Hi Ben,

In message <Pine.SUN.3.95.970115212910.3839B-100000@crystal.cs.ucl.ac.uk>
          Ben Strawson <B.Strawson@cs.ucl.ac.uk> wrote:

> I think that there are two problems here - one is getting the Cumana
> drivers (and the others) working at full speed, the other is trying to get
> the NDA lifted from the Cumana card.  Even if the Cumana card's were
> running at full speed, I would still like to see the NDA lifted.  Of

So do I.
Well, I mailed them around January 2nd about another problem and asked them
again to withdraw the NDA. Let's wait for their reply because Mark says one of
the persons at Cumana wouldn't mind lifting the NDA and Chris Manning says they
told him at Acorn World that they would remove the NDA.
Anyway, I don't see what should take so long to get this sorted out. I asked
them to withdraw the NDA about half a year ago. Mark Gillman first said they
had to insist on it to "protect secret company information" (that is, the
address of the controller in podule space) but later said during our discussion
that they would reconsider it. They also talked to the kernel team about it
which explained to them why they would like the NDA to be removed.
No news from them since except for what Mark says that one person at Cumana
gave its OK but there's somebody else who has to give his OK.

> course lifting the NDA would I believe help in solving the first problem
> of getting the cards working at full speed, since then there would be many
> other people who could look at the Cumana code and try and fix it.

It certainly would. I've been very interested to get this driver to work faster
for many months. Had the sourcecode been available, I would have tried to test
everything I can long ago. Mark asked Powertec users to try the driver with
interrupts and DMA turned on. Ale and I replied that we would have liked to be
able to do this too for the Cumana.
This doesn't mean we would have been successful but how can we know if we can't
even try. And all this because of some simple addresses. Mark once said he
could even release csc.c because this doesn't include NDA information,
everything under NDA is inside a header file, probably called cscreg.h)

So long, Markus