dmesg: re0 at pci1 dev 0 function 0 "Realtek 8168" rev 0x11: RTL8168G/8111G \ (0x4c00), msi, address 00:16:96:ec:88:51 rgephy0 at re0 phy 7: RTL8251 PHY, rev. 0
Can only survive heavy TCP traffic (e.g. large download or a test with tcpbench/iperf) on mode "media: Ethernet 1000baseT full-duplex (1000baseT full-duplex,rxpause,txpause)" for a few seconds before the interface chokes completely and the console starts emitting "re0: watchdog timeout" messages. Functionality can be restored by downing+deleting the interface and then bringing it back up with /etc/netstart. Strangely, it stays functional indefinitely if telling tcpbench/iperf to use only UDP traffic, albeit with a heavy ~25% packet loss. Perhaps a problem with TCP checksumming? If running constrained on mode "media: Ethernet 100baseTX full-duplex (100baseTX full-duplex,rxpause,txpause)" it also appears to be stable for any type of traffic and any duration, but tcpbench/iperf sometimes shows an initial packet loss of a few percent which quickly tapers off to 0%.
