On Fri, Aug 5, 2011 at 9:17 PM, Bob Pearson <[email protected]> wrote: > There are a couple of issues with SRP/RoCE. One of them is that the current > scheme for configuring connections is based on DM MADs. The other is the one > you mention. We (SFW) looked at both problems a couple of years ago and came > up with a version of SRP that runs over RDMACM. SRP was standardized before > RDMACM was a gleam in Sean's eye and assumes that it has the entire private > data area for CM MADs to pass parameters. But if you squeeze a little you > can get it all in after the area that RDMACM takes. We have versions of SRPI > and SRPT that can use RDMACM for connection setup. Since RDMACM supports > connection setup automatically for RoCE you're all fixed. In order to make > this commercially practical you would have to go back to NCITS and update > the SRP standard but I'm not too sure that there is enough interest to make > that happen. The other problem requires making changes to the special file > that sets up the connection. We also have this patch.
Has it been considered to add that ib_srp patch to the ib_rxe patch series, such that Dave and Roland can have a look at it ? Bart. -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html
