCP/IP accelerator to an application buffer associated with the application program.

22. The method of claim 15, further comprising:

transferring receive data indirectly through a receive buffer associated with the DTI to an application buffer associated with the application program.

23. The method of claim 22, further comprising:

temporarily storing a plurality of packets received out-of-order in the receive buffer associated with the DTI.

24. The method of claim 22, further comprising:

temporarily storing a plurality of packets received when no sockets receive call is pending in the receive buffer associated with the DTI.

25. The method of claim 15, further comprising:

managing the DTI accelerator and the TCP/IP accelerator using a kernel agent buffer pool.

26. The method of claim 25, further comprising:

translating a virtual address included in the translated sockets call into at least one physical address to access a memory structure included in a virtual memory space associated with the application program using a translation table included in the kernel agent buffer pool.

27. The method of claim 25, further comprising:

translating IP addresses and TCP port numbers received in a network packet into a pointer to a DTI context stored in a context table in the kernel agent buffer pool.

28. The method of claim 27, further comprising:

calculating a hashed value from the IP addresses and TCP port numbers received in a network packet and using the hashed value as an index into a context hash table in order to fetch from the context hash table a pointer into the context table stored in the kernel agent buffer pool.

29. The method of claim 25, further comprising:

translating a destination IP address of a packet being sent to a network into a local subnet address using a routing table stored in the kernel agent buffer pool.

30. A method comprising:

receiving a connection request from a client;

associating the connection request with a master DTI;

receiving a translated sockets call at the master DTI, wherein the translated sockets call is derived from an application sockets call;

acquiring a child DTI from the master DTI;

associating the child DTI with the connection request;

completing a connection sequence to provide an established connection associated with the connection request; and

notifying an ETA adaptation module of the established connection.

31. The method of claim 30, wherein associating the connection request with a master DTI further comprises:

calculating an index into a hash table.

32. The method of claim 30, wherein acquiring a child DTI from the master DTI further comprises:

acquiring the child DTI from the master DTI through a master DTI queue.

33. The method of claim 30, wherein associating the child DTI with the connection request further comprises:

calculating an index into a hash table.

34. The method of claim 30, further comprising:

sending a first connection acknowledgment to the client; and

waiting for a second connection acknowledgment from the client.

35. The method of claim 30, further comprising:

writing a connection ready notification to a master DTI event queue.

36. An article comprising a computer storage medium storing associated data, wherein the data, when accessed, results in a computer performing:

receiving an application sockets call from an application program at an adaptation module;

deriving a translated sockets call from the application sockets call;

receiving the translated sockets call at a direct transport interface (DTI) having a set of memory structures;

coupling the translated sockets call to a DTI accelerator using the set of memory structures; and

coupling the set of memory structures to a transport control protocol/internet protocol (TCP/IP) accelerator.

37. The article of claim 36, wherein the machine-accessible medium further includes data, which when accessed by the machine, results in the machine performing:

maintaining a set of logically separate control and transfer flows for a plurality of DTIs including the DTI using a context table included in a kernel agent buffer pool.

38. The article of claim 36, wherein the machine-accessible medium further includes data, which when accessed by the machine, results in the machine performing:

communicating over a default communication path between the operating system and the DTI accelerator using a raw packet driver.

39. The article of claim 36, wherein the machine-accessible medium further includes data, which when accessed by the machine, results in the machine performing:

processing a subset of packets included in a plurality of packets not processed by the operating system or the TCP/IP accelerator using an exception agent.