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

[Sean Hefty]

>> Just to emphasize what Sean has pointed out, you are asking how can a CM
>> consumer know that a **local** QPN is not in the timewait state
>> according to the **remote** CM. Since the issue is with the remote CM,
>> it seems to me that pushing down timewait into verbs is not the correct
>> direction to look at.

We should still ensure that we don't give a user a local QPN that we know is in
timewait.  For example, a user 1 connects over a QP, transfers some data, then
destroys the QP.  User 2 allocates a new QP.  Can user 2 get the same QP as the
user 1?  If so, user 2 is likely to see a stale connection.  An option at this
point is for user 2 to destroy the QP and allocate a new one.  If they do this,
will they get the same QP again?

Now imagine if user 1 had created 1000 connections.  I believe that we should
make things as easy on user 2 as possible, including reducing the chance of
giving them a QP that the remote side is likely to have in timewait.

- Sean