[openib-general] [PATCH] user_mad: In send_handler, only queue_packet on timeout if not response

Hal Rosenstock halr at voltaire.com
Tue Jun 28 13:21:43 PDT 2005


On Tue, 2005-06-28 at 16:12, Roland Dreier wrote:
>     Hal> I'm not sure what a consumer would want a timeout to a non
>     Hal> request method.
> 
> I guess my view would be that if a consumer would never want a
> timeout, then the kernel shouldn't generate them.  We be consistent
> between kernel space and user space though -- the umad module
> shouldn't be filtering out information.
> 
>     Hal> Other alternatives are: Change RMPP send timeouts to be same
>     Hal> as normal MAD sends or Not return entire send packet on
>     Hal> timeout. Is that a requirement also ?
> 
> I don't think we need to return the whole packet on timed out sends
> now that the MAD structure has the header first.  In fact it probably
> makes sense to leave the payload off of timeout responses.  We just
> need a way to return the transaction_id of the timed-out send.

OK. I'll work on this approach then.

I still think there's a semantic difference with timeouts which needs
more consistency.

-- Hal




More information about the general mailing list