> In http://www.NetBSD.org/cgi-bin/query-pr-single.pl?number=7636, I've
> documented a problem I've been having with my SPARCclassic at 1.4. Basically,
> every time I try to compile the package swig, the kernel panics on the first
> C++ compile. (I've put the gory details in the PR.) Since this machine ran
> great with a 1.3I install, I don't think it's a memory or disk problem.
> 
> Has anyone seen this bug themselves? Also, if you have any ideas as to the
> root cause but need more info, please feel free to Email me and I'll
> help you as much as I can.


I haven't seen this particular instance, but there have been reports
of similar failures all of which indicate corruption of `extent'
or `pool' objects.   The most likely cause is insufficient protection
against modification from interrupt routines. And the most likely
candidate for getting corrupted in this way seems to be `expool',
the extent allocator's private memory pool.

-pk