On Wed, Jan 23, 2013 at 09:05:24AM +0000, u-codalist-9wcu_at_aetey.se wrote:
> > 
> > all get their .. entries (should be the same directory) in this bad
> > state at once.
> 
> I would become suspicious concerning the kernel part of Coda as ".." is
> subject to lot of special handling there. I did not test NetBSD for
> several months so can not contribute with the actual experience.
> 

Agreed - that does sound like a suspect high on the list, perhaps
another ipl64 issue.

> Another guess is about the client RVM being quite large.
> This may trigger subtle bugs if some pointers are being limited
> in range (truncated?) or otherwise if something collides with
> RVM in the address space - which always is an issue when you
> mmap at a fixed address as RVM needs.
> 

Oh, the bugs are less than subtle - there are 32bit types in the rvm
structures that would wrap, no doubt about it.  I did try fixing them
but that made my coda server core dump which sort of stalled me.

-- 
Brett Lymn
Staple Guns: because duct tape doesn't make that KerCHUNK sound - xkcd.com