On 3 April 2015 at 13:46, Ben Pfaff <b...@nicira.com> wrote: > >> > Also, the commit you name was in 2013, and we've had other >> > contributions from FreeBSD contributors since then >> > (namely Ed Maste <ema...@freebsd.org>) and it seems like he >> > would have noticed if it were totally busted. >> >> You mean this message: >> http://openvswitch.org/pipermail/dev/2013-May/027594.html >> >> If so, Ed replied that he reviewed the patch but didn't apply it to test. :-) > > Ed is the author of the following 37 commits in the Open vSwitch > repository. Ed, did you really never test Open vSwitch on FreeBSD, > despite this long list of contributions?
The discussion above happened around the time I was moving away from using Open vSwitch in near-production on a daily basis on FreeBSD. I reviewed, but did not test, the change -- and clearly that review was insufficient. Prior to that time I was building, testing, and using it in a near-production environment on an ongoing basis. I know others are using Open vSwitch on FreeBSD in various capacities as well, such as Prof. Luigi Rizzo's team. I'm surprised that nobody spotted this until now. The offending flag that ends up in the sign bit is: #define IFF_MULTICAST 0x8000 /* (i) supports multicast */ so will certainly be set in many cases. Kevin, what failure mode did you observe? Note that we have Open vSwitch in the FreeBSD ports tree, and it's a relatively old version from around that time. As a result, FreeBSD users installing from ports wouldn't have encountered this issue (but would have to deal with an outdated version). Kevin's patch looks correct and I'm happy that he'll be updating the version in the FreeBSD ports tree as well. -Ed _______________________________________________ dev mailing list dev@openvswitch.org http://openvswitch.org/mailman/listinfo/dev
