On Wed, Jul 18, 2018 at 04:27:45PM +0200, Jan Stary wrote: > This is 6.3-current on and amd64 PC (dmesg below), using > > re0 at pci2 dev 0 function 0 "Realtek 8168" rev 0x06: RTL8168E/8111E-VL > (0x2c80), msi, address 50:e5:49:36:ec:0d > > as the NIC. With a hostname.re0 that says > > inet 192.168.11.3 255.255.255.0 > > it gets configured fine, but only uses 100baseTX. > Now that I got me this 1000Mbps switch, I would like to have 1000baseT. > > Why is it that the autoselect chooses 100baseT > when both the NIC and the switch it is connected to > are capable of 1000Mbps? > > When I change the hostname.re0 to > > inet 192.168.11.3 255.255.255.0 192.168.11.255 media 1000baseT > > (and run 'doas sh /etc/netstart re0'), it becomes > > re0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500 > lladdr 50:e5:49:36:ec:0d > index 1 priority 0 llprio 3 > groups: egress > media: Ethernet 1000baseT (none) > status: no carrier > inet 192.168.11.3 netmask 0xffffff00 broadcast 192.168.11.255 > > and indeed, I lose all conectivity. > > Is anyone seeing the same? > Is there something obvious I am missing? > > All I found on this was this old thread: > http://openbsd-archive.7691.n7.nabble.com/Forcing-re-driver-to-1000baseT-no-connection-4-4-release-td79375.html > > Jan
Got a similar one running up until 6.2, when i switched to a intel dualport without any problems. As the other two said, check the cable. Both devices need to have the same media settings, ie autoneg turned on or off. If you just turn it off on the card, the switch maybe doesn't get it right and disables the port. If you turn out the autoneg on the card, do so on the switchport, too. The most common error is a broken cable or a "special" cable from a vendor with only 4 cable cores for 100MBit without pins 4/5 and 7/8. hth, marc
