On Mon, 2005-12-12 at 10:28, Michael S. Tsirkin wrote: > Quoting Hal Rosenstock <[EMAIL PROTECTED]>: > > > I've coded the following up since I grew tired of patching my kernels > > > to run sdp. Seems to work fine for me, can someone please speak up > > > on why this isnt a good idea for CMA, as well? > > > > Sorry for the slow response on this. I meant to dig this out over the > > weekend. > > > > I believe the reason it was changed from searching the netdevices list > > to ip_dev_find originally was that this really is a route lookup on the > > dest addr to determine what the local outgoing interface is and that > > can't be done directly from the netdevices list if the destination is > > not (IP) subnet local (e.g. gateway cases). > > What you are saying is, the original approach didnt do ip route lookup at all, > and thats why it was changed?
Yes, I believe so. > But now we only do the list walk if the ip route resolution returns > a loopback device, so we are ok, right? That's the way it seems to me (at least for SDP but not CMA (addr)). -- Hal _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
