On Wed 22 Jan, 1997, Jason Thorpe <thorpej@nas.nasa.gov> wrote:
>On Tue, 21 Jan 1997 23:31:48 -0500 (EST) 
> Laine Stump <laine@MorningStar.Com> wrote:
>
> > Has this been noticed before (maybe even addressed?), or should I file a
> > PR? (Or is it worth it? I only did this for a quick fix to something
> > that I ended up doing the right way later.)
>
>May want to send a PR... could be a vnode locking bug...

When I first encountered this in 1.1, we put it down to the same
class of problem that rendered nullfs and unionfs useless for large
amounts of multiple process disk access. (We tried loopback mounting
nfs after failing with nullfs. It didn't help.)

John Kohl (I think) had previously submitted PRs for this problem,
but his suggested code fixes didn't help with the union/nullfs
problems (and I think he added to the PR to that fact, too). I also
commented upon this to Jan-Simon Pendry, and I think he concluded
that it was indeed a problem, but I've not heard from him recently!
(I guess it wasn't something simple to fix)

-- jrg.