Re: CMD MSCP SCSI problems? (again?)

[David Brownlee <abs%absd.org@localhost>]

On Fri, 10 Feb 2023 at 02:33, Dave McGuire <mcguire%neurotica.com@localhost> wrote:
>
> On 2/9/23 21:07, Dave McGuire wrote:
> >>>   Ragge, how sure are you that the 24VUP assumption and the timing
> >>> loop is at fault here?  It doesn't look like it from my initial tests.
> >  >
> >> Not sure at all, it may very well be something else.  I just thought
> >> at that since the mscp code works on slower machines, so it smells
> >> timing error.
> >
> >    It certainly does.  But now I can't get it to boot at all...I don't
> > know if there's a clue here or not.  I just reinstalled the tape
> > controller (TMSCP), and now that's the only mscpbus (mtc0 at uba0,
> > mscpbus0 at mtc0) that it finds.  That's on a different board (a Pertec
> > interface), not the CMD SCSI board.  The install kernel won't even boot
> > from CD now.
> >
> >    Of course proving a negative is hard, but I've tried >20 times, and
> > it definitely would've booted by now in its previous configuration.
> >
> >    VMS boots fine and sees the disks and tape.
>
>    [replying to my own post here...]
>
>    Annnnd, as if just to mock me, NOW it boots.  I had just run and shut
> down VMS, and did not power-cycle the machine, but I have no idea if
> that's related.  This time it sees mscpbus0 and mscpbus1, and comes all
> the way up.
>
>    My point behind mentioning it is that it clearly isn't some sort of
> "multiple MSCP controllers don't work" bug, but it's still something flaky.

Maybe a scattering of kernel printfs in dev/qbus/uda.c to see where it
fails to attach? (when it chooses to do so :)

David