On Wed, 6 Mar 2002, Anders Magnusson wrote:
> On 6 Mar 2002, Perry E. Metzger wrote:

> > The machine only had 18 bit pointers. It is a little light in terms of
> > real memory space -- only about a megabyte in a full Moby.

> That's not entirely true. The KL10-E (Used in 2060/2065 IIRC, someone may
> correct me) had extended mode addressing which gave you a virtual
> address space of 30 bits (but I think only 23 bits were used).
> It should be quite possible to memory-wise to get NetBSD to run on
> such machine, but I think the amount of work to make it run would
> not be negligable :-)

I wouldn't really expect there to be any (or many) real PDP-10 systems
that would run such a port port.  Given the 18-bit pointers, I can see
that being an impass, and I don't know if any of the emulators out there
specfically support the KL10-E.

My curiosity has mostly to do with wondering if it's even possible to
support the PDP-10 processors (or a subset of them).  And it seems like
it would be possible.  Then beyond that, is there any merit in doing the
port?  Not so much to provide yet another OS for the PDP-10, but more of
it being an excercise of the portability of NetBSD.  At least to
something a bit more unusual than the run-of-the-mill 32 and 64-bit
processors.  Someone pointed out the arm26 port, which may have already
achieved this.

It would be a lot of work, and there are probably are a lot more
pressing things for us spend our time solving.  But maybe there will be
some people here both curious and able to pull off a port.  There would
be some worth in the bragging rights.  Sort of like Linux has with their
IBM mainframe port.

-brian.