[openib-general] ib_mthca "NOP command failed to generate interrupt (IRQ 169), aborting."

[Roland Dreier]

 > Machine:        EM64T
 > HCA :           Mellanox Technologies MT25208 InfiniHost III Ex (Tavor 
 > compatibility mode) (rev a0)
 > Old Kernel:             2.6.14  New Kernel              2.6.15.6
 > Old OpenIB revision:    3929    New OpenIB revision:    5685

Hmm, quite a lot has changed between those two revisions, but I
haven't heard of things breaking.

Some things to try:
 - Build with CONFIG_INFINIBAND_MTHCA_DEBUG turned on, and see if
   there's anything enlightening in the kernel log.
 - Update your HCA FW to 4.7.400 -- it shouldn't matter but 4.6.0 has
   a lot of known bugs so it can't hurt.
 - Update to the latest svn rev, if you want to run svn drivers.  Or
   you could try the stock drivers in 2.6.15.6.
 - Try passing the option "fw_cmd_doorbell=0" when loading ib_mthca.
   Again, it shouldn't matter but maybe it does.
 - Try a binary search of svn revisions between 3929 and 5685 to
   figure out when the driver stopped working.

 - R.