[ofa-general] Recalculated Queue Sizes caused mthca Catastrophic Errors

[Jack Morgenstein]

On Wednesday 20 February 2008 21:50, Roger Spellman wrote:
> As I said, doubling the queue size solves the problem.  However, it
> would be better if the mthca driver did not create the problem in the
> first place.  If a QP is being created such that CQ >= SQ + RQ, then
> that relationship should be maintained.  Do others agree with me?
> 
There is no way the driver can do this.  A CQ is not necessarily restricted
to a single QP.  Furthermore, there is NO requirement that the same CQ be used
for both the SQ and the RQ of a QP.

The function ibv/ib_create_cq returns the cqe size actually created; the function
ibv/ib_create_qp also returns the number of work requests available in the sq and
the rq. 

It is the responsibility of the application to check these size values returned
by create_cq and create_qp to determine if the CQ is large enough.

- Jack