[rescue] Sun-2 in FPGA (and 3C400 are not very good)

[Romain Dolbeau]

Le mer. 2 sept. 2026 à 19:38, William Barnett-Lewis via rescue
<rescue at sunhelp.org> a écrit :
> I've been following this with some interest, but due to various
> reasons was always more interested in the Sun 3x systems. Is anyone
> aware of something like this for that 030 architecture?

Software-wise, there was a post recently from Andy Wallis on updating
TME - but there's very little architectural information on sun3x vs.
the other sun/m68k architecture, which have a lot of documentation
available. Supporting sun3x would basically be implementing what SunOS
and NetBSD support on sun3x...
Dan Moisa has created a Qemu sun3 machine, but again, not sun3x.
software emulation is fast enough that you don't need the features of
sun3x anyway :-)

FPGA-wise, I have a draft sun3 (again, no x) that boots the PROM - but
it needs to go through the same update than my draft sun2 did to
actually do something useful. And first I need to debug something in
my sun2 disk emulation which is still a bit dodgy it seems. AI slop
;-)

For these kinds of work, AI is a boon. Up to very recently (at least
Q1 26!), there were no open-source Motorola MMU implementations for
FPGA - no '851, no '030 or later. Now you can find both the MC68030
(Mac IIsi) and the MC68040 (NeXT!) - maybe still a bit rough, but
existing in MiSTeR cores that run real operating systems. Ditto for
the FPUs ('881/'882, '040 built-in). All of that is AI-enabled work.

To support sun3x (SW or FPGA), I think having an AI look at the
sun3/sun3x differences in available kernel sources would give a good
starting point - I suspect the primary difference will be the DMA;
sun2/sun3 is DVMA (all useful devices are connected on the CPU side of
the MMU), while sun3x might not be as the MMU is internal to the '030.
Whether Sun did regular physically addressed DMA or added a sort of
IOMMU, I don't know.

Cordially,

-- 
Romain Dolbeau