> BTW, I think we still need to add check for overflow in > __rte_pktmbuf_read() - > bpf jit is not the only user of it. > Unless, of-course, we want rte_pktmbuf_read() to work with wrapped- > around values.
It makes no sense for rte_pktmbuf_read() to wrap around. Its offset and length parameters are clearly specified and documented as unsigned, and there is no mention of wrap-around or implicit negative values when summing them. I have submitted a fix just now: https://inbox.dpdk.org/dev/[email protected]/T/#u -Morten
