8.0 regression: wired network slow to initialise ?
pyunyh at gmail.com
Fri Jan 8 18:23:09 UTC 2010
On Fri, Jan 08, 2010 at 06:00:10PM +0100, martinko wrote:
> John Baldwin wrote:
> >On Friday 08 January 2010 4:23:09 am martinko wrote:
> >>$ grep sk0 /etc/rc.conf.local
> >>When booting up or netif restart:
> >>Starting Network: lo0 sk0.
> >>lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> metric 0 mtu 16384
> >> options=3<RXCSUM,TXCSUM>
> >> inet6 fe80::1%lo0 prefixlen 64 scopeid 0x5
> >> inet6 ::1 prefixlen 128
> >> inet 127.0.0.1 netmask 0xff000000
> >>sk0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> metric 0 mtu 1500
> >> options=b<RXCSUM,TXCSUM,VLAN_MTU>
> >> ether 00:11:22:33:44:55
> >> media: Ethernet autoselect (none)
> >> status: no carrier
> >> ^^^^^^^^^^
> >>Please note that on older versions of FreeBSD it printed 'status:
> >>active' and also IP address obtained via DHCP. The same now happens
> >>whether booting up or restarting via /etc/rc.d/netif and is a bit
> >>confusing at best. Other network services starting on boot seem to be
> >>not affected at least.
> >SYNCDHCP would probably restore the old behavior but make your boot take
> Please note that it displays "status: no carrier". Is this really
> dependent on DHCP ? Even if I `netif stop` network interface status is
> active until I plug out network cable (only then it's no carrier).
I think link state handling of sk(4) needs more code.
dhclient(8) wants to see an established link before sending DHCP
request so driver should maintain correct link state to make
dhclient(8) happy. Due to hardware bugs of PHY sk(4) used several
workarounds but that made it harder to keep track of link state
More information about the freebsd-stable