> On Jul 12, 2017, at 5:42 PM, Ben Pfaff <[email protected]> wrote: > > On Wed, Jul 12, 2017 at 03:47:07PM -0700, Justin Pettit wrote: >> The function nx_put_raw() shouldn't append "ct_nw_proto" to nx_match if >> the corresponding mask isn't set. >> >> Signed-off-by: Justin Pettit <[email protected]> >> --- >> lib/nx-match.c | 4 +++- >> 1 file changed, 3 insertions(+), 1 deletion(-) >> >> diff --git a/lib/nx-match.c b/lib/nx-match.c >> index cb0cad8458b9..c64953b8892b 100644 >> --- a/lib/nx-match.c >> +++ b/lib/nx-match.c >> @@ -1190,7 +1190,9 @@ nx_put_raw(struct ofpbuf *b, enum ofp_version oxm, >> const struct match *match, >> nxm_put_ipv6(&ctx, MFF_CT_IPV6_DST, oxm, >> &flow->ct_ipv6_dst, &match->wc.masks.ct_ipv6_dst); >> if (flow->ct_nw_proto) { >> - nxm_put_8(&ctx, MFF_CT_NW_PROTO, oxm, flow->ct_nw_proto); >> + if (match->wc.masks.ct_nw_proto) { >> + nxm_put_8(&ctx, MFF_CT_NW_PROTO, oxm, flow->ct_nw_proto); >> + } > > Please use nxm_put_8m instead, e.g.: > nxm_put_8m(&ctx, MFF_CT_NW_PROTO, oxm, flow->ct_nw_proto, > match->wc.masks.ct_nw_proto);
I thought about that, but I think nw_proto is normally not masked, and the other "nw_proto" fields in that file are similarly handled. Do you think we should make them behave the same? --Justin _______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
