When the action is not supported by hardware, the packets will be sent to the OVS datapath to do the forwarding. Interfacing the datapath is via newly added netdevs to represent the VFs.
From: Jayanthi Annadurai [mailto:[email protected]] Sent: Thursday, April 17, 2014 7:57 PM To: Alon Harel Cc: Justin Pettit; Wes Felter; [email protected]; Rony Efraim; Or Gerlitz Subject: Re: [ovs-discuss] OpenVswitch hardware Acceleration When the hardware does not support the required action, Will the packet be routed to root or the VM that is owning the VF? On Wed, Apr 2, 2014 at 6:27 AM, Alon Harel <[email protected]<mailto:[email protected]>> wrote: We are working on enabling OVS to handle SRIOV/VFs. This includes: 1. Modeling VFs as netdevs connected to an OVS bridge instance 2. Modeling the switch (NIC embedded switch) as a netdev (there is an RFC proposed by Jiri in the netdev mailing list) 3. Adding the forwarding hardware (NIC embedded switch) underneath the OVS kernel datapath. The concept is: -- Configuration is reflected to the hardware via the netdevs. -- Forwarding is done in hardware when actions are supported or in software (kernel+user) when hardware does not support the required actions per the flow. We will be more than happy to elaborate the discussion. Alon -----Original Message----- From: discuss [mailto:[email protected]<mailto:[email protected]>] On Behalf Of Justin Pettit Sent: Wednesday, March 26, 2014 3:56 AM To: Jayanthi Annadurai Cc: Wes Felter; [email protected]<mailto:[email protected]> Subject: Re: [ovs-discuss] OpenVswitch hardware Acceleration Jayanthi Annadurai wrote: > Is the support for SR-IOV on the road map for OVS? There are no plans from any discussions I've heard. --Justin _______________________________________________ discuss mailing list [email protected]<mailto:[email protected]> http://openvswitch.org/mailman/listinfo/discuss
_______________________________________________ discuss mailing list [email protected] http://openvswitch.org/mailman/listinfo/discuss
