[ofa-general] Re: [ewg] [PATCH] IB/ipoib: avoid WARN_ON on NULL path->ah

[Moni Shoua]

Eli Cohen wrote:
> On Tue, Sep 23, 2008 at 02:15:47PM +0300, Moni Shoua wrote:
>> I agree
>> I gave a thought here.
>> It's possible, when path_rec_completion() is called with nonzero status, to do nothing with ah.
>> Only when path query finishes with success do the replacement.
>> This is good for cases when old_ah is still good (no remote LID change happened).
> 
> But what you're saying is that even though no SM responded to a path
> query, you still want to use the old ah which is an even stronger
> request than what the original path did (which waited until the new
> path resolution to complete before updating the ah). Can you think of
> scenario that can cause this?
> _______________________________________________
> ewg mailing list
> ewg at lists.openfabrics.org
> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ewg
> 
I agree that this is a stronger request now. 
Suppose a host that has an ah to another host gets a SM_CHANGE event.
I would like that the first host will send path queries to the existing path because there is a chance
that the other host changed its LID. However, the new SM might not be ready immediately so the first host
might need to retry the path refresh. In this case I would like to keep the old ah since the chances that it is 
a good ah are very high.
BTW, the description of the scenario above is in a code that I intend to send soon.