sbuf-related panic

[Ian Lepore]

On Tue, 2015-03-17 at 12:14 -0700, Craig Rodrigues wrote:
> On Tue, Mar 17, 2015 at 11:57 AM, Ian Lepore <ian at freebsd.org> wrote:
> 
> >
> > There is still a panic, one that I can't yet figure out why it wasn't
> > panicking before my changes, but I'm working on it.
> >
> >
> Can you run the kyua tests on your system?
> 
> 
> https://wiki.freebsd.org/201411DevAndVendorSummit?action=AttachFile&do=view&target=kyua_jenkins.pdf
> 
> Those tests seem to be reproducibly triggering the problem.
> 

Triggering it wasn't the problem, understanding why it didn't used to
fail and now it does was the tricky bit.  :)

It turns out it never failed before because nobody had tried to use
default allocations (pointer and size zero) with sbuf_new_for_sysctl()
before, and it didn't fail for me while I was testing my changes because
I had commented-out INVARIANTS in my kernel config and forgotten that
(doh!) so none of the assertions were actually being tested.

So I fixed my test environment, then I fixed the code.   Hopefully it's
all better now.

-- Ian