> So, as stated before, the only workaround I know of is to keep some traffic
> on the interface.. but whether or not that works for everyone, it is
> obviously still a driver problem. If I had the time, I'd try to debug it..
> but it's (obviously) not causing a kernel panic, nor any messages to
> console, so I wouldn't know where in the code to start looking.

Yes, this problem is really our driver's limitation. 
See /sys/dev/ic/i82557.c line:1472.

OpenBSD's driver has same limitation, but FreeBSD's not.


Regards,