Even though it's from almost 20 years ago, I think this talk
buy uwe@ and niallo@ is still pretty relevant at describing the
general process of porting OpenBSD to a new architecture.

https://www.openbsd.org/papers/eurobsd2005/niallo-uwe/

On Tue, Sep 03, 2024 at 06:03:04PM +0300, S V wrote:
> сб, 31 авг. 2024 г., 14:30 Janne Johansson <icepic.dz@gmail.com>:
> 
> > Den sön 25 aug. 2024 kl 16:07 skrev S V <nerfur@gmail.com>:
> > > Hello!
> > > I'm curious there to start with new arch.
> > > I got access to rare architecture with installed linux (sources for
> > kernel with support for it is open sourced) and proprietary c/c++ compiler
> > (gcc-compatible mostly) (no sources for it)
> > > Any suggestions on there to start? Anybody interested in more info/dive?
> >
> > I think there could be several reasons why you did not get an answer,
> > so I will guess what those reasons might be:
> >
> > 1. The question was very vague, as if the platform/arch was a secret
> > or something. If the receiver doesn't even know what we are talking
> > about, it might be hard to gather interest for it. Also, openbsd is
> > mostly not interested in 32bit arches, so anything not 64-bit is
> > probably not going to gather much interest right off the bat, and most
> > things that do support 64bit are already present.
> >
> 
> Yeah, I understand this, it wasnt secret, but I'm asking it more for myself
> cause arch is very rare and probably will never be accepted upstream by
> reasons of closedness and random retail shop unavailability.
> 
> My vague question was more with goal to get generic way of porting to new
> arches ( that also will be useful to next curious people with another
> arches! )
> 
> (Arch is so called e2k, russian VLIW, that recently become at least
> partially opensource in terms of linux kernel. Also I plan to get loongson
> board for experiments too.)
> 
> 2. The non-existence of a free compiler is also a huge show-stopper.
> > Who would pay for compilers to produce such a platform? Not obsd devs
> > at least. Even in the super unlikely event that someone (you?) would
> > turn this compiler into free software overnight, adding yet another
> > compiler into src/ would be a hurdle as well.
> >
> 
> I was planning to play with it myself as sort of pet. (at least I dont
> build grand plans)
> 
> 3. Given that we don't even know what machines we are talking about
> > here, we can only guess that it is some rare kind of machine, which in
> > turn could prevent it from easily becoming a supported OpenBSD
> > platform, this requires boxes to be available for builds of -stable
> > releases, -current and N+1 machines for ports building.
> >
> 
> Sure, I understand all this.

-- 
Matthieu Herrb