Nigel & Zhong,
Just got the following response on a similar problem I have in the Open-FCoE
forum. Not good news unfortunately.
Ronnie
On Tue, 2009-09-08 at 10:00 -0700, Joe Eykholt wrote:
Here are the problems I know about regarding point-to-point mode in 2.6.31
and the current initiator code:
We never did the correct point-to-point protocol. The correct operation
would be that one or both sides send FLOGI (non-FIP, just regular FCoE) from
their
"control" MAC addresses to 0e:fc:00:ff:ff:fe, S_ID = 0, D_ID = fffffe,
and one or both sides accept that with LS_ACC and S_ID fffffe and D_ID=0.
The F-port bit being off in the LS_ACC indicates it's point-to-point.
The side with the greater WWPN sends PLOGI with the FC_IDs assigned
as S_ID and D_ID.
What was done instead, was the FLOGI LS_ACC assigned the FC_IDs of
0x010101 and 0x010102 to the nodes with the lesser and greater PWWNs,
respectively. Totally wrong, but it worked as long as we were talking only
to other open-fcoe nodes. The PLOGIs proceded from both nodes.
The FLOGI response would have a non-zero D_ID which the receiver would
take as it's assignment. This is what's done in the current target code
on the open-fcoe.org site.
However, the current initiator code always uses a 0 D_ID on the FLOGI LS_ACC,
so point-to-point doesn't even work with the broken method that we had been
using.
Point-to-point doesn't get tested anymore; all testing is done with switches.
I've been planning to fix this all to be correct, but it there's always
been something more important to do.
Joe
--
This message posted from opensolaris.org
_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss
