[OMPI devel] OMPI over OFA udapl (was Re: [ofa-general] OpenMPI and RDMA-CM)

[Arlin Davis]

Andrew Friedley wrote:

>
> Yep. My memory is dim, but I think that was OFED's DAPL, or it was in 
> the generic part of DAPL that all implementations seem to share.
>
> As hinted by the comment (I wrote it by the way), I think the best 
> solution would be if dat_ep_query() returned the port number 
> correctly.  Most of uDAPL seems to just pass around pointers to 
> internal data structures (which I'm not sure is the best idea in the 
> world), so it didn't seem like a trivial fix to me at the time.  I 
> remember considering reporting this as a bug, but I didn't because the 
> uDAPL standard didn't seem to enforce any requirements on passing the 
> port number around with the address, so it technically wasn't wrong.

I tend to agree. The common code should query the actual provider to get 
local address for the EP and not assume it is the IA address from the 
HCA used during the open. They are after all different bindings. I will 
take a look at the code.

>
> Was the OFED uDAPL code switched from something else to RDMA CM at 
> some point?  I'm almost certain I was running fine on OFED's uDAPL at 
> one point (in fact, a lot of the uDAPL BTL development I did was using 
> the OFED stack).

We had several interations while waiting for the rdma_cm code to become 
available. I am guessing that you were using one the early version that 
used sockets to setup the QP's.

-arlin