Methods for task scheduling through locking and unlocking an ingress queue and a task queue

A method for parallel processing implemented in a network unit, comprising:nemploying a first core to lock an ingress queue when the ingress queue is not locked by another core, encapsulate packets from the ingress queue into tasks for scheduling and push the tasks to one of a plurality of task queues based on one of a priority of the task, a processing flow of the task, or an order type of the task when the ingress queue is locked by the first core, and unlock the ingress queue after the pushing of the tasks;employing the first core to search a first subset of the plurality of task queues for a first unlocked task queue when the ingress queue is locked by another core, lock the first unlocked task queue and place a first task from the first unlocked task queue into an order queue when the first core finds the first unlocked task queue, and unlock the first unlocked task queue after placing the task;employing the first core to search a first remaining subset of the task queues for a second unlocked task queue when all of the first subset of the task queues are locked and the ingress queue is locked, lock the second unlocked task queue and place a second task from the second unlocked task queue into the order queue when the first core finds the second unlocked task queue, and unlock the first unlocked task queue after placing the second task, wherein the first remaining subset of the task queues comprises members of the task queues outside of the first subset of the task queues; andprocessing an order task that is at a head of the order queue.

In a network unit, a computer program product executable by a multi-core processor, the computer program product comprising computer executable instructions stored on a non-transitory computer readable medium that when executed by the multi-core processor cause the network unit to perform the following:nemploying a first core to lock an ingress queue when the ingress queue is not locked by another core; encapsulate packets from the ingress queue into tasks for scheduling and push the tasks to one of a plurality of task queues based on one of a priority associated with the task; a processing flow associated with the task, or an order type associated with the task when the ingress queue is locked by the first core; and unlock the ingress queue after pushing of the tasks;employing the first core to search for a first unlocked task queue from a first default subset of a plurality of task queues when the ingress queue is locked by another core, lock the first unlocked task queue and place a first task from the first unlocked task queue into an order queue when the first core finds the first unlocked task queue, and unlock the first unlocked task queue after placing the first task, wherein the first default subset is different from a second default subset of the task queues from which a second core begins a search for a second unlocked task queue; andemploying the first core to search a first remaining subset of the task queues for a second unlocked task queue when all of the first default subset of the task queues are locked by other cores and the ingress queue is locked by another core, lock the second unlocked task queue and place a second task from the second unlocked task queue into the order queue when the first core finds the second unlocked task queue, and unlock the first unlocked task queue after placing the second task, wherein the first remaining subset of the task queues comprises members of the task queues outside of the first default subset of the task queues; andprocessing an order task that is at a head of the order queue.