There’s a thread on the freebsd-current mailing list with some posts from Matthew Dillon (look for subject: ‘Re: panic: APIC: Previous IPI is stuck’) talking about how DragonFly’s IPI model would benefit FreeBSD enormously if ported over; no new facts, but a good read.
A running description of activity related to DragonFly BSD.
The really sad thing is that most people think that Matt is joking when he suggests that FreeBSD should incorporate DragonFly’s IPI messaging subsystem, when in fact he is serious.
I’m no expert, but the more I read about both approaches, and the more benchmarks I see, the more I am convinced that the DragonFly developers ideas and implementations really are superior to the equivelant functionality and infrastructure in FreeBSD 5.