[Intel-wired-lan] anyone aware of problem with 82599ES stuck sending TX pause frames?

[Chris Friesen <chris.friesen@windriver.com>]

Hi,

We're running linux 3.10 with the 3.13.10-k ixgbe driver.  We're seeing an 
intermittent issue with our 82599ES devices where they seem to occasionally get 
stuck sending TX pause frames.

controller-1:~$ sudo ethtool -S eth26 | grep flow_control
      tx_flow_control_xon: 0
      rx_flow_control_xon: 0
      tx_flow_control_xoff: 3446364
      rx_flow_control_xoff: 0

[wrsroot at controller-1 ~(keystone_admin)]$ sudo ethtool -S eth6 | grep flow_control
      tx_flow_control_xon: 103
      rx_flow_control_xon: 0
      tx_flow_control_xoff: 9174310
      rx_flow_control_xoff: 0


Generally it seems to happen after receiving a relatively small number of 
packets (though the data set is small so this may be chance):

13: eth4.85 at eth4: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 9000 qdisc htb state UP 
mode DEFAULT
     link/ether 90:e2:ba:37:9b:ec brd ff:ff:ff:ff:ff:ff
     RX: bytes  packets  errors  dropped overrun mcast
     96681      212      0       0       0       192
     TX: bytes  packets  errors  dropped carrier collsns
     77054616   136387   0       0       0       0


29: eth26: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc htb state UP mode 
DEFAULT qlen 1000
     link/ether 90:e2:ba:4f:9e:38 brd ff:ff:ff:ff:ff:ff
     RX: bytes  packets  errors  dropped overrun mcast
     105471     213      0       0       0       129291
     TX: bytes  packets  errors  dropped carrier collsns
     49801840   88058    0       0       0       0

ifdown/ifup seems to be sufficient to cause it to start behaving normally again.

Anyone seen anything like this?

Thanks,
Chris