[openib-general] ucma into kernel.org

[Sean Hefty]

Michael S. Tsirkin wrote:
> What I am saying that giving the application control over the timeouts
> seems more like a workaround than a solution.

The CM timeout depends on both the round trip time, as well as the time it takes 
the remote service to respond to the connection request.  The errors that we've 
seen were related to the amount of time that it took the remote service to 
respond, which was on the order of 30+ seconds.  I believe that this is an 
application specific problem that requires the application to determine the 
proper timeout.

Note that the underlying IB CM timeout already includes timeout information 
based on the obtained SA path record.  The exposed values are application 
response timeout values only.  For the retry count, I'm not aware of any way to 
programmatically determine the proper retry count.  Additionally, CM timeouts 
are linear, not exponentially increasing.

- Sean