Flow Cache

The PASS Flow Cache operation consits of the following three sub-operations:

The flow cache operation can be used to help glue the ingress path to the egress path to establish a fully automated processing chain without host intervention. It may be only applied to egress packets from the host to establish fully-offloaded egress chain operation with or without flow cache classification.

The flow cache classification module resides at PASS Egress0 clsuter and it supports 256 LUT1 entries, each entry consists of any combination of IP and UDP/TCP parameters as defined at paFcInfo_t. Each flow cache entry is also assocaited with a pre-determined egress flow where the egress packets will undertake up to 4-level of packet modification defined by the egress flow packet modification records, which are described at the next section. The API Pa_addFc() and Pa_delFcHandle are used to add and delete the Flow Cache entries respectively.

Each egress flow consists of up to four packet modification stages where the packet can undergo transformations in the egress path, before it is sent out from PASS. The four packet modification stages are

The detailed operation of each packet modification stage is defined at its corresponding egress flow record defintion sections(paEfRecLevel1_t, paEfRecLevel2_t, paEfRecLevel3_t and paEfRecLevel4_t) The API Pa_configEflowRecords is used to add/delete single or multiple packet modification records.

The ingress traffic may be routed to the egress path at each of the following four classification stages:

The advanced routing information data structure paRouteInfo2_t with flow cache operation data structure paEfOpInfo_t are used to define flow cache route.


Copyright 2014, Texas Instruments Incorporated