On 07-Jul-23 1:51 PM, Xiao Liang wrote:
sa->hdr_len and prm->tun.hdr_len don't include L2 length so both should start in the diagram at the end of the ETH header. So the right way to compute datagram length is dgram_len = mb->pkt_len - sqh_len - sa->hdr_l3_off - sa->hdr_len + sizeof(struct rte_udp_hdr)|<- mb->pkt_len - sqh_len ->| |<- sa->hdr_l3_off ->|<- sa->hdr_len ->| |<- udph->dgram_len ->| +--------------------+------------+-----+-----+---------+-----+ | ETH | IP | UDP | ESP | payload | sqh | +--------------------+------------+-----+-----+---------+-----+ |<- sa->hdr_l3_off ->|<- l3_len ->| |<- sa->hdr_len ->| If hdr_len doesn't include L2 length, I would agree that dgram_len = mb->pkt_len - sqh_len - sa->hdr_l3_off - sa->hdr_len + sizeof(struct rte_udp_hdr) But then what's the point of sa->hdr_len - sa->hdr_l3_off in lib/ipsec/sa.c?
I will defer to Konstantin for a definite answer, that is if sa->hdr_len is supposed to include l2 length / offset or not. If it does, then the change that triggered this discussion is correct and we don't need to account for hdr_l3_off there.
