[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 12:51:34 PDT 2005


On Tue, 2005-06-28 at 15:43, Roland Dreier wrote:
>     Hal> Why does user space need to see all the timeouts ? I thought
>     Hal> it was only request timeouts which were needed to be
>     Hal> seen. The semantics for RMPP timeouts appear to be different.
> 
> Well, can the MAD core even generate timeouts for trap repress or
> response messages?

Yes (At least if the send was RMPP'd). That's what started this thread.
Not sure about normal MAD send of those. Sean ?

>   If so, I don't see a good reason for these events
> to be available to the kernel but not to userspace consumers.

The semantic of timeout for RMPP send appears to me to be different.

> It's much easier for consumers to ignore events they're not interested
> in than it is for consumers to deal with not getting events they want.

I'm not sure what a consumer would want a timeout to a non request
method.

Other alternatives are:
Change RMPP send timeouts to be same as normal MAD sends
or
Not return entire send packet on timeout. Is that a requirement also ?

-- Hal




More information about the general mailing list