For closure purposes .. We resuscitated our roach with a USB wiggler, upgraded to the (not quite latest) tgtap and tcpborphserver in http://www.mail-archive.com/[email protected]/msg01370.html, and now the tut2 10GbE tutorial works (after removing the "name" argument in tgtap)! (We actually want this version of tgtap to stay compatible with a colleague's roach).
Thanks for the help! Steve On Thu, Jun 10, 2010 at 3:19 AM, Jason Manley <[email protected]>wrote: > > > > Looks like I _do have a configuration problems: 1) a different IP > > address than what 10gbe_demo.py uses in tap_start() and 2) "Fabric > > interface is currently: Disabled". > Yeah, from your printout I can see that the link is up and correctly > established, but your 10GbE core is not configured correctly. Try'n do > a "ps aux" on your ROACH and make sure tgtap is running with the > correct parameters. You can even start it manually if you want to aid > debugging. > > > if my model name is "r_10gbe" and block name is "gbe0" than is my > > complete block a combination of those? (or does it include hard > > drive path, etc.) > Sorry, I didn't explain this very well. For the purposes of tgtap (and > BORPH in general), only the block names are important (not the model > names). So in this case your core would be called "r_10gbe". If it > were in a subsystem though, then it'd be mysubsystem_r_10gbe. As you > can see, in these situations the name gets very long and the older > tcpborphserver/tgtap falls over. > > > I was wondering if my yellow 10gbe blocks are too "new" for my older > > configuration, but they are dated before the 10gbe tutorial files. > The only recent change to the 10GbE cores/block involves the support > for the old 10GbE (version "1") core, which I highly recommend you use > because the new one (v2) still has a few kinks in the receive side. > Unfortunately the old core doesn't run at higher clock speeds (good > luck meeting timing much over 220MHz). Beware though that very old > compilations won't work even with the v2 core (there were bus changes > towards the end of 2009). If it's dated in 2009, I'd be wary and try a > recompile with the latest SVN versions. > > > Maybe I'll just bite the bullet and see if upgrading my > > configuration at least gets the (newer) tutorial to work ... > I'm sorry this didn't work out for you. We're working on making this > process easier on ROACH-II. It will feature an onboard JTAG-USB > bridge, so that in theory you can bring the whole board up from > scratch with just a single USB cable. > > Jason >
