From: Auke K. <auk...@in...> - 2006-06-26 14:37:35
|
David Greaves wrote: > dmesg is *full* of these: > > e1000: eth0: e1000_clean_tx_irq: Detected Tx Unit Hang > Tx Queue <0> > TDH <b> > TDT <b> > next_to_use <b> > next_to_clean <1f> > buffer_info[next_to_clean] > time_stamp <72cb9> > next_to_watch <1f> > jiffies <734b3> > next_to_watch.status <0> > NETDEV WATCHDOG: eth0: transmit timed out > e1000: eth0: e1000_watchdog_task: NIC Link is Up 1000 Mbps Full Duplex > e1000: eth0: e1000_clean_tx_irq: Detected Tx Unit Hang > Tx Queue <0> > TDH <69> > TDT <69> > next_to_use <69> > next_to_clean <d2> > buffer_info[next_to_clean] > time_stamp <74b4f> > next_to_watch <d2> > jiffies <74c54> > next_to_watch.status <0> > > Can you suggest anything? > turn off tso - this should stop the tx timeouts from happening. You can do this with ethtool so it doesn't require a driver recompile. As for the problem itself: we're still working on it. Jesse might get back to you about it. Cheers, Auke |