In a custom user space networking stack, backed by Netmap or PF_RING, is the ability to use TCP Segmentation Offload and/or Large Receive Offload lost?
The only place I can find anything that mentions NIC offloading w.r.t. Netmap or PF_RING is in Netmap: a novel framework for fast packet I/O, but it isn't apparent to me if it's usable or if it's operating transparently in the NIC:
Popular hardware features related to TCP acceleration, such as hardware checksumming or even encryption, Tx Segmentation Offloading, Large Receive Offloading, are completely orthogonal to our proposal: they reduce some processing in the host stack but do not address the communication with the device.
Thank you for your time and happy holidays!