From: jsmith@dcs.warwick.ac.uk (James Smith)
Newsgroups: comp.sys.transputer
Subject: Re: D7405 'rspy -x' user supplied rsc's
Date: 19 Aug 1998 15:05:44 +0100
Organization: Department of Computer Science, Warwick University, England
Message-Id: <6rem3o$kd8$1@bise.dcs.warwick.ac.uk>
References: <K4TU5VAYSX110Ams@oxim.demon.co.uk>


Hi. I worked on this stuff for Inmos about 4-5 years ago. My memory of it
is a little hazy and I don't work for them any more, so they might have
changed a few things after I left. The example in the manual should be fine
- there's not much to go wrong. :)  

The first question is whether rspy works normally without the -x option? If
some of the transputers aren't resetting properly then this will cause
problems when rspy tries to look around the network. If it works ok then it
is more likely to be the rsc you created that is causing the problem, as a
lot of rspy is actually implemented through the same mechanism. The part
testing software is actually run as an rsc after the network has been
booted. You can even do

rspy /x part

to run the part.r2h and part.rah that come with rspy. It's therefore more
likely that your problem is with the rsc itself. As far as arguments are
concerned the code in the example should be fine. The code loading software
that rspy uses to run the rsc will provide those arguments itself in a
fixed format, so they should not be changed. I remember using a strange
line of code like

#OPTION "WUNA"

at the top of some code I was working on, but I can't even remember what
that does or whether it's relavent to this, so it's a bit of a long shot.

Building the rscs can't be done in the standard way as they can't use any
of the toolset libraries and things. There should be an example in the
manuals. I assume you're using this as well? I can't remember enough about
the details to be of any more help I'm afraid, and all the output from rspy
tells us is that sent and run the rsc to the root transputer ok, but then
failed to recieve and results back from the network. That indicates that
something is wrong with the rsc. It might be of use to do an idump and
idebug straight away afterwards to see what the root transputer is up to at
the time.

That's about where my knowledge starts to thin out unfortunately. It
definitely worked when I'd finished with it. They were always a bit tricky
to do, but the idea of the example was to provide people with some code
that worked so they could get working easily. I guess something must've got
broken.

-- 
James Smith
jsmith@dcs.warwick.ac.uk
http://www.dcs.warwick.ac.uk/~jsmith/

