[Oberon] Tracing Oberon-0 boot problems

[muller at inf.ethz.ch]

Doug Danforth <danforth at greenwoodfarm.com> is having problems 
getting any version of Native Oberon to work on his machine.  
We are now systematically trying to trace the problem.  Since this 
might be of interest to others, and others may want to contribute 
with their experiences, I'm taking the discussion to the mailing list.

To recap, Doug said (speaking of the Oberon-0 boot diskette):
> Under both V2.3.6 and Alpha no cursor appears on my screen.  (I believe under
> some other version of Native I had a red cursor which now I can not reproduce).

And I asked: That probably means booting did not work out.  We should 
concentrate on finding out why.  What exactly do you see after booting 
the Alpha version of Oberon-0?  If you don't have a visible cursor, 
what do you see in the System.Log?  Do the Oberon tracks even appear,
or is the screen just black?

To which he replied:
> Yes, the tracks do appear.  The system track is as expected and the
> user track is all grey.  There simply is no cursor and the keyboard
> does not seem to do anything.

And now we are "live" again with this email... (BTW we've decided to 
concentrate on the Alpha version).

These are startup commands in Oberon-0 (from Install.Oberon.Text in
Alpha/Build/Build.zip), one of them is apparently hanging:

{ System.OpenLog }
{ System.Open Install.Tool }
{ UsbSystem.Start }
{ UsbMouse.Init }
{ UsbKeyboard.Init }
{ ConfigInput.WatchMouse 3 }
{ Config.Oberon0 on }

My guess would be one of the Usb commands, since the first two
commands have apparently been executed (you see the log and Install.Tool
in the system track), and the last two are quite harmless.

When it is hanging, do the NumLock/CapsLock keys toggle the 
corresponding LEDs?  If not, keyboard interrupts are probably 
not working.  If the LEDs work, try Ctrl-Break (once or twice) 
and let us know what the resulting trap viewer says.  This should
show us which command above is hanging.

> Yes, I used the same diskette.  I can not reproduce the results of
> Partitions.Show for 2.3.6 because it now traps (which is what it has
> always done except for that one time).

Ok, let's ignore 2.3.6.

> > > Warning: Logical drive chain points to sector without partition table.
> > 
> > This is potentially a problem, but we need more information from
> > Oberon itself.  If you answer the questions above we can proceed.
> 
> Ok (see above)
> So why should I have a working ME system that does not complain and
> yet partinfo doesn't like it?

The primary partition table format is quite old (DOS 2.0?) and 
the extended partition table format also (DOS 3.0?).  These 
formats were never very well documented by Microsoft (especially
in earlier years), so a lot of the information about them is based 
on folklore collected over the years.  Therefore different programs 
often make different assumptions about what the tables should look 
like.  This is especially a problem with the extended partition 
table and logical drives, which have a relatively tricky nested 
linked list structure.  partinfo is probably being conservative,
and ME not.

> I have *not*
> systematically tried every combination.  I have tried explicitly setting PS2=1
> and also MouseType=0 thru 6 but to no avail.

With the Alpha version it is not really necessary to configure
the mouse beforehand using OBL config strings.  Paul Reed modified 
the post-2.3.6 Input module to detect the mouse automatically and 
this is successful in most cases.

The "ConfigInput.WatchMouse 3" command in the Oberon-0 startup
is a fall-back for the rare case where the autodetection fails.
It installs a task that "watches" the mouse cursor for 3 seconds.
If any movement is detected, the task prints "Mouse movement 
detected" and terminates.  If not, it prints "Press Shift to 
configure mouse" and when the SHIFT key is pressed, it calls 
the ConfigMouse.SelectMouse command, which allows manual cycling 
through the mouse drivers and ports with the keyboard.

To conclude, once we have the answer to the questions about the keyboard
LEDs and Ctrl-Break above, we can continue.

-- Pieter

--
Pieter Muller, Computer Systems Institute, ETH Zurich / MCT Lab, Zurich
Native Oberon OS: http://www.oberon.ethz.ch/native/