udhcpc timeout with tg3

[Ferenc Wagner <wferi@niif.hu>]

Hi,

Testing the squuze beta1 installer I ran into a networking problem.

    Primary network interface:
      1. eth0: Broadcom Corporation NetXtreme BCM5701 Gigabit Ethernet [*]
      2. eth1: Broadcom Corporation NetXtreme BCM5701 Gigabit Ethernet    
    Prompt: '?' for help, default=1> 
    
    Configuring the network with DHCP  ..13%..20%..33%..40%..53%..60%..73%..80%..93%..100%
    Network autoconfiguration failed
    
    Your network is probably not using the DHCP protocol. Alternatively, the DHCP 
    server may be slow or some network hardware is not working properly.
    [Press enter to continue] 

Up to this, my DHCP server doesn't receive a single request.  But a
couple of seconds after this, a request comes and an IP gets leased to
the installer.  If I back out and execute a shell from the installer
main menu, I can see the IP address assigned to eth0, although no udhcpc
process is running by this time and eth0 is down.  Running udhcpc eth0
manually confirms that it fails to get a request out during its first
round of three packets, then it waits for too long before trying again:

~ # ip addr
1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN 
    link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
    inet 127.0.0.1/8 scope host lo
2: eth0: <BROADCAST,MULTICAST> mtu 1500 qdisc mq state DOWN qlen 1000
    link/ether 00:06:5b:8e:71:d5 brd ff:ff:ff:ff:ff:ff
3: eth1: <BROADCAST,MULTICAST> mtu 1500 qdisc mq state DOWN qlen 1000
    link/ether 00:06:5b:8e:71:d6 brd ff:ff:ff:ff:ff:ff
~ # dmesg -n8
~ # udhcpc eth0
 0.00 udhcpc (v1.17.1) started
      [ 1512.784911] ADDRCONF(NETDEV_UP): eth0: link is not ready
 1.14 Sending discover...
      [ 1515.784728] tg3: eth0: Link is up at 1000 Mbps, full duplex.
      [ 1515.796034] tg3: eth0: Flow control is off for TX and off for RX.
      [ 1515.808388] ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
 3.83 Sending discover...
 6.69 Sending discover...
 9.81 udhcpc: has been called with an unknown param: leasefail
29.85 Sending discover...
      Sending select for 10.253.2.7...
      Lease of 10.253.2.7 obtained, lease time 43200

(The first column contains my stopwatch readings).

Uping the interface takes 3 seconds to complete, so the switch doesn't
reach forwarding state (albeit portfast is set) before the first three
packets are sent.  Then netcfg times out before udhcpc would even try
again 20 seconds later.  It looks like the distribution of DHCPDISCOVER
packets is less than ideal.
-- 
Regards,
Feri.