Hi,

we see a problem with the e1000 module and SL5 machines running  kernel.
2.6.18-53.1.14.el5 #1 SMP Wed Mar 5 10:08:25 EST 2008 x86_64 x86_64

Upon boot the e1000 based NIC is recognized, however once the device is
used by the dhcp client, it starts to time out: 
/-------
Apr 25 21:04:22 localhost kernel: e1000: eth1: e1000_watchdog_task: NIC Link is  Up 1000 Mbps Full Duplex, Flow Control: RX
Apr 25 21:04:22 localhost kernel: ADDRCONF(NETDEV_CHANGE): eth1: link becomes ready


Apr 25 21:04:22 localhost kernel: e1000: eth1: e1000_clean_tx_irq:
Detected Tx Unit Hang
Apr 25 21:04:22 localhost kernel:   Tx Queue             <0>
Apr 25 21:04:22 localhost kernel:   TDH                  <4>
Apr 25 21:04:22 localhost kernel:   TDT                  <4>
Apr 25 21:04:22 localhost kernel:   next_to_use          <4>
Apr 25 21:04:22 localhost kernel:   next_to_clean        <0>
Apr 25 21:04:22 localhost kernel: buffer_info[next_to_clean]
Apr 25 21:04:22 localhost kernel:   time_stamp           <fffc438b>
Apr 25 21:04:22 localhost kernel:   next_to_watch        <0>
Apr 25 21:04:22 localhost kernel:   jiffies              <fffc4b58>
Apr 25 21:04:22 localhost kernel:   next_to_watch.status <0>
\------
After about 20-30 repetitions of this message (with different
timestamps), the machine would crash. Has anyone seen this effect or any
recommendations?

The card in question identifies itself as follows:
/----
$ lspci 
01:07.0 Ethernet controller: Intel Corporation 82541PI Gigabit Ethernet Controller (rev 05)
\----



Thanks and kind regards,
	Felix 
-- 
  Dipl.-Ing. Felix Engel        | mail: [log in to unmask] 
  fon: +49 241 80 27629         | fax: +49 241 80 28306                
  RWTH Aachen University, SSS 611920, Templergraben 55, D-52056 Aachen