Newsgroups: comp.sys.transputer
From: peterm@senet.com.au (Peter Morris)
Subject: Re: ST20450 and Inquest, are they even from the same planet?
Date: Fri, 25 Jul 1997 06:09:42 GMT
Message-ID: <33d842fd.0@kastagir.senet.com.au>

Mark@nlcc.demon.co.uk (Mark Ian Barlow) wrote:

>Having been involved in at least one project with the older T-series
>transputers where the hardware people did not see fit to implement
>the analyse signal ("what do you want that for? Sorry, the board
>schematics have already been sent off...") I know very well what I'm
>now facing and I HATE IT!

I have a customer who had this situation.  Their software engineers
worked for three years without post mortem debugging. One day
I visited to solve a problem and after not satisfactorily being
able to get an analyse line connected, decided to try using the
post mortem debugger after MANUALLY resetting the system. After
all, the difference between resetting and analysing is that 
analysing is supposed to preserve system state, but I reasoned
that even after resetting, there still might be useful information
in the network.

In fact this turned out to be the case.  I was able to read the
all values I wanted, and even jump down channels. 
This was with T805 transputers and the D7314A ANSI C Toolset.
 
According to the "Transputer Instruction Set" most of the registers
are not initialised when the machine powers up or is reset, but
the external memory interface is initialised. Evidently, in the
system I was debugging, the external memory was not being
destroyed as a result of being initialised.
OR... perhaps the system was wired so that "resetting" it actually
analysed it, so a genuine reset occurred only after power on.

I am sorry I cannot help with Inquest and the ST20450. I would
be surprised though if SGS is not concerned about its reputation
so I would suggest letting them know how you feel.

For real time systems post mortem debugging is such a valuable
tool it would be a tragedy if they don't get it working.

Before I read your post I was on the point of recommending the
ST20450 and its derivatives to a large multinational company,
but I shall try now to get a committment from SGS-Thomson to
solve this problem first.
 
So, hoping something can be done,


Peter Morris

------------------------------------------------------------------------
Transputer Systems, 14A First Avenue, Nailsworth, South Australia, 5083.  

Email: peterm@senet.com.au   Tel/Fax: Intl +61 8 8344 4776
------------------------------------------------------------------------