Re: SGI Indigo 2, machtype and timers

[Adrian Chadd <adrian%freebsd.org@localhost>]

On Mon, 20 Apr 2026 at 17:48, Steve Rumble <stephen.m.rumble%gmail.com@localhost> wrote:
>
> On Mon, Apr 20, 2026 at 5:12 PM Adrian Chadd <adrian%freebsd.org@localhost> wrote:
>>
>> I have a feeling I'm going to need to commit some super early
>> reach-through sins for this stuff so Indy
>> and Indigo2 machines with the 3D framebuffers aren't absolutely
>> terrible during early boot.
>
>
> I ran into something similar a couple of months ago due to gio and imc depending on the Indy's subtype being set already by ioc. The attached kludge resolved those both (since gio attaches to imc).
>
> It's a bit ugly, but the same could probably be done for int.

Yeah, but it doesn't fix the int2 gfifo check slowing things down.
Maybe I can drive that down from 5ms
to something smaller just to get to int0 probing.

As much as I like cleanly built stuff, honestly I'm tempted to just
probe the subtype for indy/indigo2
mainbus.c after it probes the mach_type so the rest of the hardware
setup will work better-er.
We know there's going to be the IOC status/config stuff for
indy/indigo2 so it should be fine to
poke it directly.

I don't have an Indigo1 or earlier hardware that would use the grtwo
style hardware to see what
weird workarounds are required there.



-adrian