[ofiwg] question about FI_CLAIM/FI_DISCARD and CQ event generation

Hefty, Sean sean.hefty at intel.com
Wed Sep 30 12:28:30 PDT 2015

> In implementing the FI_DISCARD feature for the gni provider,
> we are trying to decide whether a call to fi_trecvmsg with flags
> set to FI_CLAIM | FI_DISCARD (following a previously issued
> fi_trecvmsg with FI_PEEK | FI_CLAIM)  should generate a
> CQE.
> It appears that the the sockets and psm(at least version 2)
> do generate a CQ event for FI_DISCARD option,  but given there aren't
> any fabtests that test FI_DISCARD yet, we thought it might
> be good to double check on the mail list.

I think the intent was yes, there should be a CQE.  But I can see the confusion, considering that fi_cancel does not generate a CQE for the cancel call itself, and discard is somewhat similar to cancel.

IMO, it seems more consistent and easier for apps if a CQE *is* generated, and we match the sockets provider behavior.

- Sean 

More information about the ofiwg mailing list