On Wed, 2008-09-10 at 01:17 -0400, Peter Memishian wrote: > > The following change is significant, and should be reviewed with > > scrutiny. More specifically, in order to get all of the ipnet_accept() > > cases to work properly, I had to re-design its structure a bit: > > > > http://zhadum/ws/seb/seb-ipobs/webrev.accept/ > > > > I'm going to push this change prior to receiving comments, as it fixes > > bugs that I'd like John to test against. I can address issues in the > > gate if any come up. > > (I reviewed this over IRC with Seb; I found some minor issues which > shouldn't affect testing that Seb will fix soon).
Yep, thanks. I've made the changes, and another change that followed from our discussion on broadcast handling. Basically, what we agreed upon was: because broadcast are locally looped back, we should be able to see them on output. My code previously restricted the observability to input only unless DL_PROMISC_PHYS was enabled. The additional change I made is related to multicast, in that there is equivalent looping back of multicast within IP for applications that want to receive multicast packets sent by other applications in the local stack. The result is: http://zhadum.east.sun.com/ws/seb/seb-ipobs/webrev.accept-2/ -Seb
