TCP socket problem on Sept. -CURRENT snapshot

[Petr Holub]

> > It occurred from several different addresses and results in being
> > unable event to ping that IP until reboot.
> 
> This is "normal" for the current time being, as the output in question
> is being used to help track down a TCP-related bug in -CURRENT:

Thank you! Actually, the more painful problem is that it made some
IP addresses unreachable afterwards: I was unable to ping them, 
to make any TCP connection, etc. Though I was unable to ping
(almost all) other IPs and I was also able to ping/connect those
``stuck'' IPs from another host siting on the same network. So it
looks like the problem is actually some per-IP bug... Any reports
of this behavior reported before before I start looking into this
deeper?

Just for reference, I'm using 200709 snapshot of 7.0-CURRENT
for amd64, running on Core2 Duo with a custom kernes with
WITNESS turned off.

Thanks,
Petr