| 摘要 |
A device and a method for processing a race condition in a multithread based network processor are provided to prevent performance degradation by enabling other threads to process other sessions without waiting completion of a task performed by the thread, as session data to be processed is stored to a queue of the thread processing the specific session without waiting. A thread queue stores the data to be processed by the thread. A CAM(Content Addressable Memory) stores information of the session processed by the thread. If other threads access the session to process the session while the thread of the thread queue processes the specific session, other threads store the session data to be processed to the thread queue of the thread. The thread continuously processes the data of the thread queue until the thread queue gets to be empty. The CAM stores a table mapping a list of the threads operated in a network processor to the information of the processed session.
|
