Method for thread reduction in a multi-thread packet processor
Key lookup operations are broken into two instructions: a Key Dispatch Instruction (KDI), and a Return Result Instruction (RRI). The thread uses KDI to dispatch key information to a selected coprocessor to initiate a key lookup operation. Upon dispatch of the key value to the coprocessor, the KDI is retired to enable the thread to continue to dispatch and retire addition instructions in the pipeline and does not go idle. Subsequently, the thread will issue a RRI to obtain the key lookup result from the coprocessor. While a thread is executing, it maintains, as part of its context, a busy flag per coprocessor in a scoreboard register and a return result register per coprocessor. KDI causes the corresponding busy flag in the scoreboard register to be set. When the key lookup operation is complete, the busy flag is cleared and the result is stored in the return result register.
1. A method of thread reduction in a multi-thread packet processor, the method comprising the steps of:
executing, by a thread in a multi-thread packet processor, a first plurality of stages of a packet processing pipeline;
dispatching, by the thread, a key dispatch instruction to a coprocessor to cause the coprocessor to initiate a key lookup operation;
executing, by the thread, a second plurality of stages of the packet processing pipeline;
issuing, by the thread, a return result instruction to obtain a result of the key lookup operation from the coprocessor;
wherein the step of executing the second plurality of stages occurs after the step of dispatching the key dispatch instruction and prior to the step of issuing the return result instruction.
2. The method of claim 1 , further comprising the steps of receiving a packet and assigning the packet to a free thread in the multi-thread packet processor.
3. The method of claim 2 , wherein each thread is assigned to a separate packet.
4. The method of claim 1 , further comprising maintaining a busy flag for the coprocessor.
5. The method of claim 4 , wherein the busy flag is set when the key dispatch instruction is dispatched to the coprocessor.
6. The method of claim 4 , wherein the busy flag is cleared when the coprocessor completes execution of the key lookup operation.
7. The method of claim 6 , wherein the return result instruction is not issued until the busy flag is cleared.
8. The method of claim 6 , wherein the coprocessor stores the result of the key lookup operation in a return result register, and wherein the return result instruction causes the result of the key lookup operation to be read from the return result register.
9. The method of claim 1 , wherein the step of executing, by the thread, the second plurality of stages of the packet processing pipeline occurs in parallel with implementation of the key lookup operation by the coprocessor.
10. A network element, comprising:
a coprocessor; and
a multi-thread packet processor, each thread configured to sequentially implement a plurality of stages of a packet processing pipeline, the plurality of stages including a first plurality of stages, a key dispatch instruction, a second plurality of stages, and a return result instruction,
wherein:
the key dispatch instruction is configured to cause the coprocessor to initiate a key lookup operation;
the return result instruction is configured to obtain a result of the key lookup operation from the coprocessor; and
the thread executes the second plurality of stages of the packet processing pipeline after implementing the key dispatch instruction to cause the coprocessor to implement the key lookup operation and prior to implementing the return result instruction to obtain the result of the key lookup operation from the coprocessor.
11. The network element of claim 10 , wherein each thread of the multi-thread packet processor is assigned to a separate packet.
12. The network element of claim 10 , further comprising multiple coprocessors and a scoreboard table maintaining a set of flags, each flag being associated with one of the coprocessors.
13. The network element of claim 12 , wherein a busy flag for a respective one of the coprocessors is set in the scoreboard table when the key dispatch instruction is implemented to cause the respective one of the coprocessors to implement the key lookup operation.
14. The network element of claim 13 , wherein the busy flag for the respective one of the coprocessors is cleared in the scoreboard table when the respective one of the coprocessors completes execution of the key lookup operation.
15. The network element of claim 14 , wherein the return result instruction is not issued until the busy flag is cleared.
16. The network element of claim 14 , further comprising a set of return result registers, each of the result registers being associated with a corresponding one of the plurality of coprocessors.
17. The network element of claim 14 , wherein the respective coprocessor stores the result of the key lookup operation in its associated return result register, and wherein the return result instruction causes the result of the key lookup operation to be read from the return result register of the respective coprocessor.
18. The network element of claim 10 , wherein the network element is configured such that execution of the second plurality of stages of the packet processing pipeline occurs in parallel with implementation of the key lookup operation by the coprocessor.