[ofa-general] Re: IPoIB path caching

[Sean Hefty]

Michael S. Tsirkin wrote:
>> I believe a better solution is for everyone to use cached records, if 
>> they exist, with a feedback mechanism from the CM that removes paths on 
>> a connection failure or path migration event.
> 
> Ack timeout on an RC QP is also a good indication we should redo the lookup.

Do you know if we get a specific event for this?  (I don't remember.) 
Both the ib_cm and rdma_cm have interfaces that allow a user to report 
events on a connection.  They are used for path migration today, but we 
could easily extend them.

To minimize issues, I think we'll want some sort of feedback mechanism 
in place before enabling caching by default.

- Sean