Re: VME Bus Error handling on MVME3100 and 6100 boards

[Tim Mooney <mooney@aps.anl.gov>]

Ernest L. Williams Jr. wrote:
> > ...
> > So, I have filed a technical concern with both Motorola and WindRiver.
> > Of course, this will lead to TUNDRA but we need to get this resolved if
> > we want to move forward and have reliability.
> >
> > I will post the results back here hopefully in the near future.
>
>
> ========== Response from Motorola below ===============================
>
> Motorola has received the fiollowing response from an engineer that was
> involved with the TEMPE and from Tundra:
>
> Tundra states that "the error handling with respect to a BERR
> termination as a VMEbus master works as described in the Users Manual.
> Tundra has no plans of changing this."  If the customer doesn't feel
> that signaling via an interrupt is insufficient then we have no
> recourse.  Please note that both Mac and Tundra agree that a failed read
> of a valid VME address would likely occur only as a result of a
> catastrophic system failure, i.e. the target board smoked.  The
> probability that a failed read, when (or if) retried would return data
> is essentially zero.
> ========================================================================
>
> Not sure, what to make out of that initial response?

"...failed read of a valid VME address..." suggests that they interpreted
                     ^^^^^
your concern differently than you intended.  Or maybe hardware guys just
assume software developers are so good that no software would *ever* try to
read something that isn't there.  Kind of flattering, yes, but I, for one,
don't deserve that level of respect.  Anyway, the answer seems clear, even
if their justification for it isn't quite on point.


--
Tim Mooney (mooney@aps.anl.gov) (630)252-5417
Beamline Controls & Data Acquisition Group
Advanced Photon Source, Argonne National Lab