Multiple cores/race conditions in IPv6 RA

[Jason Wolfe]

Word around town is Kubilay Kocak leaked this on Wed, Dec 09, 2015 at 09:02:48PM +1100:
> On 9/12/2015 7:58 PM, Andrey V. Elsukov wrote:
> > On 09.12.15 11:54, Kubilay Kocak wrote:
> >>> some time ago Mark Johnston has published there the patch related to
> >>> this problem:
> >>> https://lists.freebsd.org/pipermail/freebsd-net/2013-February/034682.html
> >>>
> >>> Maybe Mark has something to say about it.
> >>>
> >>
> >> Is it worth creating an issue report to track/resolve this, with 10.3
> >> coming up?
> > 
> > This problem exists since 4.x-5.x, so, I don't think that creating a
> > report will automatically resolve it :)
> > 
> 
> Not automatically no. But perhaps never being tracked has contributed to
> its non-resolution?

Andrey/Kubilay,

Thanks for the responses.  I was actually able to grab 2 patches from markj that we are currently testing.  The first is actually in Phab, maybe we can bribe markj to pick it back up :)

https://reviews.freebsd.org/D2254

The second one actually gets into the NDP code and is similar to the one posted on the mailing list that you linked.  It's a few months newer from May of 2013:

https://people.freebsd.org/~markj/patches/ndp-locking.diff

Initial testing is going well, I'll report back once we have more conclusive results.  It appears based various threads and reviews that the issues are a bit nebulous as there are many locking issues in this code.  Thank you to markj and others for going after the hot spots, and I've created a bug report to track the overall progress of this:

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=205165

Also, I don't believe these issues have been present for too long.  I see the entry deletion bug reported on 4.X/5.X, and that seems to be more related to the ndp binary itself.  Our 8.3-RELEASE installs did not exhibit the race conditions in the kernel code.

Thanks again!
Jason