On Tue, Jun 16, 2015 at 12:38:06PM -0300, Thadeu Lima de Souza Cascardo wrote: > On Tue, Jun 16, 2015 at 08:16:33AM -0700, Ben Pfaff wrote: > > On Mon, Jun 08, 2015 at 01:05:41PM -0300, Thadeu Lima de Souza Cascardo > > wrote: > > > Support IGMPv3 messages with multiple records. Make sure all IGMPv3 > > > messages go through slow path, since they may carry multiple multicast > > > addresses, unlike IGMPv2. > > > > > > Tests done: > > > > > > * multiple addresses in IGMPv3 report are inserted in mdb; > > > * address is removed from IGMPv3 if record is INCLUDE_MODE; > > > * reports sent on a burst with same flow all go to userspace; > > > * IGMPv3 reports go to mrouters, i.e., ports that have issued a query. > > > > > > Signed-off-by: Thadeu Lima de Souza Cascardo <casca...@redhat.com> > > > > Thanks! I have a few comments. > > > > I am surprised that igmpv3_header in packets.h is marked OVS_PACKED, > > because it looks like all of the fields are naturally aligned. Does > > this structure need to be packed? > > > > Similarly for igmpv3_record. However, in this struct I would also > > change 'maddr' from ovs_be32 to ovs_16aligned_be32, because there are > > situations where packet headers can be aligned on an odd 16-bit boundary > > and using ovs_16aligned_be32 makes it easier to handle that safely on > > RISC architectures. > > That was me basing my first version of the code on LACP. I was about to > test my v3, I will take the chance and apply your suggestions before I > test and resubmit.
Thanks for the hint about LACP. It looks like one of the structures there doesn't really need to be packed either. I'll send a patch. _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
