On Thu, 26 Nov 1998, Charles M. Hannum wrote:

> I think there is actually only one case where what you say can happen
> -- that being a fault causing a pagein over NFS.  In this case, the
> process will do synchronous network I/O, which goes through the code
> in uipc_socket.c that uses uiomove().

Well, Murphy strikes again... Sure, I reckon my Shark is now less
susceptible to SEGVs, but I just did some more stress testing and would
you know it, I got a couple.

*sigh*

Cheers, Steve