rpaulo@netbsd.org writes:
> Synopsis: Panic in tcp_input() by integer division fault
> please take a look at this PR

Ugh.  I'm thinking that I should back out revision 1.144 of tcp_output.c,
until we have a test program that excersizes these corner cases and fixes
for the bugs.  I'm pretty sure that backing out the patch will not stop
a carefully crafted program from panicing the kernel, but at least
it will not happen so easily.

Darrin