[ofa-general] Re: [RFC][PATCH] last wqe event handler patch

Eli Cohen eli at dev.mellanox.co.il
Thu Jun 26 08:18:44 PDT 2008


On Thu, Jun 26, 2008 at 08:05:12AM -0700, Roland Dreier wrote:
>  > So what you're saying is that the scenario Shirley described is not
>  > possible...
> 
> Seems like it... so the "one at a time" draining bug fix is probably not
> right.  The IB spec says:
> 
> If the HCA supports SRQ, for RC and UD service, the CI shall generate a
> Last WQE Reached Affiliated Asynchronous Event on a QP that is in the
> Error State and is associated with an SRQ when either:
> 
>     . a CQE is generated for the last WQE, or
>     . the QP gets in the Error State and there are no more WQEs on the RQ.
> 
> It is slightly ambiguous what "generated" means for a CQE -- it is
> conceivable that someone could interpret that to mean a CQE is "in
> flight" but has not necessarily become visible.
I believe "generated" means here that the CQE has been pushed into the
CQ (posted write) so if you drain the CQ you should see it (if the
chipset maintains cache coherency for the CQ buffer).

> 
> Anyway I think we need to debug this further.
> 
>  - R.
> _______________________________________________
> general mailing list
> general at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
> 
> To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general



More information about the general mailing list