[ofa-general] IPoIB-UD post_send failures (OFED 1.3)

Roland Dreier rdreier at cisco.com
Thu May 8 10:42:37 PDT 2008


 > Using OFED 1.3, we see what I believe is the same problem, but it
 > looks a little different. We don't get "NETDEV WATCHDOG", but 
 > we get an endless string of "post_send failed".

That's bad.  Did you check if the send is failing due to overrunning the
send queue?

 > (I suspect, but haven't verified, that the difference is due to 
 > the sharing of ipoib_dev_priv's tx_outstanding member between 
 > the UD and CM IPoIB QPs, the value of tx_outstanding is used 
 > to determine when to call netif_stop_queue().)

A while ago, I was worried about the handling of tx_outstanding and how
the driver makes sure that it doesn't post too many sends, but I managed
to convince myself that the code was OK.  Guess we should check it over
one more time.

 - R.



More information about the general mailing list