Currently I got a HP DL 120 in production with a HP NC107i PCIe Gigabit NIC (NetXtreme BCM5723) and FreeBSD 9.0-RELEASE #0 installed. At first those entries in /var/log/messages were ignored because everything was running smooth and the system seemed normal and stable:
bge0: watchdog timeout -- resetting
bge0: link state changed to DOWN
bge0: link state changed to UP
With increasing network load there were a few unusual latency peaks and finally the NIC shut down or the system was frozen (w/out any messages of course). I couldn’t figure that one out, it is a remote box w/out KVM access. After some research I found that In order to get rid of those messages you need to disable bge0 ASF driver features by adding
to your /boot/loader.conf – and yes, you really need to reboot your machine since sysctl hw.bge.allow_asf=”0″ will fail with “sysctl: oid ‘hw.bge.allow_asf’ is read only”. Those messages are gone now and the system seems stable as it should be.
p.s.: edit 25th of June 2012: The system is stable, issue solved.