From mboxrd@z Thu Jan 1 00:00:00 1970 From: Rolf Manderscheid Subject: Re: [PATCH] ibsim: support LID portions of directed-route SMPs Date: Thu, 11 Aug 2011 15:27:07 -0600 Message-ID: <4E44492B.6080101@obsidianresearch.com> References: <20110809145018.GA2056@calypso.mtl.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <20110809145018.GA2056-iQai9MGU/dze+A/uUDamNg@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Alex Netes Cc: linux-rdma-u79uwXL29TY76Z2rM5mHXA@public.gmane.org List-Id: linux-rdma@vger.kernel.org Hi Alex, > route_MAD() breaks the LID route support. I think the change to route_MAD() is sound, there's no change in behaviour for LID-routed packets. The path argument is expected to be NULL in the case of a LID routed MAD, which causes an early return, but this is only an optimization. Even if you forget to pass in a NULL path (which I did in one case as you pointed out!), the path->cnt will be zero for LID routed MADs which causes the direct_route_*_MAD() functions to just return the same port that was passed in, so it works either way. I tested this with some simple topologies, running opensm and then doing a perfquery to each simulated node. I also used python-rdma with a new python simulator client which is the motivation for this patch. Without this change, python-rdma exposes the missing handling of leading and/or trailing LIDs in directed-route SMPs. I'll send an updated patch which passes a NULL path for LID-routed packets at the second call site. Rolf -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html