Marvell Kirkwood - anyone?

[Rodney W. Grimes]

> On Tue, Apr 23, 2019 at 03:19:15PM +0100, Balanga Bar wrote:
> > Is it possible to get a version of portsnap from that point?
> 
> I don't know enough to answer that question.  I think it would be fair
> to assume "no".

You would not want ot use portsnap I do believe,
but what you may want to start with is the ports.txz file
that should of shipped on the i386/amd64 disc1 .iso

That would of been the copy of the ports tree at the time
the system was released.  Your going to have probably significant
paints in finding tar balls that match, but it is not impossible.

The shorter your list of needed ports the better.

Your are then going to have to check for CVE's against
all that code and either fix or mitigate the issue, if
it is a non connected low access device CVE's may not
matter at all.

For OS sources I think I would either grab from svn the version
at releng/8.x that matches what you decided on, or a point 
on stable/8 slightly after this.  My reasoning here is that you
would be able to pull in specific changes from later in the
life of stable/8 that you may need fairly easily, ie svn merge.


> IIUC you seem to be looking for an _easy_ way to get Kirkwood back up
> and working.  I'm going to be honest and say there isn't one.

I Concur.

> Here are the approaches I think you can take:
All very reasonable too.

> 
>  - stay on 8.x; bring individual port updates to it from ports-head and
>    build your own ports.  Difficulty: hard.
> 
>  - figure out what src changes after 8.x regressed Kirkwood; check 
>    out src 12-STABLE, build your own src, and use FreeBSD.org packages.
>    Difficulty: expert.
> 
>  - stay on 8.x; attempt to bring a modern ports tree to it and build
>    your own ports.  Difficulty: challenging.
> 
> The difficulty level of the first approach depends on which ports you
> are going to try to use.  shells/bash?  Probably not too hard.  Anything
> GUI-related?  Very hard.
> 
> None of these approaches are achievable within hours; they will take
> days, or, in the case of the third approach, weeks.
> 
> fwiw, the second approach is the only one where your fixes could be
> merged back into FreeBSD.  If I were personally determined to run
> Kirkwoord, that's the approach I would take.  (I gave my GuruPlug
> away some time ago.)
> 
> I'm sorry that I can't be more encouraging.
> 
> mcl

-- 
Rod Grimes                                                 rgrimes at freebsd.org