[ofa-general] RDMA write completion question

[Michael Krause]

At 06:59 PM 5/24/2007, Steve Wise wrote:


>Devesh Sharma wrote:
>>On 5/24/07, Gleb Natapov <glebn at voltaire.com> wrote:
>>>Hi,
>>>
>>>  Does local RDMA write completion guaranties that a data that was RDMAed is
>>>already accessible in a destination's host _memory_?
>>Local RDMA write completion guarantees that the data you have RDMAed
>>has been copied into the remote buffer, without any data corruption.
>For iWARP, the local write completion simply means you can reuse the local 
>buffer and the the transport will deliver it or kill the connection.  The 
>data _could_ be queued in the local rnic and anywhere else in the tcp cloud.

This is where IB and iWARP do differ slightly.   iWARP may indeed not have 
transmitted anything to the remote RNIC while in IB, a completion should 
equate to the data being received by the remote CA.

As a general point though, the source RNIC is unlikely to issue a RDMA 
Write completion if it has only injected the packet into the Ethernet 
fabric.  It may issue prior to injection or in response to a TCP ACK 
indicating the RDMA Write was received by the remote RNIC but I doubt any 
do something in between.

Mike