[openib-general] GSI compromise

[Sean Hefty]

On Wed, 04 Aug 2004 16:18:19 -0400
Hal Rosenstock <halr at voltaire.com> wrote:

> Looks good although I need some more time to think about redirection
> before commenting on those aspects of the API.

*nods*  I also haven't given too much thought to the use of the API over QP0, so some additional tweaks may be necessary.
 
> In terms of timeout_ms, would 0 or some other value be no timeout (for
> request/response matching) ? 

I was assuming that a timeout_ms of 0 would be no timeout.  I.e. it's just a send.
 
> Rather than ib_mad_reg/dereg_class, it looks more like
> ib_mad_register/deregister to me now :-) 

agreed