[openib-general] Question about QP's in timewait state and CM stale conn rejects

[Sean Hefty]

Arlin Davis wrote:
> How can a consumer know for sure that the new QP will not be in a 
> timewait state according to the CM?

Given that the QP may have been in use by another process, I don't think that 
there's any way for the new owner to know.

> Does it make sense to push the timewait functionality down into verbs?

This may be a clean way of handling the issue, but... see below.

> If not, is there a way for the 
> CM to hold a reference to the QP until the timewait expires?

For userspace QPs, the CM doesn't have access to the QP, so some sort of special 
call into verbs would be needed.

Even if we pushed timewait handling under verbs, a user could always get a QP 
that the remote side thinks is connected.  The original connection could fail to 
disconnect because of lost DREQs.  So, locally, the QP could have exited 
timewait, while the remote side still thinks that it's connected.

- Sean