[Xen-users] network packet storm e1000_clean_tx_irq: Detected Tx Unit Hang

["Luke S. Crawford" <lsc@xxxxxxxxx>]

I've got a multi-bridge setup here;  xenbr 0 through 5.   In fact, we have 
several identical setups.  Well, one of them gives us a packet storm that 
looks a lot like what you get when you bridge two switches twice without 
STP.   The bridges look okay, though;  also, I notice the TX packet count 
is massively high on peth2, but it is reasonable on both eth2 and vif0.2 
(and on eth0 within the DomU attached to xenbr2)

I get the following errors in dmesg when the switch port for peth2 is 
enabled (shortly thereafter it eats the network)
e1000: peth2: e1000_watchdog_task: NIC Link is Up 100 Mbps Full Duplex
xenbr2: port 2(peth2) entering learning state
xenbr2: topology change detected, propagating
xenbr2: port 2(peth2) entering forwarding state
e1000: peth2: e1000_clean_tx_irq: Detected Tx Unit Hang
  Tx Queue             <0>
  TDH                  <a0>
  TDT                  <a2>
  next_to_use          <a2>
  next_to_clean        <a0>
buffer_info[next_to_clean]
  time_stamp           <1007bc816>
  next_to_watch        <a0>
  jiffies              <1007bc93e>
  next_to_watch.status <0>
e1000: peth2: e1000_clean_tx_irq: Detected Tx Unit Hang
  Tx Queue             <0>
  TDH                  <a0>
  TDT                  <a2>
  next_to_use          <a2>
  next_to_clean        <a0>
buffer_info[next_to_clean]
  time_stamp           <1007bc816>
  next_to_watch        <a0>
  jiffies              <1007bca06>
  next_to_watch.status <0>



Ideas?  do you think the 'tx unit hang' is causing the network spew?  or 
that the network spew is causing the 'tx unit hang' ?    I was guessing 
the former simply because the network spew is only seen on peth2;  it's 
not seen on any other interface.

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-users