[openib-general] Handling SM class (SMInfo vs. other queries)

Mon Oct 25 10:41:15 PDT 2004

On Mon, 25 Oct 2004 10:34:09 -0700
Roland Dreier <roland at topspin.com> wrote:

>     Sean> If the MAD is not consumed by the driver, the MAD
>     Sean> layer may update the MAD and call process_local_mad a second
>     Sean> time, correct?
> 
> Sure, I guess so -- nothing should break if the MAD layer does this
> (since the driver can't remember that it already saw the MAD).  But
> when do you see this being done?  If the driver didn't handle the MAD
> the first time around, it's unlikely to do anything different on the
> second try (unless the MAD layer does something extreme like change
> the attribute ID, and I can't think of a time when we'd do something
> like that).

I was thinking of DR MADs, that have several checks, plus updates to the hop_ptr.  Are we talking about handing MADs to the driver immediately after it is received, or immediately before it would be dispatched to other clients?