[openib-general] ucma into kernel.org

[Michael S. Tsirkin]

Quoting r. Or Gerlitz <ogerlitz at voltaire.com>:
> > With regard to timeout, can CMA use the ARP roundtrip time to estimate it,
> > instead of asking the user or assuming hardcoded values?
> 
> ARP resolution might be cached by the OS - so the CMA can not count on 
> the round trip time

It's a problem, I agree, but hard-coding timeouts still does not make sense
to me - I honestly don't see how will an application know which value to
use here, since the roundtrip really depends on the topology.

Any ideas on how this can be handled correctly? Does CMA at least back off
exponentially on timeout?

-- 
MST