[ofiwg] definition of a completion in OFI

Jason Gunthorpe jgunthorpe at obsidianresearch.com
Thu Mar 5 11:31:23 PST 2015

On Thu, Mar 05, 2015 at 07:18:23PM +0000, Underwood, Keith D wrote:

> > Destroying a RC QP while there is pending send q work on either side does
> > not have well specified behavior in IB, and that is exactly the case Sean is
> > probing with his sockets test.
> Which gets back to: the app is wrong.  No server should ever be
> written that way.  No app that it is ok to write that way is going
> to lack a "finalize()".

That is a fine answer - but in sockets land functions called shutdown and
close *DO* have well defined semantics for incomplete data, that are
completely different than the above.

Don't call the libfabric versions shutdown and close, or give them
socket like semantics.


More information about the ofiwg mailing list