Lustre devel - Jan 2007 - [Bug 11616] ASSERTION(error != 0 || conn->ibc_state >= IBLND_CONN_ESTABLISHED) causes nmi

Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11616

           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|bugs@clusterfs.com          |eeb@clusterfs.com
             Status|NEW                         |ASSIGNED


Created an attachment (id=9440)
Please don''t reply to lustre-devel. Instead, comment in Bugzilla by
using the following link:
 --> (https://bugzilla.lustre.org/attachment.cgi?id=9440&action=view)
debug patch

Can you attach (dmesg | grep o2ib) to this bug?  Any errors reported before the

assertion failure should provide a clue.  It would also help to enable network
error printks (echo + neterror > /proc/sys/lnet/printk) beforehand.

FYI, this assertion states that the only reason for closing a connection before

it becomes established is if there has been an error.  A trawl through the code

hasn''t thrown up any obvious errors and AFAICS the only way this could
happen
is
if the CM delivered RDMA_CM_EVENT_DISCONNECTED before
RDMA_CM_EVENT_ESTABLISHED,
which I believe shouldn''t be possible (but I''ll check).  In
any case, this
patch
adds some debug in case this is occurring.