what is the significance of using two hash flow tables, kernel space and user space? can you please point out method where the packet verdict ( drop / normal ) is taken after flow match ?
Thanks --Mandar ? -----Original Message----- From: Ben Pfaff <[email protected]> To: [email protected] Cc: [email protected] Sent: 14-Apr-2010 12:06:30 +0000 Subject: Re: [ovs-discuss] flow hash calculation in open vswitch. On Wed, Apr 14, 2010 at 02:38:00AM -0400, [email protected] wrote: > We are trying to control FTP traffic and for that we need support of port range. > We have updated? struct odp_flow_key and struct ofp_match to support port range ( for eg: 'tp_src'? as lower port limit and 'tp_src1' as upper port limit). > The memcmp of flow key is been replaced by individual struct member compare with an extra? port range bound check. > Still we are few steps short, can you suggest where else is the changes are needed ? A hash table cannot support range matches, so you will need to adopt a new data structure.
_______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss_openvswitch.org
