Minxi Hou <[email protected]> writes: > The ovskey flow-string parser has no OVS_KEY_ATTR_SCTP entry, so a > flow string containing sctp(src=.../dst=...) parses without error but > silently drops the L4 key. The resulting flow carries only > ipv4(proto=132), and the kernel rejects it: match_validate() in > flow_netlink.c requires OVS_KEY_ATTR_SCTP when the IP protocol is > IPPROTO_SCTP and returns -EINVAL for the missing key. > > Register OVS_KEY_ATTR_SCTP in the parse table and add unit tests > (runnable via ovs-dpctl.py --test) covering SCTP and the neighboring > TCP/UDP/ICMP/IPv4 key parsers. > > Signed-off-by: Minxi Hou <[email protected]> > --- > .../selftests/net/openvswitch/ovs-dpctl.py | 129 ++++++++++++++++++ > 1 file changed, 129 insertions(+)
[...] > @@ -3165,5 +3171,128 @@ def main(argv): > return 0 > > > +def _init_ovskey_nlas(): > + """Initialize required NLA classes for ovskey parsing.""" > + nlmsg_atoms.encap_ovskey = encap_ovskey > + nlmsg_atoms.ovskey = ovskey > + nlmsg_atoms.ovsactions = ovsactions > + > + > +def _parse_flow(flowstr): > + """Parse a flow string and return the key.""" > + key = ovskey() > + key["attrs"] = [] > + key.parse(flowstr) > + return key > + > + > +def _find_attr(key, attr_name): > + """Find an attribute in parsed key.""" > + for attr in key["attrs"]: > + if attr[0] == attr_name: > + return attr[1] > + return None > + > + > +class TestOvsKeyParse(unittest.TestCase): > + """Unit tests for ovskey.parse() flow string parsing.""" [...] I don't think we need unit tests for the unit test... That seems like extra working making tests all the way down. Maybe someone disagrees, but to me a test case here feels a bit excessive.
