Pann McCuaig wrote:
> 
> [root@kappa ~]# modprobe -r forcedeth ; modprobe forcedeth
> PCI: Enabling device 0000:00:08.0 (0000 -> 0003)
> PCI: Enabling device 0000:00:09.0 (0000 -> 0003)
> eth0: no link during initialization.
> eth1: no link during initialization.
>
> 
> It's an nVidia nForce NIC:
> 
> nVidia Corporation MCP55 Ethernet (rev a3)

Forcedeth ?
There is a known problem with the forcedeth driver, which causes the NIC 
to stop working when used with certain networks.
In one case we had, the NIC performed fine through tests with our 
network switch, and stopped working when used with the customers.
(Though the failure mode there was quite annoying - NIC came up and 
worked for a short period, then stopped sending packets)



 From the forcedeth source:

  * Known bugs:
  * We suspect that on some hardware no TX done interrupts are generated.
  * This means recovery from netif_stop_queue only happens if the hw timer
  * interrupt fires (100 times/second, configurable with NVREG_POLL_DEFAULT)
  * and the timer is active in the IRQMask, or if a rx packet arrives by 
chance.
  * If your hardware reliably generates tx done interrupts, then you can 
remove
  * DEV_NEED_TIMERIRQ from the driver_data flags.
  * DEV_NEED_TIMERIRQ will not harm you on sane hardware, only 
generating a few
  * superfluous timer interrupts from the nic.
  */









-- 
      John Hearns
      Senior HPC Engineer
      Streamline Computing,
      The Innovation Centre, Warwick Technology Park,
      Gallows Hill, Warwick CV34 6UW
      Office: 01926 623130 Mobile: 07841 231235