It appears that this is an illegal instruction (illegal operand) trap in a
modified Rhat4U4 kernel. I am not sure about the line number, but perhaps
sdp_cma_handler() is processing an RDMA_CM_EVENT_ROUTE_RESOLVED event.
Based on the backtrace, this should be an RDMA_CM_EVENT_CONNECT_REQUEST
event.
I would verify that whatever structure that is associated with a
listening rdma_cm_id is still valid until after the listen has been
destroyed.
- Sean
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general