[openib-general] Solicited response with no matching send request

[Sean Hefty]

Hal Rosenstock wrote:
>>If we want to route the MAD to the corresponding agent, however, we can 
>>do that.  But doing this only seems useful if a client is duplicating 
>>functionality, which only makes sense to me for user-mode clients.
> 
> 
> If we want to limit this to user mode clients only, we would need an
> extra parameter on register to indicate whether the client was kernel or
> user mode. This clearly wouldn't be very trustworthy. Is there a better
> way ?

Since the registration would actually be done in the kernel, I think 
that we can trust it.  It's just that before supporting this, I'd like 
to make sure that routing unmatched responses is really the right solution.

I.e. Is this something that kernel mode clients would need?  Does it 
make sense for clients to duplicate additional functionality, such as 
RMPP?  Would a solution that duplicated RMPP functionality in user-mode 
be better than one that only allowed for managing timeouts?

- Sean