Bewerken

Delen via


Splitting IPv6 Frames

To support header-data split, a NIC must support splitting IPv6 Ethernet frames without any IPv6 extension headers. The NIC must be able to split such frames at the beginning of upper-layer-protocol header.

Support for IPv6 Ethernet frames with IPv6 extension headers is optional. A NIC can support some IPv6 options and not support others. The NIC must not split IPv6 frames that contain IPv6 extension headers that is does not support. The header portion of a split frame must contain the entire IPv6 header and all of the IPv6 extension headers that are present.

The NIC can also support header-data split for fragmented IPv6 frames. For more information about fragmented IPv4 frames, see Splitting Fragmented IP Frames.

Note  Supporting an IPv4 option, an IPv6 extension header or a TCP option, for the purposes of header-data requirements, implies the ability of the NIC to recognize the element, determine its length, include it in the header MDL and locate its end and the beginning of the next element in the frame.

If the header-data split provider splits an IPv6 frame, the indicated NET_BUFFER_LIST structures must have the NDIS_NBL_FLAGS_IS_IPV6 flag set in the NblFlags member. For complete information about setting header-data split flags in the NET_BUFFER_LIST structure, see Setting NET_BUFFER_LIST Information.

Additional Ethernet frame characteristics determine how to split IPv6 frames. If the frame is fragmented, see Splitting Fragmented IP Frames. If the frame contains TCP information, see Splitting Frames at the TCP Payload. If the frame contains UDP information, see Splitting Frames at the UDP Payload. For all other cases, see Splitting Frames Other Than TCP and UDP.