Ilya Maximets <i.maxim...@ovn.org> writes: > On 4/14/22 17:34, Paolo Valerio wrote: >> Signed-off-by: Paolo Valerio <pvale...@redhat.com> >> --- >> lib/meta-flow.xml | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> > > Hi, Paolo. Thanks for the patch! > See some comments inline. > >> diff --git a/lib/meta-flow.xml b/lib/meta-flow.xml >> index 28865f88c..3445246f4 100644 >> --- a/lib/meta-flow.xml >> +++ b/lib/meta-flow.xml >> @@ -4101,6 +4101,15 @@ r r c c c. >> opcodes greater than 255 are treated as 0; this works adequately >> because in practice ARP and RARP only use opcodes 1 through 4. >> </p> >> + >> + <p> >> + In the case of fragmented traffic, a difference exists in the way >> + the field acts for IPv4 and IPv6 later fragments. > >> Because of the >> + headers structure, the protocol type is not processed for IPv6 >> + fragments with nonzero offset, meaning that matches based on this >> + field are not effective for those packets. > > This part reads as nw_proto makes no meaningful value in case of > IPv6 later fragments, but that is not correct. It has a value 44 > and users can match on it, IIUC. Could you clarify this in the > text? You may also look at how this is described in the design > doc: Documentation/topics/design.rst.
You're right. Re-reading I realize that it's badly phrased as the intention was to state the fact that upper layer protocols could not be matched as - and this is the missing piece you spotted - nw_proto for later ipv6 frags is set to IPPROTO_FRAGMENT regardless. I will fix the description specifying only the different behavior between ipv4 and ipv6 (dropping the header structure part as well). Thanks, Paolo _______________________________________________ dev mailing list d...@openvswitch.org https://mail.openvswitch.org/mailman/listinfo/ovs-dev
