Roland Dreier wrote:
> If not, can some notes be sent to the list? I say lets learn from what
> you did so far...
In my experience, getting code to work over both IB and iWARP isn't that
hard. The main points are:
- Use the RDMA CM for connection establishment (duh)
- Memory regions used to receive RDMA read responses must have "remote
write" permission (since in the iWARP protocol, RDMA read responses
are basically the same as incoming RDMA write requests)
- Active side of the connection must do the first operation
- Don't use IB-specific features (atomics, immediate data)
Dunno the exact semantics for IB, but: write and send completions for
iWARP only indicate the buffer for the IO operation can be reused. It
does not indicate the data has been placed in the peers memory.
Steve.
_______________________________________________
general mailing list
[email protected]
http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
