[Bug 221385] [Regression] v6 mapped v4 addresses not working in 11.1

[Lewis Donzis]

Would appreciate some guidance on this.  It seems like a reasonably serious regression, so I’m surprised it hasn’t already been fixed.

Since all of our code binds a particular library, we were able to work around it by overriding the weak referenced socket() with our own version that creates the socket and explicitly turns off the V6ONLY option, but that doesn’t help with third-party applications.

Thanks,
lew


> On Sep 13, 2017, at 7:47 AM, Lewis Donzis <lew at perftech.com> wrote:
> 
> Hello.
> 
> This particular bug is a real problem in our embedded system, and we’re trying to decide whether to go back to 11.0 or wait for a patch for 11.1.
> 
> We downloaded just the one file containing the fix, but it wouldn’t compile, so we downloaded -CURRENT and verified that it works fine.  But it’s unclear to me whether the code in -CURRENT is slated for an 11.1 patch, or if it won’t show up until 11.2 or even 12.0.  I see the "mfc-stable11 <https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=221385#h5>” flag on the bug set to a value of “?”.  Is that significant?
> 
> Perhaps someone can help us understand how to determine when and in which version a fix will show up.
> 
> Thanks,
> lew
>