[openib-general] [PATCH] agent: Fix agent_mad_send PCI mapping and gather address and length

[Hal Rosenstock]

On Wed, 2004-11-10 at 10:36, Roland Dreier wrote:
> Yes, IPoIB works for me again.

Thanks for validating.

>     Hal> Also, it seems to me that no response needs to be handed to
>     Hal> process_mad.  Does this optimization make sense ?
> 
> I'm not sure I understand the question.  process_mad definitely needs
> a buffer to return a response in.  Are you suggesting that process_mad
> overwrite the input buffer when it generates a response?  That's
> probably OK although I'm not sure if it's much of an improvement
> (process_mad will probably have to allocate a response buffer
> internally and copy the response when returning).

I'm asking about also checking the method prior to calling process_mad. 
If the method is a response method (e.g. GetResp for one), we could
bypass calling process_mad. Or is this not worth the extra checks in the
MAD layer as it is low enough overhead and adds additional protocol
knowledge into the MAD layer ?

-- Hal